TWI489407B - System and method for generating decentralized options trading strategy - Google Patents

Description

Decentralized option trading strategy generation system and method

The present invention relates to a decentralized option transaction strategy generation system and method, and in particular to a decentralized option transaction strategy that can generate decisions containing different weights for profit dispersion decentralized option transactions. The strategy generation system and method are related.

In recent years, due to skyrocketing prices, investment in wealth management has become an important conduit for wealth accumulation, given that public wages have not increased and deposit rates have fallen. Option trading has the characteristics of low cost, low risk and multiple trading portfolