TW554276B - Methods, system and computer for determining a winning bid for a sealed bid auction at an optimal bid price - Google Patents

Abstract

A method (32) of valuation of groups of assets (12) by partial full underwriting (14), partial sample underwriting (34) and inferred values (40) of the remainder using an iterative and adaptive statistical evaluation of all assets. Statistical inferences drawn from the evaluation are applied to generate the inferred values. The assets are collected into a database (76), catagorized, subdivided by ratings as to those variables and then rated individually. The assets are then regrouped according to a bidding grouping and a collective valuation is established. Simulated bid scenarios are examined for combinations of bid prices and a best bid price according to risk and return is determined.

Description

554276 A7 _______B7 V. Description of the Invention (,) Background of the Invention The present invention relates generally to the valuation method of financial instruments, and more specifically to the rapid valuation of a large number of financial instruments. Large amounts of assets such as loans (for example, tens of thousands of yuan in loans) or other financial instruments ’are sometimes made available for sale due to economic conditions, planned or unplanned releases, or legally awarded compensation. Thousands of sales of commercial loans or other financial instruments, which sometimes involve assets equal to billions of billions, sometimes need to take place in months. Of course, the seller of the asset wants to obtain the best price for its portfolio assets and will sometimes group the assets into " asset parts (t r a n c h e s). The term "asset portion" as used herein is not limited to foreign instruments, but may also include financial instrument groupings of unrelated nationality or jurisdiction. Bidders may bid on all asset portions, or only on certain assets. Some asset parts. In order to get an asset part naked, bidders usually need to propose the highest price for that asset part. In determining the bid for a particular asset part, bidders will usually hire a certifier to evaluate as much as possible More assets, and are valid for a limited time. When the moment of bidding will expire, the bidder will evaluate the assets that have been certified at that moment, and then try to extrapolate the assets that have not been analyzed by the authenticator Traditionally, the creation of bids and offers is based on the inherent price of an asset and an understanding of the competition. The investment and / or division of the entity that performs the bidding has reached an agreement on the minimum and maximum price that it will unilaterally determine. This process Upon completion, bidders may significantly underestimate an asset portion and bid on uncompetitive bids, or may bid higher than The price of the certification is subject to an unknown amount of risk. Of course, because the goal is to use the Chinese paper standard (CNS) M specification (210X297 mm) for the paper size that the bidder can profit from. -4- (Please read first Note on the back, please fill out this page again.) Order it printed by the Ministry of Economic Affairs, the Intellectual Property Field Employee Consumption Cooperative. Losing the bid for the asset part because of the obvious undervaluation of the asset part represents a lost opportunity. I hope to provide a system that helps to accurately value a large number of financial instruments in a short period of time and understands the relevant gains of a bid It is more desirable to be able to determine the bid of each asset part of the combined assets of a financial instrument in a sealed auction. The invention is described in an exemplary embodiment, providing a method of iteration and adaptation, in which a combined asset is Divided into three main valuations. Full certification of the first type valuation of a portfolio asset is based on a negative sample. OK. The second valuation type is effectively sampled from the types of common descriptive attributes, while the assets in the selective random sample are fully certified. The third valuation type is a statistical inferred valuation, which uses the certified price and the first Differences from Part 2 and applying statistical inferences to individually value each of the assets in Part 3. Aggregation and data reduction are used in the valuation of Part 3. As more assets are certified as the process proceeds, then It has an increase in the number of assets at the price established in the first and second part and a decrease in the number of assets in the third part, and the difference in valuation of the assets in the third part becomes more and more limited. More specifically, The assets in Part III are evaluated by grouping the assets into clusters, based on the similarity of the valuation of the assets in Parts 1 and 2. Speculative bids are generated using estimates to determine their best bids within a range of parameters set by the bidder. This optimal bid is obtained through an iterative bid generation process. Take a bid at a sealed bid This paper size applies to Chinese National Standard (CNS) Λ4 size (210X29 * 7mm) (Please read the precautions on the back before filling this page)

-5-554276 A7 B7 V. Description of the invention (3) (Please read the notes on the back before filling out this page) The bid of each asset part of the portfolio assets of financial instruments on sale is simulated by a bid disclosure process and The best bid that determines its highest probability of recovering a successful bid. Brief description of the graphics_ Figure 1 is a flowchart showing a known method for valuing a combination of assets Figure 2 is a flowchart showing valuing a combination of assets according to an embodiment of the present invention; Figure 3 is a procedure Figure, which shows in more detail one embodiment of the first part of the rapid valuation method for a large number of portfolio assets, which divides assets into types of differences; Figure 4 is a flowchart showing the first part of the rapid valuation method for a large number of portfolio assets , Which is gathered from a foundation to an asset part or a combined asset foundation; Figure 5 shows the probability distribution of a model asset that has been deduced for its recovery; printed by the Consumers ’Cooperative of the Intellectual Property Field of the Ministry of Economic Affairs. Flow chart of the supervised learning steps of the method; Figure 7 is a flowchart of the unsupervised learning steps of the method of Figure 3; Figure 8 is an embodiment of the unsupervised learning method; Figure 9 is the first step of the rapid asset valuation method 1st generation (first pass) embodiment; Figure 10 is a flowchart of one of the fuzzy clustering methods used in unsupervised learning of Figure 8; This paper scale is applicable to the Chinese National Standard (CNS) ) A4 specification (210 X 297 mm) -6-554276 A7 B7 V. Description of invention (4) Figure 11 is a set of tables showing examples of model selection and model weighting for a rapid asset valuation method; (please first Read the notes on the back and fill out this page) Figure 12 is a table showing the demonstration of a rapid asset valuation method; Figure 13 is a cluster diagram of a demonstration cluster method of a rapid asset valuation method; And _ 14 is a computer network map. Ministry of Economic Affairs, Intellectual Property, Xiaogong Consumer Cooperative Printed Component Comparison Table 10 Method 1 2 Portfolio Assets 1 4 Certification 1 6 Part 1 18 Untouched Remainder 2 0 Rough Inference 2 2 Valuation 2 4 Valuation 2 6 Bid 2 8 Quick Valuation system 3 0 jm. Lane 3 2 Valuation method 3 4 Sampling 3 6 Part 2 3 8 The computer scale is applied to the Zhongguanjia Standard (CNS) Λ4 specification (21G X297 mm) 554276 A7 B7 V. Description of the invention ( 5) Printed by the Intellectual Property of the Ministry of Economic Affairs, Smart Industrial and Consumer Cooperatives 4 0 Book of δ 4 4 Second part 4 4 Generated 4 6 Uncertified part 4 8 Type 5 〇 Type 5 2 Cluster 5 4 Cluster 5 6 Cluster 5 8 times Cluster 6 〇 cluster 6 2 cluster 6 4 cluster 6 6 tree view 6 8 valuation block 7 〇 asset part 7 2 asset part 7 4 asset part 7 6 database 7 8 captured information 8 〇 Standard 8 2 Form 8 4 Group 8 5 Flow chart (Please read the precautions on the back before filling this page)

、 1T This paper size applies Chinese National Standard (CNS) A4 specification (210X 297mm) -8- Intellectual property of the Ministry of Economic Affairs ^ Printed by 7M Industrial Consumers Cooperative 554276 A7 B7 V. Description of invention (6) 86 Complete cash method 8 8 Partial Cash Method 90 Group 9 2 Group 9 4 Certifier 96 Full Price Table 98 Certification Group 10 10 Certifier 1 0 2 Partial Price Table 1 0 4 Partial Price Group Full Certification Group 10 6 Complete Sampling Procedure 1 0 8 Partial Sampling Procedure 110 100% Sampling 1 2 Complete Sampling Group 114 Rule 116 Complete Sample Data Price Table 12 0 Cluster Sample Groups 22 Re-certification of Asset Level 125 Manual Data Entry 12 6 Initial Credit Analysis Form 12 8 Asset Level 13 0 Adjusted Credit Analysis Table 13 2 Partial Sampling Letter Usage 134 Statistical Inference Algorithm This paper size applies Chinese National Standard (CNS) Λ4 specification (210X297 mm) (Please read the precautions on the back before filling this page)

-9-554276 A7 B7 V. Description of the invention (7) 136 Authentication cluster table 13 8 Adjusted credit score (please read the notes on the back before filling this page) 14 0 Adjusted credit table 14 2 Inferred credit estimate 144 Untouched Assets Table 146 Asset Level Valuation Steps 14 8 Golden Flow Bridge 1 50 Cash Flow 152 Inferred Cash Flow Bridge 15 4 Proposed Asset Part Bid 156 Asset Part Model 15 8 Thresholds 160 Thresholds Conditions 161 Simulation Bid Public Analysis 16 2 Management 16 4 Assets Part Bidding Ministry of Economic Affairs Intellectual Property ^ H Industrial Consumer Cooperative Print 166 Valuation Phase 168 Bid Preparation Phase 17 0 Type 17 2 Type 17 4 Type 17 6 Data 178 Hard Disk Storage 18 0 Minimum Three Points Asset valuation This paper applies the Chinese National Standard (CNS) A4 specification (210X 297 mm) -10- 554276 Intellectual property of the Ministry of Economic Affairs ^; Printed by the employee consumer cooperative A7 B7 V. Description of the invention (8) 18 2 Vertical axis 18 4 Horizontal axis 186 Worst case percentage 18 8 Denomination line 190 Best case percentage 19 2 Most likely case percentage 19 4 Probability 19 6 Point 19 8 Area 2 0 0 Curve 2 0 2 Area 2 0 4 100% probability line 2 0 6 Supervised learning 208 Unsupervised learning 2 1 0 Method 2 12 Definition 2 14 Expert opinion 2 16 Sample authentication method 2 18 Coordination 2 2 0 Setting 2 2 2 Classification 2 2 4 Application 2 2 6 Rule 2 2 8 Credit analysis sheet The paper size applies to Chinese National Standard (CNS) A4 (210X 297 mm) (Please read the precautions on the back before filling this page)

-11-554276 A7 B7 V. Description of the invention (9) 2 3 〇Data acquisition module 2 3 2 Variables > EB module 2 3 4 Hierarchical segmentation module 2 3 6 Fuzzy clustering method module 2 3 8 Authentication view Module 2 4 〇 Method 2 4 2 First step 2 4 4 First step 2 4 6 Second step 2 4 8 Fourth step 2 5 〇 Fifth step Intellectual property of the Ministry of Economy ¾ Printed by Industrial Consumer Cooperatives 2 5 2 No. Six Steps 2 5 4 Seventh Step 2 5 6 Court Auction Price 2 5 8 Market Price 2 6 〇 Tree Table 2 6 2 Shaker Tree 2 6 4 Shaker Tree 2 6 6 Shaker Tree 2 6 8 Shaker Tree 2 7 〇 Shaker Tree 2 7 2 Shaker Tree 3 〇〇 System 3 〇2 Server (Please read the precautions on the back before filling this page)

This paper size applies the Chinese National Standard (CNS) Λ4 specification (210X297 mm) -12- 554276 A7 _____B7__ V. Description of the invention (10) 3 0 4 Computer 306 Database server. (Please read the notes on the back before filling in this I) Detailed description of the preferred embodiment Figure 1 is a graphic 10 illustrating a known method for valuing a large number of portfolio assets through an authentication cycle 12 to facilitate Bid to buy (for example) portfolio assets in an auction 1 2. Figure 1 is a high-level view of a non-repetitive and non-automated typical authentication and inference method. In Figure 10, the certifier authenticates (printed by the Goods and Consumers Cooperatives of the Intellectual Property Supervision Bureau of the Ministry of Economic Affairs of the PRC 14) the combination of individual assets in the asset 12 to produce a certified first part 16 and an uncertified remainder Part 1 of 8. Before any asset is certified, the first part 16 is zero percent and the remaining part 18 is 100 percent. As the authentication process proceeds, the first part 16 increases and the remaining part 18 decreases. The goal is to certify as many assets as possible before bidding on portfolio assets. The team of certifiers continues to individually certify (1 4) until a bid must be made. A rough inference 20 is made to evaluate the remainder 18. Inferred 値 20 becomes uncertified inference 値 24. A rough inference yields a valuation of the remainder 18 of 24. The valuation 22 is only the sum of the individual assets 中 in Part 1 16. However, valuation 24 is a group valuation generated by inference and may therefore be undervalued. Valuations 22 and 24 are then added up to produce a portfolio asset 値 26. The valuation process is performed on each asset component (t r a n c h e) of the portfolio asset. Figure 2 is a diagram illustrating an embodiment of a system 28 for rapid asset valuation. Figure 2 contains a representation of the procedural steps taken by the system 28 to value the portfolio asset 12. The system 28 individually values each asset "touched", except that it is considered to be statistically insignificant or financially irrelevant for the untouched capital paper size applicable to China National Standard (CNS) A4 specifications (210X 297 mm) -13- 554276 A7 B7 —— ____-.................. V. Description of the invention (11) Except for a very small amount of 30. Specifically, except for a very small amount of 12 in the portfolio asset All assets other than 30 undergo repeated iterations and adaptive valuations 32, of which the combined assets 12 are individually valued, individually listed in the form, and then selected and grouped from the form into any desired or required group or The purpose of the asset part is to facilitate the bidding (as described below). As shown in Figure 10, the certifier starts a complete certification (1 4) of the individual assets in the combined asset 12 to generate the first part of the complete certification of one of the assets 16. The certifier also authenticates (3 4) a sample of one of the assets of the second part 36 of the portfolio asset 12 and a computer 38 statistically infers (40) the price of one of the third part 42 of the portfolio asset 12. The computer 38 also Generate (44) the form repeatedly (described below), The waiting table shows the puppets (as described below) assigned to the assets in sections 16, 36, and 42. In one embodiment, the computer 38 is constructed as a separate computer. In other embodiments, the computer 38 is The system is constituted as a server, which is connected to at least one client system (shown and described in FIG. 14) through a network, such as a wide area network (WAN) or a local area network (LAN). For example, refer to FIG. 2. One of the unsampled and uncertified portion 46 of the third portion 42 of the portfolio asset 12 accepts a statistical inference procedure 40 using a fuzzy C-means cluster ("FCM") and a composite high / expected / low / time series / Risk ("HELTR") score to generate two types of 48 and 50. HELTR is defined as Η—high cash flow, E—expected cash flow, L—low cash flow, and T—the timing of cash flows (eg month: 0) -6, 7-18, 19-36, 37-60), and R—assetment of the borrower (assessment) (9-box used by credit analysts). Type 48 is considered to have sufficient commonality Sex is valued as a whole. Type 50 is further divided into clusters 52 and 54, which Then it is further subdivided. This paper size applies the Chinese National Standard (CNS) Λ4 specification (21〇X: 297 mm) (Please read the precautions on the back before filling this page), \ 2 口 Ministry of Economics Smart Money / i- ^ a (Printed by Gongxiao Bone Cooperatives-14-554276 A7 B7 V. Description of the invention (12) (Please read the notes on the back before filling out this page) Cluster 52 is divided into sub-cluster 56 and 58 and cluster 54 is Subdivided into sub-cluster 60, 62 and 64. Clusters and sub-clusters are both shown in the " tree " diagram 66 and become boxes in the valuation block 68. These individual asset prices are then reorganized into asset sections 70, 72, and 74 to facilitate bidding. Any number of The asset part can be assembled by the seller in any configuration group. The individual asset information (not shown) of each asset in the portfolio asset 12 is entered into a database 76, and the selected data 78 is based on one of the iterative and adaptive procedures 32 established criteria 80 is retrieved from the database 76. When a standard 80 for the valuation of any asset is established, the established standard 80 is stored in the database 76 to evaluate other data sharing this established standard Asset data in library 76. Iterative and adaptive valuation procedures 32 thus form (82) valuations (described below) and group them (84) for bidding. Intellectual Property of the Ministry of Economic Affairs ^ M Industrial Consumer Cooperative Cooperative Prints 3 and 4 together form a flow chart 8 5 which illustrates the essential features of an embodiment of a system 28 (shown in Figure 2) for valuing a large number of portfolio assets 12 (shown in Figure 2). Valuation procedures 1 4, 3 4 and 4 0 (also (Shown in Figure 2) Sequentially used in the system 2 8 in the following manner. As mentioned above, the full certification 14 is the first type of the valuation process. The grouping and sampling certification with the full certification of the sample 34 is the second type of the valuation process. Statistical inference 40 is the third type of valuation procedure, which is an automatic grouping and automatic valuation. Procedures 14, 34 and 40 are based on objective criteria as described below. "Certification" as used herein refers to a processing procedure in which one person "Certifier") to review an asset in accordance with established principles and determine the current cost of purchasing the asset. During certification, the certifier uses a pre-existing or established standard 80 to facilitate valuation. " Standard, refers to the asset price and the paper size applicable to the Chinese national standard (0 milk) 8 4 specifications (210 \ 297 mm) -15- 554276 A7 _____ _B7 _ V. Description of the invention (13) According to this Rules of rating of other types. For example, taking a standard as an example, an authenticator can decide that the borrower's three-year cash flow history is the type of information related to asset valuation, and can give a certain rating to cash flows at various levels. Full certification is performed in two ways, a full cash-based approach 86 and a partial cash-based approach 88. Both the full cash-based approach 86 and the partial cash-based approach 88 began with their asset groups 90 and 92 (14), which were individually and completely reviewed (see Figure 2). This complete review14 is usually based on a large amount, or other appropriate currency, the amount of its assets being viewed in relation to other assets in the portfolio, or on the basis that its borrower is a well-known person or is sufficiently reliable that its assets can be quickly and easily Reliable and complete certification, or their assets are ordered in the market so that the difference between the prices of these assets is minimal. The asset group 90 is evaluated by the certifier 94, and each asset in the group 90 has a small difference in valuation, such as its assets backed by cash or tradable goods with a cash price, and is placed in a complete price list 9 6. The individual frames selected from the assets in Table 9 6 are stored as a complete certification group (98). Group 92 is evaluated by a certifier team 100, which may be the same as team 94, but each of its assets receives an undervalued or partial price and is placed in a partial price table 102. The individual selected assets in the asset section of Table 102 are stored as part of the price (full certification group). The standard cash-based method 86 and the partial cash-based method 80 (shown in Figure 2) are stored in a database 76 in a digital storage memory (not shown) of a computer 38 (shown in Figure 2). (As shown in FIG. 2) for supervised learning 206 and unsupervised learning 208 of automatic valuation 40. This paper size applies to China National Standard (CNS) A4 (210X 297 mm) (Please read the precautions on the back before filling this page)

Printed by 1T Intellectual Property Workers 'Consumer Cooperatives, Ministry of Economic Affairs-16-554276 A7 B7 V. Description of Invention (14) (Please read the precautions on the back before filling out this page) Intellectual Property of the Ministry of Economic Affairs ^ -θΜ Industrial Consumer Cooperatives' Printing Disaster Sampling Certification 34 is accomplished using two procedures, a full sampling 106 procedure and a partial sampling 108 procedure. A full sample 106 is used for a large number of asset types, and includes 100% of the sample group 110 in the sampled asset type. The assets in the full sample 106 are not individually certified, but are certified to the full sample group 1 1 2 according to a predetermined commonality. A resulting complete sampling group valuation (not shown) is generated and then abolished quarantine in accordance with Rule 1 14 to produce another complete sample asset price (Table 1 16). The individual complete sample assets in Table 1 16 are then electronically loaded with any complete sampling group valuation required for their bids 1 1 8 as suggested by the group of assets in the asset section. The number of assets in a certification sample group can be from one to any number of assets. Partial sampling 1 08 is used for assets and other types, and includes a sampling method from a percent of a representative group in one of the clusters sampled and random sampling of other groups in the cluster to form a Cluster sample groups 1-20. In the partial sampling 108, all groups were sampled, and some groups were partially evaluated by inference from the cluster sample group 120. Partial Sampling] 08 includes the use of manual data to enter asset levels of 1 25 and re-certify 1 22 to generate an initial credit analysis table 1 26, which provides an asset level adjustment 128 to generate an adjusted credit analysis table 130. As described above, the individual assets are selected from the adjusted credit analysis table 1 30 according to the asset portion group to generate a partially sampled credit card Π 2 for the asset portion 70 (shown in Figure 2). The automatic evaluation program 40 uses a supervised learning method 206, an unsupervised learning method 208, and uploaded from a statistical inference algorithm 134 to generate an authentication cluster table 136 stored in a digital storage device. Supervised learning method 206 This paper size applies the Chinese National Standard (CNS) A4 specification (210 > < 297 mm) -17- 554276 A7 ______ B7 5. In the description of the invention (15) 'One person knows what questions to ask Experienced certifiers who set up prices help computers decide whether an asset is a good investment and how to value it. In the unsupervised learning method 208, the computer classifies and classifies assets and objectively evaluates these assets based on feedback from the data. A certifier periodically reviews the unsupervised learning method 208 to determine whether the computer makes a reasonable certification conclusion. Point Brain uses statistical algorithms 1 to 34 to perform its inferences. For example, not limited to this approach, one embodiment uses the Design For Six Sigma ("DFSS") quality paradigm, which was developed and used by General Electric Company and applied to a product development using multiple generations ("MGPD") model of labor-demanding (Due Dillgence ("DD")) asset valuation method to evaluate its asset information with increasing accuracy. Learning methods 206 and 208 use their accumulated knowledge as the valuation progresses into the probability of cash flow recapture and recapture calculations in a continuous, immediate manner. Supervised learning method 206 uses business rules to identify clusters of assets that have common ground for valuation. The unsupervised learning method 208 uses feedback from previous data valuations performed by the process 40 to determine whether progress has been made in increasing confidence in valuation. Thanks to the use of high-speed computers, the identification of all available raw data and the reacquisition of the interrelationships between the clusters of these available raw data can be achieved, as described below. In an exemplary embodiment, an unsupervised organization's fuzzy cluster average ("FCM") method using HELTR scoring technology is used to infer a credit score valuation on assets in a portfolio asset, as described below . These cluster technologies have been developed in response to more complex classification segments to describe the assets in the portfolio assets that need to be estimated within a period that cannot be processed manually and the paper dimensions are applicable to the Chinese National Standard (CNS) A4 specification (210X297) )) &Quot; -18- (Please read the notes on the back before filling this page)

1T Smart Production ^ H (Printed by Industrial-Consumer Cooperative Printing 554276 A7 _____B7_ V. Description of the Invention (16) High Total Assets. An example method first organizes the valuation score (fixed and / or recaptured) at A computerized system. Then adjust the valuation scores for specific factors and business decisions. Then perform a coordination that describes multiple valuation scores for the same asset and an overall adjustment to the interview / retraction inference valuation. Organizational valuation scores are obtained by ( (In electronic form) according to the effectiveness of each cluster's descriptive attributes to organize each cluster's valuation: cluster number, cluster name, descriptive attributes of the cluster, probability of regain (such as HELTR score), and Perform with confidence. A cluster number is a unique identifier of a descriptive attribute of the facts about a particular group of assets that is used by an assessment expert to estimate the value of an asset. Examples of descriptive attributes include (but are not limited to) : Payment status, asset type, creditworthiness of the borrower in terms of points, ownership and eligibility. In one embodiment The cluster name is an alphanumeric name that describes the descriptive attribute or source of the cluster. An example of a descriptive attribute can be found in Figure 12 below, as described below. Descriptive attributes are the facts or ranges or directions used to generate asset prices. Computer logic Used to examine replicated clusters and warn analysts or certifiers. Because each asset can be described with a combination of many descriptive attributes, encounters at various levels of the same asset can occur. The possibility of regaining or Any number of credit scores or asset prices is an indicator of prices specified at different asset levels. All information from various descriptive attributes is combined so that its purchase price or sale price can be determined as a fixed price or a possible price.値. The illustrative example used here is the HELTR score. The paper size of each cluster applies the Chinese National Standard (CNS) A4 specification (210X297 mm). One---19- (Please read the notes on the back before filling in (This page), 11 Printed by the Intellectual Property Sentence of the Ministry of Economic Affairs Employee Cooperatives 554276 A7 B7 V. Description of Invention (17) A unique set of descriptions Descriptive attributes and assigned HELTR scores. The unique attributes of each cluster contribute to the valuation of the cluster price. Different combinations of attributes provide a higher confidence or confidence interval for the score of a particular cluster. For example, if any asset is A green paper described as having a height equal to 2.5 "and a width equal to 5" may have a price of between 0 and 1,000 yuan and has little confidence in this estimate. If the same asset is described as another fact or attribute or The direction is a real $ 20 dollar bill, which will generate a very high confidence factor for the $ 20 cluster price.-The valuation and confidence of the cluster is determined at a single point and recorded. Sometimes new ones will be obtained The information or analyst wants to change its price. The price is manually changed 'or automatically with a data field and decision rules in an automatic way via computer code. The previous price is used to reflect new information. As an illustrative example, suppose that the previous cluster confidence was recorded as 0.1, and it was discovered that a different asset with such descriptive attributes in such a cluster has just been sold for more than the predicted "most likely" price. Then apply the rule when this happens, that is, the cluster confidence is multiplied by 10. 〇 1 X 1 〇 = 1 is the changed cluster confidence. The purpose of this method is to reconcile multiple scores of the same asset and control confidence in each valuation source associated with each valuation range. The use of HELTR as an illustrative example with a sample loan point on a specific asset is as follows: This paper size applies the Chinese National Standard (CNS) Λ4 specification (210X 297 mm) (Please read the precautions on the back before filling this page ) L0. Order printed by the Intellectual Property Bureau of the Ministry of Economic Affairs, M Industrial Consumption Cooperative, -20-554276 7 Β V. Description of the invention (18) Cluster number Cluster name Expected low time sequence Valuation Confidence Bureau expected low time sequence 1 Mortgage level. 85.62 • 15 3 .3 (.3 / 1.65) (. 85) (.3 / 1.65) (. 62) (.3 / 1.65) (. 15) (.3 / 1.65) (3) Placement-Recourse 2 Assets .45 .4 • 31 3 .7 (.7 / 1.65) (. 45) (.7 / 1.65) (. 4) (.7 / 1.65) (. 31) (.7 / 1.65) (3) Classification- Industry-Year 3 Coordination · .9 .5 .2 2 • 65 (.65 / 1.65) (. 9) (.65 / 1.65) (. 5) (.65 / 1.54) (. 2) (.65 / 1.65 ) ⑵ Use-Borrower η X 1.65 .6999 .4792 .2374 2.6059 (Please read the notes on the back before filling out this page) Group Sense (c ο nsensus) Valuation System. 6 9 9 9 The highest value, most likely • 4792 Low as .2374 A timing to 2.6059. Different logic can be applied to handle any weighting. The Intellectual Property Bureau of the Ministry of Economic Affairs (industrial and consumer cooperatives printed consensus scores was developed with the background of an overall hypothesis. Assuming that an overall assumption has changed, method steps 1 2 and 1 3 8 are included in the calculation method to weight the consensus scores. Illustrative examples are the erroneous reacquisition of certain valuation factors, including: overall economic changes, replaceable market prices established by an asset class, and inferred asset valuation methods compared to other profitable calculations Loss or improvement of the method. This paper size applies the Chinese National Standard (CNS) Λ4 specification (210X 297 mm) -21-554276 A7 B7 V. Description of the invention (19) (Please read the precautions on the back before filling this page ) In another embodiment, an interactive correlation tool is used to quickly understand and describe the composition of a portfolio asset. Typically, the tool is used to correlate responses from a user-selected variable to a portfolio asset Other variables. This tool quickly identifies unexpected high or low correlations between two attribute variables and response variables. Attribute variables are of two types : Continuous and type. The calculation of cross-correlation is based on the correlation of all relevant variables with their frames or levels, and (in one embodiment) is presented as a two-dimensional matrix for easy identification of combinations Trends between assets in the asset. First, the cross-correlation tool identifies all attribute variables in the combined asset as one of two types (continuous or type). For each variable, the aggregation level is calculated in boxes on the continuous Variable, and the value is calculated in the type of variable. A user who wants to use the tool to identify correlations will select a response variable, Y, for example, for expected retrieval. For all attribute variable pairs (x 1 and X 2) and its level (a and b), according to the following formula to calculate the mean 値 of the response variable, Yr: Y > Two sum (Y (xl a and x2 = b)) / count (xl = a and x2 = b). The expectations of the printed cooperative response variables of the cooperatives of employees of the Intellectual Property Bureau of the Ministry of Economic Affairs are calculated based on the following formula: Y Expected two (sum (Y (x 1 = a)) * count (xl two a ) + sum (Y (x2 dib) * c〇unt (x2 dib)) / (count (xl = a) * count (x2 dib )) · Expected 値, Y ..., from the individual use of the weighted events of x 1 = a and x 2 = b, the selected variable, the error of Yp, Y, the following formula is used to calculate Y crr ο I Ύ " γ-Y cxpcct. This paper size applies to the Chinese National Standard (CNS) Λ4 specification (210X297 mm) -22- Intellectual property of the Ministry of Economic Affairs, printed by the Industrial and Commercial Cooperatives 554276 A7 B7 V. Description of the invention (20) In the example, the expected volume and error are displayed in a multidimensional display to easily identify variables from the expected volume. In another exemplary embodiment, a transfer function method for converting raw data into a final bid is used, as described below. Table 136 is an electronically adjusted credit analysis table using the revised coefficients obtained in procedures 14, 34, and 40 to provide a coefficient adjustment of the asset's credit score of 38 and generate inferred individual asset credit cards. Adjusted Credit Analysis Table 1 40 . Individual assets are taken from a table 140 (if required) by asset grouping to generate an inferred credit valuation 142. An inference is finally performed on the negligible remaining portion 30 of the "untouched" asset to generate a table 144 of untouched assets.値 from Table 144 was selected to generate an untouched asset valuation. Full cash appraisal 98, partial cash appraisal 104, full sample credit appraisal 1 18, partial credit card 132, inferred credit card 142, and any designations from untouched asset table 1 44 are cumulative and mutually exclusive, in order of priority From the complete cash valuation 98 to the inference credit card 142. The sum of the valuations represents the value of the portfolio assets. FIG. 4 is a flowchart of the bid preparation stage 168 performed by the system 28 (shown in FIG. 2). Cumulative valuations 98, 104, 118, 132, 142 and 144 are combined in the risk preference loan level valuation step 1 46. A deterministic cash flow bridge 148 is generated using a cash flow time series 50 to make a speculative cash flow bridge 152. A speculative or probable cash flow bridge 152 is generated and used to determine a proposed asset part bid 154, which is applied iteratively with an asset part model 156 until a certain threshold 158 is reached. The threshold 値 158 is (for example) This paper size is applicable to China National Standard (CNS) A4 specification (210X297): -23- ---------- Φ „(Please read the precautions on the back before filling (This page) Order line 554276 A7 _B7_ V. Description of the invention (21) (Please read the notes on the back before filling this page) A return ("IRR") internal rate, which is greater than a certain amount, A certain profit time ("TTP") and a positive net present value ("NPV"). Generally, NPV is defined as: _ = c. + &Amp; (equation A) where C. is the time 0 For investment, Ci is the expected return at time 1, and r is the deduction factor. The basic idea is that one yuan today is greater than one yuan tomorrow. In the case of insurance strategies, 'NPV is defined as:

=-Ye-fVr) A Λ (Equation B) where P is the premium, E is the expected nominal cost, and c is printed by the Intellectual Property of the Ministry of Economy 笱 Employee Consumer Cooperative as the claim cost. Essentially, Equation B is about how to generate net income, the difference between its profit and weighted expected risk. Note that their sum totals all the strategies in a particular section. It is also noted that all premiums, nominal costs, and claim costs have been discounted before being substituted into the equation. As a result, a profitability score is generated. If the threshold condition 1 60 is satisfied, the bid 1 54 is subjected to a simulated bid public analysis 1 6 1 to predict whether the bid can be a winning bid. The results of a sealed bid auction are based on the amount of bids from each bidder. The execution of the auction involves disclosing all bids and selling the auctioned items to the best bidders. In a traditional sealed bid auction, once a bidder submits a bid, it is not allowed to change its bid, and the bidder does not know that the paper size of other bidders is applicable to the Chinese National Standard (CNS) 8-4 specification (210X 297 public envy) "-24- 554276 A7, ___B7 V. Description of the Invention (22) (Please read the notes on the back before filling this page) Bid until its bid is made public, making the auction result uncertain. By offering a higher price, the chance of a naked auction is increased, but the profit obtained from a naked auction at a lower price is reduced. The simulated competitive bidding can increase the chances of obtaining the highest profitability. By setting a range of bids / sale prices, it has a tendency to deplete any competing bidder's resources so that it retains the highest capital before its own resources are exhausted. Trade your most wanted asset. Focus on bidding decisions with an analytically sound process, because purely entertaining business judgments can be amplified by a way of obtaining information without a hidden agenda, personality, or unilateral knowledge. Printed by the Consumer Cooperative of the Intellectual Property Field of the Ministry of Economic Affairs Each potential bidder has a range of possible bids that could be used to participate in a sealed bid auction. The bid range can be expressed as a statistical distribution. By speculatively sampling the distribution of bid prices, a possible auction scenario can be simulated. Further by using an oversampling technique, such as a Monte Carlo analysis, many scenarios are simulated to produce a distribution of results. The distribution of the results includes the odds and price profit of the naked auction items. By changing the cost of your own bid, you can determine the probability of an auction that is barely your own bid. The following core elements are used to simulate a competitive bidding profit: compile market rules and contracts into computerized business rules; compile potential competition / market power, forecasted budgets and priorities into a preference matrix; personal bidding power, preferences ; Risk / return trade-offs that have been agreed into a preference matrix; and a computerized speculative optimization. Analysis 1 60 simulates a competitive environment, and applies other national paper standards with various financial resources to the Chinese National Standard (CNS) A4 specification (210X 297 mm) " -25- 554276 A7 B7 V. Description of the invention (23) Silai The bids calculated by the bid countermeasure system 2 8. In one embodiment, analysis 1 60 (by way of example and not limitation) includes a total bid limit, such as may be the case where the total price of assets exceeds the financial capabilities of the entity using system 28. In Yishi ;, analyze the profitability of various combinations of assets that can be estimated in the 160, in the case of limited bidding resources. Analysis 1 60 also considers its history of bids against known competitors and information about the various asset types preferred by competing bidders. In analysis 1 60, the asset portion bid is then evaluated and transmitted by management 1 62 to make a final asset portion bid 1 64. All valuations prior to bid 1 64 can be repeated as needed. Furthermore, because this method is self-adjusting and iterative, the asset part bid 1 64 may increase, with more and more prices being discovered with each repeated operation performed by the system 28. The method described in flowchart 85 includes a valuation phase 166 (shown in Figure 3) and a bid preparation phase 168 (shown in Figure 4). Valuation phase 1 66 includes procedures 14, 34 and 40. The valuation phase 166 continues until it terminates, with its automatic valuation procedure 40 and sampling procedure 34 attempting to find additional prices in various assets of the asset type. Referring again to FIG. 2 and based on the rapid asset valuation, the data types 170, 172, and 174 in the combined asset 12 are identified in each asset and stored in the database 76. The iterative and adaptive valuation method 32 uses a portion of the selected data 78 and statistically applies the standard 80 to those selected data 78 to increase the known asset price, rather than roughly inferring the asset price 20. According to Method 28, the increase in production is divided into at least the first portion 16, the second portion 36, and the third portion or the remainder 42. If you use Procedure 14, the assets in Part 16 will be fully recognized. The paper size applies to Chinese National Standard (CNS) A4 (210X 297 mm) (Please read the precautions on the back before filling this page), 1T Ministry of Economic Affairs Printed by the Intellectual Property Bureau, M-Industrial Cooperative Cooperative, -26- 554276 Ministry of Economic Affairs, Smart Financial Sentence, W-Industrial Co-operative Cooperative, printed by A7 B7 V. Description of the invention (24) The standard of this valuation is 80. Using procedure 34, method 28 samples the amount of an asset from the second portion 36 representing the group in the second portion 36 to determine the complete sampling group valuation 118 and the partial sampling credit 132 of the second portion 36, and establishes The additional criterion for this valuation is 80. Using the program 40, the partially supervised learning method 206 and the partially unsupervised learning method 208 are executed with an automatic analyzer, such as the computer 38 of FIG. 2. For learning, the automatic analyzer extracts the established criteria 80 and selected data 78 about the third part or the remainder 42 and divides the third part 42 into parts 46, and then further divides each part 46 into types 48 and 50 and The type 50 is divided into clusters 52, 54 and the clusters 52, 54 into sub-clusters 56, 58, 60, 62, and 64, which use the standard 80 input from the database 76 and each method 206 and 208. Individual asset valuation of assets in the sub-clusters 56, 58, 60, 62, and 64 by statistical inference. Individual asset valuations are included in the cluster table 1 36 (see Figure 3) and after adjustment 138 are included in the credit analysis table 140. The established standards 80 are objective because they are derived from the database 76, and these standards are determined during the full certification process 14 and the sample certification process 34. In other words, the information obtained in the complete price list 96, partial price list 102, table 116, initial credit analysis table 126, adjusted credit analysis table 130, adjusted credit analysis table 140, and untouched asset table 1 44 for all assets It is placed in a database 76 in a digital storage device (such as the hard disk storage 17 8 of a computer 38), and is associated by program 40 with a standard 80 from programs 14 and 34. During procedure 40, a criterion 80 (which has a statistically acceptable degree of reliability) is entered. That is, the program 40 learns iteratively when it evaluates and establishes the standard 80. Supervised learning method This paper size applies Chinese National Standard (CNS) Λ4 specification (2 丨 O X297 mm) (Please read the precautions on the back before filling this page)

-27- 554276 A7 B7 V. Description of the Invention (25) (Please read the notes on the back before filling this page) Method 206 and unsupervised learning method 208 The assets in part two 36 are linked to the established standard 80 in database 76 to increase the accuracy of statistical inference estimates 142. Selection data 7 8 (similar to the selection data 78 for assets in Sections 16 and / or 36) relating to one or more of the assets in Part 3 4 2 is then placed in a database 76 and then inferred by statistics The price of each asset in the third section 42 is determined from the information entered. Printed by the Intellectual Property Sentence of the Ministry of Economic Affairs, Consumer Cooperatives During the method described in flowchart 85, assets are valued at a different asset level, and individual asset prices are tabulated or grouped into one or more combinations. In order to have maximum flexibility for various bid scenarios, any subset of the portfolio assets 12 are individually evaluated and priced at a specific time frame. In the known method 10, if the seller of an asset reorganizes its assets, for example, from the group of the asset company to the group of the geographical location of the borrower, the revaluation of the bid may be insufficient, because the rough inference 20 Need to be implemented. When using system 28, because individual asset prices are generated and listed in tables 96, 102, 116, 130, 140, and 144, these can be electronically reorganized into different valuations 98, 104, 118, 132, 142, whose "food chain" selection criteria are mutually exclusive and can be selected by the analyst performing the evaluation, which will be further described below. If the sellers group the assets, they can easily make groupings based on the seller's group or asset portion, and can easily generate an appropriate valuation of the asset portion. 46 The individual assets of the third part 42 are thus easily reorganized to objectively obtain the inferential valuation of the group or asset part. Many methods are available to establish asset prices. According to the objectives of valuation, the relative advantages of different valuation methods establish a valuation technique for a specific asset. The paper dimensions are applicable to the Chinese National Standard (CNS) Λ4 specification (210X 297 mm) -28- 554276 A7 B7 employees of the Ministry of Economic Affairs ’s intellectual property field Consumption Cooperatives Printing V. Advantages of Invention Statement (26). One approach is similar to a "food chain" in that it retains the hypothetical production method and chooses the interval with the highest confidence interval. In the foreword of the illustrative example of the food chain, someone may prefer to value a financial asset by buying and selling similar assets in the open market, rather than relying solely on personal opinion. In the ranking, the market-to-market price is selected to surpass one's opinion. In the same way, assets in a portfolio asset with estimated cash flow recapture can be valued using several valuation techniques. A typical goal (with a high probability of success) is to establish future cash flows. Valuation methods are ranked according to their ability to accurately quantify cash flows, or cash equivalents, with estimates of the least adverse variables and / or the most favorable variables. Assets are valued by all practicable methods, or they can have business logic rules to eliminate duplicate work. When it is known that once the best method has been used, a more accurate method will eliminate the need to estimate the price of an asset. . In order to provide the best prediction of asset prices, assets are evaluated using each method in a food chain until each particular asset is valued by the best available method. Once this optimal price is found, the asset is set to this price, regardless of other lower prices (which have more variables) in the food chain, and this price is transmitted to completion. For example, a portfolio asset is valued using a food chain. The first valuation method in the food chain is the method that best meets the valuation goals. One is to find the price tag with the highest accuracy (closest confidence interval). Once the asset is valued with a strategy (where the price of the unique asset is established), the price is sent to the valuation form and any further steps in the food chain are removed. I -------- f · " '(Please read the notes on the back before filling in this page) The size of the paper is applicable to the Chinese National Standard (CNS) Α4 specification (210X29 * 7mm) -29 -554276 A7 _____B7 V. Description of the invention (27) The assets listed in the original portfolio assets that conform to any valuation method are maintained in the untouched balance sheet. The goal is to leave zero assets in this untouched table. An example of a food chain is as follows, in order of preference. (A) assets with 100% cash ’(b) assets with partial cash, (c) liquidity market prices of similar assets, (d) direct certification, and (e) inferential certification. The food chain approach enables: to find the best probability distribution model, reduce the probability distribution variables (especially regarding unfavorable tails), provide the ability to quickly build probability distributions while preserving all available knowledge in the customer base, and provide the best price predictions to regain Ability at any moment in the process. As shown in Figure 4, the general structure of the bid preparation phase 1 68 is to determine a price of 1 64, similar to the buying and selling privilege valuation paradigm, in which the winning investor will have the right (but not the obligation) to regain the investment. The price is divided into three parts for each asset part, namely the time price of the currency component, the inherent price component, and the possible cash flow component. The time price and inherent price of money are calculated decisively and rarely change once established. The time price of money is calculated by multiplying the company cost of a low-risk investment capital by its investment representing a viable period of another investment opportunity that was abandoned due to the execution of the investment. The inherent price is a known liquid asset price that exceeds the purchase price and is available immediately after controlling the asset. One embodiment is a properly traded security that is purchased as part of a portfolio asset at a price below the market price. Possible cash flow variables are a function of the assumptions made by an appropriate diligent group and the method chosen by the group to convert raw data into cash flow and regain data flow. The system described here is used to reduce negative variables and find the price tag. This paper size applies the Chinese National Standard (CNS) Λ4 specification (210X297 mm) (please read the precautions on the back before filling this page), 11 lines. Intellectual Property Bureau of the Ministry of Economic Affairs (printed by Industry and Consumer Cooperatives -30- 554276) A7 B7 V. Description of the invention (28) (Please read the notes on the back before filling in this page) Intellectual Property of the Ministry of Economic Affairs Gou S (printed by the Industrial and Consumer Cooperatives 5 is a typical triangle with a minimum three-point asset valuation 1 8 0 Probability distribution chart. According to the procedure 40, the minimum value of each of the three conditions of each financial instrument is evaluated. A vertical axis 1 8 2 represents an increasing probability and a horizontal axis 1 8 4 represents an increasing part of the recovery. Figure The worst case percentage of the denomination line 1 8 8 is 186, the best case percentage of the denomination 188 is 19%, and the most probable case percentage of the denomination 188 is 192. The probability of the worst case percentage 1 8 6 is Zero, best case percentage 1 90 The probability is zero and the probability 192 of the most likely recovery 192 is 値 represented by point 196. Below the curve 200 is defined by the points 186, 196, and 190 connected The size of area 198 is representative The price of the mark. The price of the marked asset maintained in the area 202 of a rectangle (defined by the line segment 204 of 100% of the denominated 88% of the 100% probability of regaining 88%) is the denomination of the asset that can be attributed to the curve 200 Measurement of the part of 1 88. Points 1 86, 196 and 190 and line segments 188 and 204, and areas 198 and 202 will be reacquired with the selected data selected by the relevant asset 78 and the standard 80 and asset price applied to the asset. The attribution probability changes. The horizontal axis 1 84 can be expressed as a percentage of denominations (eg, yuan) instead of denominations. When a currency unit is used, the area 198 below the curve 200 of different assets will be a currency unit, and Therefore, the amount of region 198 is related to each other, so it is important to total bids 70, 72, and 74. The more you know about the asset, the more accurate curve 200 can be. When criteria 80 is established, statistics are applied to the curve 200 to assist in establishing the positions of points 1 86, 196, and 190, thus establishing region 198 and establishing the expected value of the asset. The timing of the cash flow (its impact price) can be based on the results of the statistical graph of the timing attribute. For example, cash flow The time sequence obtained can be divided into three segments: 0-6 months. The paper size is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) -31-554276 A7 B7. 5. Description of the invention (29) (Please read first Note on the back page, please fill in this page again) 7-12 months, 13-18 months, etc. Using the automatic analyzer of algorithm 134 3 8 can select the frame width and learn the transaction relative to the sensitivity of the timing of the valuation relative to Estimated recapture and costs that may be determined by the certifier. In an exemplary embodiment, a minimum of 4 boxes should be used when the deduction factor exceeds 25%. For deductions between 10 and 25, a minimum of 6 boxes should be used to cover the possible recovery cycles. In accordance with Procedure 40 ', other sources of information are selected so that a certifier can use them to evaluate the value of financial instruments. In this case, the standard 80 established by the certification team 94, 100, 114, 122, and 140 in procedures i 4 and 34 is useful. According to the method described in flowchart 85, the original data is transformed into a retrieved and a rule set is selected to apply a valuation to the raw data, and this rule set is coded into the valuation database in the form of standard 80. Each time a cluster is touched multiple times during the valuation of procedures 14, 34, or 40, a consensus forecast is generated and applied to the cluster. According to system 28, the probability distribution of cash flows and the timing of the asset level are determined by generating a valuation transfer function 1 46 at the asset level. This function will use the original data, infer the assumptions it will generate, and aggregate the assets Valuation of individual assets in a section. Ministry of Economic Affairs, Intellectual Property, Employee Consumption Cooperation, and Du Printing Because not all recaptures are uniform, a method is provided to establish the variability of cash flow recapture. Individual assets are aggregated for group exposure. Traditionally authenticating as many denominations as possible within the allowable time, identifying a sizable sample to remain in the cluster. The cluster reserve was predicted using a recovery analysis of 155 denominations plus a sample size of 2.65% and variables. The sample size of 30 thus generated is the denomination of 100 assets, the denomination of 150 to 1,000 assets, the denomination of 400 to 5,000 assets, and this paper size applies the Chinese National Standard (CNS) A4 specification (210X29 * 7mm) -32- 554276 A7 B7 V. Description of the invention (30) 500 denominations of 10,000 assets and 600 denominations of 20,000 assets. (Please read the notes on the back before filling this page) During the statistical inference process 40, the remaining assets in the third part 42 of the portfolio asset 12 were clustered with descriptive authentication attributes or standards 80, and random samples were taken from Each cluster and certified samples. In one embodiment, when the average value of the asset level is less than 10%, the sampling in the program 40 is stopped. In another embodiment, the sampling is stopped when the average part variable level of the asset is less than 15%. The combined asset average variable is not used as a stopping point if the potential unit sold is smaller than the entire combined asset. According to procedure 40, the re-evaluation of cluster sampling is deduced to the corresponding total number of clusters. When using system 28, the goal is to reach each inferred asset valuation through three or more unique clusters. During process 40, the relevance of a cluster of certification confidence and descriptive attributes is weighed. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economics. For example, non-limiting, 0 = no confidence in the descriptive attributes of this cluster will provide a meaningful valuation; 1 = full confidence in the descriptive attributes of this cluster will provide An accurate valuation as if each instrument was individually certified, while a number between 1 and 0 indicates part of the confidence in the valuation. These reconciliations occur in the adjusted confidence analysis table 130. In procedure 40, the cash flow at the asset level is then adjusted in the adjusted confidence analysis table 140 by the overall economic coefficient. In one embodiment, the overall economic coefficient is related to major asset classes, such as (but not limited to) real estate residential loans or commercial equipment loans. These factors can be applied globally, such as (non-limiting) legal trends, " GDP " estimates, guarantor trends, deposits efficiency, borrower group codes, etc. This paper Standards for Finance, Guanjia County (CNS), M specifications (2H), X 297, Chu) -33- 554276 A7 B7 V. Description of the invention (31) A method for sampling a combination of assets includes searching for key assets i mentioned A. The owner of the watch uses the affiliated property insurance assets of the exemplary Fante listed below to influence the valuation of the assets in the wind and the wind. (Please read the note on the back first. (Fill in this page again)

Customs printed by the Intellectual Property of the Ministry of Economic Affairs of the Employees 'Cooperatives printed on this paper are applicable to the Chinese National Standard (CNS) A4 specification (210X 297 mm) -34- 554276 Printed by the Intellectual Property of the Ministry of Economics Employees' Cooperatives of Consumers A7 B7 V. Invention Description ( 32) _Table A: How much is the portfolio asset property borrower (by borrower group UPB) safe organization (yes / no) secured loan type (period, revolving, etc.)% UPB from mortgage in the first position Deposit score (0 = poor, 1 = excellent) UPB's 12-month deposit %% of the final payment of the principal #borrowed part of the loan borrowed by the owner UPB single family residence residential retail industrial hospital understanding most households have developed land / Undeveloped / other office stock / margin loan asset attribute segmentation is accomplished by encoding the attributes as "analog variables" This paper size applies the Chinese National Standard (CNS) Λ4 specification (210 × 297 mm) (Please read the back Please fill in this page again for precautions), 1T -35- 554276 A7 B7 Intellectual property of the Ministry of Economic Affairs ^ Printed by the co-operative cooperatives V. Invention description (33). For example, a common asset attribute is "Is the borrower paying within the last 12 months?", If the answer is yes, it is coded as "1" with a variable, otherwise it is coded as "0 ':. A similar" analog variable "is used Other asset attributes. The segmentation procedure is completed by using any statistical procedure that processes the encoded asset attributes to divide the combined asset into groups of similar assets. One algorithm is a k-means clustering. In an example , Which uses three types of asset attributes: · unpaid principal balance (UPB), probability of payment (score from 0 to 1), and security score (probability guaranteed by real estate collateral) 'assets can be classified as having similar The five groups of attributes. Once the grouping of assets is performed, the number of samples that are to be adopted and proposed for further authentication review is calculated by: establishing confidence levels (using which Statement of total recapture), establishing accuracy (total recapture in each segment (h) that someone wants to build), and providing a range of on-site predictions and recaptures The percentage of total unpaid principal balance (UPB) (R), based on: L " J 「令 广 ―77ΓΤ— λ ~ LI _η two sample size N = cluster size xi two sample i UPB yi two sample i Regained (Please read the notes on the back before filling this page) • Γ.

Line 1T · This paper size applies to China National Standard (CNS) A4 specification (210 × 297 mm) -36- 554276 Μ B7 V. Description of the invention (34)

R Σ is I • A _ / VΣ 'cluster is expected to regain% h2 ^ k2 xn 1

N Σ; Σ () ά ·) 2 yV — 1

(Equation C Λ = error tolerance for estimating ^ = with Ϋ IR ", n, Λ a / Σ brother N Σρ〆, n almost one ί = ι Σ ~,-/ = 1 ^ = constant in TchebysheVs Formula: " ~ μ ^ | ~ k4Vard) with probability 2 1 ~ 丄

(Equation D (Please read the notes on the back before filling this page>

By solving equation C 'of η, the required sample size for a given cluster can be obtained. Solving equation C further allows the user to determine n (the probability of the sample size calculated by IB / k2), and the relevant authentication 値 will predict the total cluster regain to within the error of h, assuming its total partition regain prediction When it is determined using equation D. In fact, it is not easy to predict the variability of total recovery without data available. A blank table program implements the above work, by generating data in a Monte Carlo simulation, and analyzing the results to guide the user until a suitable sample size is obtained. Table B provides an example of research from a group of 20 loans, with predicted (expected) recovery between 20% and 30% of UPB, and the range of UPB between 1MM and 2MM. Eight samples are necessary to predict that the total return of 20 loans with a 75% confidence index will be within 10% of the actual. This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) Order% Wisdom of the Ministry of Economic Affairs: Printed by the Production Bureau Staff Consumer Cooperative-37- 554276 A7 B7 V. Description of the invention (35) Table B: Sample size blank Form program help system

(Please read the notes on the back before filling out this page) Printed by the Intellectual Property Co., Ltd. Consumer Cooperative of the Ministry of Economic Affairs Appropriate variable adjustment forecasts are executed for each asset and the valuation table is constructed to include each asset in the portfolio assets. The continuous probability of regaining a unit that is valued for sale, which in one embodiment is the asset portion. When using system 28, the internal rate of return (" IRR ") and variables will then be evaluated. The better asset part has lower variables to a given IRR. Use the deduction rate of the plan And it is estimated that the net present value (" NPV ") of each asset part is greater than zero. A deduction rate is the risk of determining the opportunity cost from the principal, plus the FX transaction cost, and the general uncertainty inherent in the variable that predicts cash flow recapture. If there appears to be more than five percent uncertainty about the project's plan to have a negative NPV, no bid is offered. Transaction evaluation is based on the asset part with the following decision criteria: IRR, risk of the IRR in an asset part, pre-paper size applicable to China National Standard (CNS) A4 specification (210X 297 mm) -38- Economy Printed by the Ministry of Intellectual Property and Employee Consumption Cooperatives 554276 A7 B7____ 5. Description of the Invention (36) The measured willingness and ability to pay for the asset part, the moment of profit (“TPP”) and the risk of repayment from the asset part, And the NPV of the expected cash flow from the asset part deduction to the risk-free rate. In a competitive bidding environment (when the content of a portfolio asset is non-negotiable), the investor or seller has a strong financial incentive to select only the portion of the total assets available for a transaction that will provide the best risk / return of its overall financial structure . Higher probability assets that meet the minimum risk / return expectations and have the most favorable odds are more attractive to investors. Entire group; is divided into individual saleable sub-portfolio assets or asset parts. Each asset segment has an estimated cash flow probability distribution and duration from previous analysis. These asset parts are then given a trial price. The new assets are combined with the existing asset performance of the seller or buyer, and executed by Monte Carlo instances (related to the interactions they use). The asset part selection process includes randomly selecting the part of the asset that you do not want to buy. Once the portfolio asset performance presents a pattern, statistical optimization can be used to find the best choice for what price to buy the restricted asset portion. The use of NPV may be misunderstood. Due to the benefits of connection and double deduction, the double deduction will occur when the pessimistic instance scenario is deducted to obtain PV. Using delayed profit can overcome this limitation, and marginal capital costs or risk-free rates are used as a deduction when the analyst performs the valuation to determine. The supervised learning method 206 of the inference valuation process 40 and the steps 120, 122, and 126 of the partial sampling process 108 are essentially similar, that is, the authenticator actively intervenes in the method, but the method is automated. Figure 6 is a first-class process diagram to illustrate a method for automatically authenticating a divisible financial instrument asset. The gold paper size is applicable to the Chinese National Standard (CNS) A4 specification (210X 297 mm) (Please read the precautions on the back before filling out this page}, 11 line · -39- 554276 A7 _ B7 V. Description of the invention (37) (Please read the notes on the back before filling this page) The first cluster of financial tools is defined by common attributes (2 1 2). One expert opinion on the price 2 1 4 is provided to it from the definition based on attributes Sample selection of clusters. This opinion is used in the sample authentication method 2 16 and the combination of attributes is reviewed and coordinated (2 1 8). Method 2 1 0 then selects and sets the individual attributes to be used (220) , And then classify individual assets as clusters (222). Cluster valuations are applied to each cluster asset (224). Using the cluster valuations, the prices are abolished and segregated (226) according to a rule to generate a credit Analysis table 228. Printed by the Ministry of Economic Affairs Intellectual Property Employee Consumer Cooperative Figure 7 is a flowchart of an exemplary embodiment of an unsupervised study 208 containing several modules. A data acquisition module 230 any available relevant information 78 . The variable selection module 232 identifies the asset-related variables that it considers important through credit inspection, or those that have the most significant influence when separating each asset group. The one-level segmentation module 234 is based on the key variables selected by the analyst To divide the entire portfolio of assets into boxes. An FCM module 236 further classifies each box into clusters based on the natural structure of the asset data. A certification review module 23 8 specifies the estimated cash flow and risk score for each cluster 138 (As shown in Figure 3). This score is then supplied to individual asset prices in the credit analysis table 1 36 of its assets from the cluster that has been adjusted in the program 40 to generate an adjusted credit analysis table 1 40 This procedure is iterative and continuous and can be executed by a computer to make it sustainable when standard certification is being performed elsewhere. Figure 8 illustrates another exemplary inferential valuation that replaces the method described in Figures 3 and 4. Method. In another method 240, a seven-step method is used to quickly value a real estate loan portfolio asset using a fully certified, paper rule Applicable to China National Standard (CNS) A4 specification (210X 297 mm) -40-554276 A7 B7 V. Description of invention (38) Combination of partial certification and inferential valuation. First, assets are sampled based on risk (242). Second, the asset is certified (244) and its valuation is recorded. Third, market price clusters are formed (246), for example by FCM (as described below). Fourth, the regression model of certified assets Is established (248). Fifth 'selects one of the best models of certified assets from those previously established (248) (250). Sixth, calculate the counts of the selected model (252). Seventh, the model selected in (250) is applied to the uncertified or inferential valuation portion 42 of the portfolio asset 12 'to predict the individual price of each uncertified asset in a weighted manner. The individual asset prices generated in accordance with method 240 are then placed in an adjusted credit analysis table 140 (see Figure 3). When sampling assets (242), the certifier uses stratified sampling to select assets for detailed review. Its hierarchy is constructed from collateral attributes. Examples of collateral properties of real estate portfolio assets include: collateral use (commercial or residential), previous valuation total, market price / cluster (forecasted from previous valuation total), land area, construction area, current valuation total, court auction Realization price, real estate type and real estate location. Generally, assets are sampled in a reverse manner, by deliberately selecting from a table sorted by decreasing unpaid principal balance ("UPB") or previous total valuation ("PAA"). Authentication (244) is a largely manual process, in which expert certifiers attribute price notations to collateral assets. The certified estimates are stored in a master database table, such as database 76 (shown in Figure 2). The valuation is usually summarized in monetary units (for example, 100,000 KRW). This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) --------- (Please read the notes on the back before filling in This page) Order% Printed by the Employee Consumption Cooperative of the Intellectual Property Bureau of the Ministry of Economy -41-554276 Printed by the Intellectual Property of the Ministry of Economic Affairs ^ Printed by the Employee Consumption Cooperative A7 _________ B7 V. Invention Description (39) At the prevailing market price. Figure 9 is a high-level diagram of the automated part of the method used by system 28. Figure 290 ° The automatic procedure is used by the certifier to assist in complete certification according to procedure 34 (see also Figure 3). The information obtained in Procedure 34 is used to infer valuation procedure 40 'to reduce the cost and uncertainty between due diligence valuations of financial instruments, and to reduce the costs and changes between labor valuations. These valuations use a cash flow model that includes asset level valuations 146, a deterministic cash flow bridge 148, a speculative cash flow bridge 152, and a cash flow statement 150. The resulting bid estimate 154 undergoes a gambling strategy 160 and management adjustment 162 to produce a final bid 164. FIG. 10 is a flowchart of an exemplary embodiment of forming a cluster 246. In forming the cluster 246, the certifier uses an algorithm (eg, Algorithm 1 34 (shown in Figure 3)) to perform an analysis using a model based on classification and recovery tree ("CART"), which obtains collateral Usage and market price (" CUMV ") The grouping of UW assets in the group, using the previous total valuation (" paA ") as the driving variable. The following two methods are used to estimate the performance of the CART-based model. One Method The ratio of the sum of squared errors of a CART-based method to the sum of squared errors of a simple model is called the error ratio. A simple model is a model that specifies an average asset price for all assets. The second The method is to calculate a certain coefficient, marked as R2, and defined as R2 = 1-(SSE / SST), where SST is the sum of squares. (Please read the precautions on the back before filling this page) f Paper size applies Chinese National Standard (CNS) A4 specification (210 X 297 mm) -42- 554276 B7 V. Description of invention (40) R2 is the contribution of a single asset in each division relative to the entire total The higher the R2 値 of an asset in a particular partition, the higher its contribution. The division of different portfolio assets is evaluated according to two methods. These two methods include: How good the model's predictive power is; by setting (for example) the price of each asset part to provide a suitable t level. (Please read the notes on the back before filling out this page} Intellectual Property Bureau of the Ministry of Economic Affairs Printed by the Employees 'Cooperative Cooperatives ——- This paper size applies to the Chinese National Standard (CNS) A4 (210X 297 mm) -43- 554276 A7 B7 V. Description of the invention (41) Printed by the Employees' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs Comments made 0.18% 0.06% 0.46% C R-squared loan credit loan each other C 0.733617 0.614882 1 1.579237 error ratio Total 728,651,119 VO 672,301,116,218,504 278,965,783,682,227 '477,337,270,451,034 j 438,544,547 r- CN 1,071,733,684,904,320 89,989,160,092,084 143,404,241,449,409 1,075,884,830 r < 3,886,895,042,611,430 795,206,763,683,411 971,553,092,439,969 U 82,692,009 r-Η 72,331,126,127,460 26,877,527,094,865 36,637,006,656,009 ί 379,765,147 00 1 H t- < l, 039,4〇U35,208,180 83,849,226,818,428 136,366,441,963,041 276,915,573 σ \ 1,017,087,163,438,760 | 65,902,258,632,574 41,730,444,375,417 PQ 645,959,109 MD 599,969,990,091,044 | 252,088,256,587,362 440,700,263,795,025 58,779,400 ON 32,332,549,696,133 6,139,933,273,655 7,037,799,486,368 798,969,257 OO 2,869,807,879,172,670 729,304,505,050,836 929,822,648,064,552 Information Current UPB THB sum loan SST sum SST (CART) sum SSE (simple) sum Old UPB THB sum loan SST sum SSE (CART) sum SSE (simple) sum Current UPB THB sum loan amount 1 SST sum SSE (CART) sum SSE (simple) sum asset part C0 CO01 CO 02 CO 03 (Please read the precautions on the back before filling this page)

This paper size applies to China National Standards (CNS) 8-4 specifications (210X 297 mm) -44-554276 A7 B7 V. Description of the invention (42) Printed by the Intellectual Property of the Ministry of Economic Affairs, Industrial and Commercial Cooperatives 0.11% 0.14% 0.22%) 7 1.4% 88.9% 77.5% 1.472316 1.196196 0.976192 1,101,110,287 1-4 1,151,224,039,958,210 422,217,344,655,182 751,266,118,230,950 263,274,692 un 434,634,503,617,058 38,738,988,293,867 37,416,524,313,367 3,607,465,475 CJs billion \ 7,216,788,387,309,520 1,625,118,040,406,770 2,380,977,246,884,730 184,828,399 〇〇CN 223,991,862,418,471 92,347,778,018,417 62,722,788,782,158 41,505,412 On 164,601,058,694,453 10,191,006,095,769 8,519,509,247,449 965,706,540 916,281,888 1-4 inch 2,517,412,345,887,330 279,167,796,660,054 285,976,191,024,073 927,232,177,539,735 329,869,566,636,764 688,543,329,448,792 221,769,281 CO 270,033,444,922,605 28,547,982,198,095 28,897,015,065,918 2,641,758,934 wn CO 4,699,376,041,422,190 1,345,950,243,746,720, 860,001, 860,001 SST sum SSE (CART) sum SSE (simple) sum current UPB THB sum loan loan SST sum SSE (CART) sum SSE (simple) sum C0 04 CO 05 current UPB THB sum Total loan total SST Total SSE (CART) Total SSE (simple) Total R square (CA1 touch), ^^ side «Job Zhao Hang Ugly_ 蹯 Skimming: 3 wash w wide 9 ^ 90000a (酹 鮰) ις ^ β (Please read the notes on the back before filling out this page) This paper size applies to Chinese National Standard (CNS) Α4 size (210 X 297 mm) -45- 554276 A7 B7 V. Description of the invention (43) First The steps are to define the relevant portfolio asset split. These divisions can be a pre-defined portion of an asset, for example, based on corporate, total unpaid principal balance (UPB), regional or consumer risk. The above table C is an example of the division according to the definition of asset part and asset rating (B or C). Table C provides an example output from a study of a portfolio asset with five asset components and two different asset types (B and C). The table shows how the error ratio is rated equal to different divisions. At the same time, R2 of each asset is also calculated as the asset of type C in each division. The second step is to calculate the SSE price of each relevant portfolio asset split of the CART model and the simple model (inference of average price). The error ratio is calculated by dividing the SSE according to the CART model by the SSE according to the simple model. If the error ratio is less than one, the CART-based model is a better prediction than the simpler model. More advantageously, an optimal model can be combined with CART and simple models to become a "hybrid" model, by selecting the best model to perform in each segment according to the error ratio measure (metnc). The third step is to calculate R2 値 for each asset in each portfolio asset split. R2 of each asset is calculated as (SST per division-SSE per division) / (Total SST of all assets X Number of assets in each division). Finally, all divisions are rated based on the error ratio calculated in the second step and R2 値 calculated in the third step. This model can accurately predict its segmentation price which is rated high by both measures (error ratio and R2), and the best model is combined using these measures. Table D shows the relative rating of the five asset components of the Type C (from Table C) assets based on two performance measures. This paper size applies to China National Standard (CNS) A4 specification (210X297 mm) ----------- (Please read the notes on the back before filling this page), 11 Intellectual Property of the Ministry of Economic Affairs ^ a ( Printed by the Industrial and Consumer Cooperatives-46- 554276 A7 B7 V. Description of the invention (44) Table D: Portfolio asset segmentation and assessment of the assets part c〇CR square rating error ratio rating R square CO 01 0.73 0.18% 2 2 CO 02 0.61 0.06% 1 5 CO 03 1.58 0.46% 5 1 CO 04 1.47 0.11% 4 4 CO 05 1.20 0.14% 3 3 (Please read the precautions on the back before filling this page) Figure 1 0 is a flowchart to illustrate the use of FCM An exemplary embodiment of forming a cluster 2 4 6 is selected to form a cluster of computers. A computer 3 8 (shown in FIG. 2) forms a cluster 246 by using the selected data 78 and performing an FCM analysis to generate the cluster. FIG. 11 shows the creation of a model 248 Select the best model 250 and calculate the total number 252, of which six databases are created using database 76. Computer 38 (shown in Figure 3) performs this procedure. Model creation 248 is used to assist the certifier in prioritizing assets Order based on complete certification 14 and sample-based certification 34, and inferential valuation. Printed by Intellectual Property of the Ministry of Economic Affairs ^ Employee Consumption Cooperative Figure 1 The lower half of 1 is a table showing an exemplary implementation of selecting the best model 250 from the six models based on the model 248d For example, these models change depending on which variable is used as X's. All models use the CUMV cluster (these are present in all assets). The model from the established model 248 is used to predict the court auction price ("CAV") 256, and market price ("MAV") 25 8. Other embodiments (not shown) use other models to predict other prices. This paper size applies the Chinese National Standard (CNS) A4 specification (210X 297 mm) -47- 554276 Printed by the Intellectual Property of the Ministry of Economic Affairs and the Consumers' Cooperative A7 B7 V. Invention Description (45) In selecting the best model 250, the best model of the K recovery model to be considered (here, K = 6) is selected. Best The model is selected for each UW asset, according to the following measures: do-what), 1 5 "}, where y is the UW 値 to be predicted and is the prediction from the klh recovery model, where k = 1, 2, ..., K. In the total number of calculations 252, the number of times of each K model selected in each of its CUMV clusters is calculated. Figure 11 contains these totals for CAV and MAV simulation examples. Other simulation examples are used in other embodiments. When applying model 254, a weighted average forecast from all models from which it generates a forecast for each uncertified asset is used. These weightings constitute a self-calculated frequency of 252, and the predictions are derived from simulation procedures. In one embodiment, a commercial statistical analysis software (SAS) is used to generate the model. Using an artificial part of the SAS system for each uncertified (non_uW) asset will obtain a predicted UW from each model in which each input variable (ie, "X variable") appears in the uncertified asset. Other simulation software packages share this feature. The following equation E details this procedure. (Equation ε) ij.k In equation E, if the model k produces a prediction of one of the assets 1, then Iu 2 1 and otherwise it is zero; fi "k two models k are selected in the ilhCUMV type (i = 1, 2) and the number of UW assets in the j1 h CU MV cluster (j2 1, 2, 3). Ten paper sizes are applicable to the Chinese National Standard (CNS) A4 specification (210X 297 mm) — one --- --48- (Please read the notes on the back before filling this page)

1T 554276 A7 B7 V. Description of the invention (46); and = prediction from model y 1. It should be noted that only one contribution comes from (please read the notes on the back before filling this page). Each of the assets has a prediction simulation method, and each of them is weighted by the number of times that all UW assets in the same CUMV cluster are selected by its simulation method. . Procedure 240 is also used to evaluate the lower confidence limit (1 \\ 1-1-Confidence Limit "LCL") and upper confidence limit (UCL) of the average prediction, instead of the corresponding statistics in Equation E. The Ministry of Economic Affairs, Chi-I-I, Japan-I, Japan, Japan, Japan, Japan, Japan, Japan, Japan, Japan, Japan, Japan, Japan, Japan, Japan, Japan, etc. With reference to Figure 3 again, the supervised learning method 206 and the unsupervised learning method 208 use clusters (c 1 u s t e r i n g). "Cluster" is a tool that attempts to access the relationship between the types of data sets. It is set by organizing each type into a group or cluster, so that the type in the same cluster is more than the type in different clusters. More similar. That is, the purpose of clustering is to extract a natural group of data from a large number of data sets, which produces a concise representation of a system characteristic. The unsupervised learning step 208 utilizes a fuzzy clustering method ("FCM") and knowledge engineering to automatically combine assets to facilitate valuation. FCM is a known method that has been widely used and applied in statistical simulations. The purpose of this method is to minimize the intra-cluster distance and maximize the inter-cluster distance. Euclidean distances are usually used. FCM 248 (see Figure 10) also minimizes the intra-cluster distance and maximizes the inter-cluster distance. Euclidean distances are usually used. FCM is an interactive optimization algorithm that minimizes the cost function. J = (Equation F) This paper size applies to Chinese national standards (CNS> A4 specifications (210X; 297 mm) -49- 554276 A7 B7 _ __ V. Description of the invention (47) where η is the number of data points; c is The number of clusters' Xk is the k-th data point; Vi is the center of the 1st cluster; // is the membership level of the k-th data in the i-th cluster; m is a constant greater than 1 (usually m = 2). Note its // Is a real number and is between [0, 1]. // 1 means that its first data is determined to be in the k-th cluster ', and β u == 〇 means that its first data is determined to be not in the k-th cluster. If / z ^ Two 0.5 'means that its 1st data is partially in the kth cluster to reach level 0 · 5. Intuitively, the cost function will be minimized' If each data point does belong to a particular cluster and has no members Part of the rank is in any other cluster. That is, there is no ambiguity when designating each data point to the cluster to which it belongs. The rank of the member // ik is defined as

A (Please read the notes on the back before filling this page) Σ

Xk ^ V i (Equation G) X k — V J · Intuitively, the membership level of the data point Xk at the cluster center Vi, // ik, increases as xk approaches Vi. Meanwhile, // ik will decrease as xk moves further away from Vi (other clusters). Printed by the Consumer Cooperative of the Intellectual Property Field of the Ministry of Economic Affairs, the i-th cluster center Vi is defined as (Equation Η)

IkT intuitively, the center of the i-th cluster, Vi, is the weighted sum of the coordinates of xk, where k is the number of data points. First use the desired number of one of the clusters c and the center of each cluster Vi. The paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) -50- 554276 A7 B7 V. Description of the invention (48) (Please read the back first Please note this page before filling in this page) i = l, 2, ..., c, the initial prediction, then FCM will converge to one of the solutions of Vi, which represents a local minimum 値 or a saddle point of the cost function (saddle pomt ). The quality of the FCM solution, like most non-linear optimization problems, is mainly based on the choice of initial chirp (number c and initial cluster center Vi). In an exemplary embodiment, the entire portfolio asset 12 is divided into unsupervised fuzzy clusters, and each cluster is inspected by a certification expert to assist the certifier in selecting financial instruments for the full certification 14 and the sample certification 34. On the other hand, this FCM can only be applied to section 42. As a result, each cluster was assigned a HELTR composite score for the purpose of adjusting (138) (see Figure 3). In essence, the HELTR composite score obtains the expected range and scope of cash flow, its timing, and the risks associated with each cluster. The Ministry of Economy ’s intellectual property is printed by industry and consumer cooperatives. Now refer to Figure 2. The ratio of the complete certification part 16 to the total portfolio assets 12 is (in an exemplary embodiment) 25% of the assets and all assets 60% of denomination. The complete certification of these assets is guaranteed by their size and price. However, this certification is fairly consistent for all certifiers, so this certification is unlikely to produce significant bid variation. However, the remaining 40% includes sections 36 and 42, which (in the exemplary embodiment) constitute 75% of the asset, but only 40% of the denomination is highly uncertain until certification. To the extent that the price can be seen in sections 36 and 42f For example, without any restrictions, there is an extra 5% to its total inference. The difference represents the difference between naked or losing the entire portfolio asset bid or the entire asset part bid, which is a profit of several hundred million yuan. In the case of an insurance strategy, according to Procedure 40, statistical data is used to try to answer three basic questions: (a) How should we collect the data? (B This paper is a Chinese national standard. (CNS) A4 specification (210X297 mm) -51-554276 A7 B7 V. Description of invention (49) (Please read the notes on the back before filling this page)) How should we describe the information we collect? And ( c) How accurate is our data description? Algorithm 1 34 answers question (c), and it is a computer-based method that does not require complex evidence of reasoning. Algorithm 1 for insurance policy inference valuation is suitable for answering 34 A statistical inference that is too complicated for traditional statistical analysis. The algorithm used for insurance policy valuation 134 simulates the distribution of statistical predictions by replacing repeated sampling. This algorithm usually consists of three main steps: ( I) replace sampling, (II) evaluate relevant statistics, and (III) predict standard deviation. According to insurance algorithm 1 34, the prediction of NPV standard error is performed as follows. For each risk model and for those models For each segmentation, assuming that there are N strategies in the segmentation, n samples are selected instead of sampling (for example, n = 1 00). In this example, each sample also contains N Strategies. For each sample, and for all historical strategies: -A-_ Z (Act) (Equation I) " ~ 07285 ^ ~ Printed by the Ministry of Economic Affairs Intellectual Property Field Employee Consumption Cooperative. Next, the net current NPV = Σ Ρ- Σ Ε- (Σ C) x A / Ew (Equation I) is derived from the most recent strategy. Calculate the sample standard deviation of nNPV 値. In Equation I, Act is the actual claim and Wtdexp The weighted expected claims for each individual strategy. Figure 12 is a table of Model Criterion 80 and Model Rule Group 138 for credit scores. Other criteria can also be selected, based on the type of financial instrument and special offer conditions or any other needs or preferences of the bidder. This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) -52- 554276 A7 B7 V. Description of the invention (50) Figure 1 3 is similar to the tree diagram 6 6 (see the lower part of Figure 2) More detailed tree diagram 260. In Figure 13 the isolation is based on (a) whether it is safe, (b) whether it is revolving, and (c) whether the final payment is zero. The result is six clusters 262, 264, 266, 268, 270, 272, generally known as " shaker trees. &Quot; Figure 14 shows an exemplary system 300 according to one embodiment of the invention. System 300 It includes at least one computer configured as a server 302 and most other computers 304 coupled to the server 302 to form a network. In one embodiment, the computer 304 is a user system including a web browser, and the server 302 Is accessible to the computer 304 via the Internet. In addition, the server 302 is a computer. The computer 304 is interconnected to the Internet through a number of interfaces, which include networks such as a local area network (LAN) ) Or Wide Area Network (WAN)), dial-up connections, cable modems and special high-speed ISDN lines. The computer 304 can be any device capable of connecting to the Internet, including an internet phone or other network connectable device (including Wireless network and satellite). The server 302 includes a database server 3 06 connected to a central database 76 (also shown in FIG. 2), which contains data describing a combination of assets. In one embodiment , The central database 76 is stored on the database server 306 and accessed by a user on one of the computers 304 through one of the computers 304 to log in to the server subsystem 302. In another embodiment, the central database 76 Stored remotely from the server 302. The server 302 is further configured to receive and store the information of the asset valuation method described above. Although the system 300 is described as a network-connected system, the methods described herein to facilitate review And the use of algorithms and portfolio assets can also be applied to this paper size China National Standard (CNS) Λ4 specification (210X297 mm) (Please read the precautions on the back before filling this page)

, 1T Printed by the Intellectual Property of the Ministry of Economic Affairs, the Goods and Workers Consumer Cooperatives -53- Fan and Shenjingwei A-B Special Interests Please apply for a clear explanation of TTT1. Renovation of technical skills, technical practice, Xiang Zhang, Xiwei, printed by the Intellectual Property Bureau of the Ministry of Economic Affairs, M Industrial Consumption Cooperative, 554276 A7 B7 V. Invention Description (51) was implemented as an independent computer system not connected to other computers. Although the present invention has been described by specific embodiments, those paper sizes are applicable to the Chinese National Standard (CNS) Λ4 specification (210 × 297 mm) (Please read the precautions on the back before filling this page)

-54-

Claims (1)

554276 A8 B8 C8 D8 6. Scope of Patent Application 1 · A method of determining the winning bid of a sealed bid auction at the best price (3 2). The method includes the following steps: (Please read the precautions on the back before filling in this Page) Determine the distribution of possible bid prices from competing bidders; choose a bid price; randomly sample other bid prices to generate a possible auction scenario; and determine the probability of a naked auction relative to the selected bid price. 2. The method (3 2) of item 1 of the patent application range, wherein the step of randomly sampling other bid prices further includes the step of using a repeated sampling technique to generate a distribution of auction results. 3. The method (32) of item 2 of the scope of application · patent, wherein the step using an iterative sampling technique further includes the step of analyzing using Monte Caiio. 4. The method (32) of the first patent application scope, further comprising the following steps: selecting different bid prices; the consumer cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs prints random sampling of other bid prices to generate a possible auction scenario; and Determines the probability of a naked auction relative to the selected bid price. 5. The method (32) of item 4 of the scope of patent application, wherein the step of randomly sampling the bid price further includes the step of using an iterative sampling technique to generate a distribution of auction results. 6. The method (32) of claim 5 in the scope of patent application, wherein the step using an iterative sampling technique further includes the step of analyzing using M o n t e Ca a · 10. -55- This paper uses the Chinese National Standard (CNS) A4 size (210 X 297 mm) 554276 A8 B8 C8 D8 6. Scope of patent application (please read the notes on the back before filling this page) 7. If you apply The method (32) of the first scope of the patent, wherein the step of determining the distribution of possible bid prices from the competitive bidders further includes a step of determining the financial capabilities of at least one possible competitive bidder. 8. The method (32) of the first patent application range, wherein the step of determining the distribution of possible bid prices from competing bidders further includes the step of compiling market rules and contracts into computerized business rules suitable for simulation 〇9. The method (3 2) of item 1 of the scope of patent application, wherein the step of determining the distribution of possible bid prices from the competitive bidders further includes the potential competition, market power, forecasted budgets, priorities, risks, and The step of compiling at least one of the revenue trade-offs into a preference matrix. 10. The method (3 2) of item 1 in the scope of patent application, wherein the step of determining the distribution of possible bid prices from the competitive bidders further includes according to the asset portion (70, 72, 74) preferred by the competitive bidders. ) Pattern of steps to compile the past bid history of competing bidders. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. A system (300) of a sealed bid auction (300) that determines the asset part (70, 72, 74) of the combined asset (12) at the best price. This system Including: a computer, which is assembled as a server (302) and further equipped with a database (76) of portfolio assets; at least one user system (304), which is connected to the server via a network ' The server is assembled to determine the distribution of possible bid prices from competing bidders, select a bid price, randomly sample other bid prices to generate a possible auction scenario, and determine that naked auctions are applicable relative to the selected paper size China National Standard (CNS) A4 Specification (210X297 mm) -56-: 554276 A8 B8 C8 D8 (Please read the notes on the back before filling out this page) 12. If the system (300) of the 11th scope of the patent application, the server (302) is assembled to use an iterative sampling technology to generate auction results. distributed. 13. The system (300) according to item 12 of the patent application scope, wherein the device (302) is assembled to use Monte Carlo analysis as a repetitive sampling technique. 14. If the system (300) of item 11 of the scope of patent application is applied, the server 12 (302) is equipped with: selecting different bid prices; randomly sampling other bid prices to generate possible auction scenarios; and determining Probability of a naked auction relative to the selected bid price. 15. The system (300) according to item 14 of the patent application scope, wherein the server (302) is equipped to use an iterative sampling technique to generate a distribution of auction results. 16. The system (300) according to item 15 of the patent application scope, wherein the server (302) is equipped to use Monte Carlo analysis as a repetitive sampling technique. 17. If the system (300) of the 11th scope of the patent application, the servo consumer printer (302) of the Intellectual Property Bureau of the Ministry of Servicing Economy is assembled to determine the financial capabilities of at least one possible bidder. 08. For example, the system (300) of the scope of patent application, where the server (302) is assembled to compile market rules and contracts into computerized business regulations. 3 0 0), where the server (302) is assembled to apply the potential competition, market power, forecasted budget, excellent paper size to Chinese National Standard (CNS) A4 specifications (210X297 mm) · 57-554276 A8 B8 C8 D8 6. At least one of the order of patent application scope, risk and return tradeoffs is compiled into a preference matrix. (Please read the notes on the back before filling out this page) 20. If the system (300) of the 11th scope of the patent application, the server (302) is assembled according to the asset portion (70 , 72, 74) to compile past bid history of competing bidders. 2 1 · — A computer (38) that determines the winning bid of the asset part (70, 72, 74) of the combined asset (1 2) at the best price, the computer containing a database (76) of one of the combined assets, the computer It is regulated to: determine the distribution of possible bid prices from competing bidders; select a bid price; 'randomly sample other bid prices to generate a possible auction scenario; and determine the probability of a naked auction relative to the selected bid price. 22. If the computer (38) of the scope of patent application is applied, it is regulated to use an iterative sampling technique to generate the distribution of auction results. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 23. If the computer (38) in the scope of application for patent No. 22 is applied, it is analyzed by the method of repetitive sampling using M ο n t e C a r 10. 24. If the computer (38) of the scope of patent application is applied, it is regulated to select different bid prices; randomly sample other bid prices to generate a possible auction scenario; and decide the probability of winning the auction relative to the selected bid price. 25. If the computer (38) in the scope of the patent application is applied, it shall be subject to the Chinese National Standard (CNS) A4 specification (210X297 mm) at this paper standard. -58-554276 A8 B8 C8 D8 An iterative sampling technique to produce a distribution of auction results. 26. For the computer (38) under the scope of application for patent, it is regulated to use Monte Cado analysis as the repeated sampling technique. 27. If the computer (38) of the scope of patent application is applied for, it is regulated to determine the financial capabilities of at least one possible bidder. 28. For a computer (38) under the scope of application for a patent, it is regulated to codify market rules and contracts into business rules. 29. If the computer (38) of the scope of patent application is applied for, it is regulated to compile at least one of potential competition, market power, forecasted budget, priority order, risk, and benefit trade-off into a preference matrix. 30. If the computer (38) of the scope of patent application is applied, it is regulated to compile the past bidding history of the competitive bidder based on the understanding of the type of asset part (70, 72, 74) preferred by the competitive bidder. . (Please read the precautions on the back before filling out this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 59 The paper size is in accordance with China National Standard (CNS) A4 (210X297 mm)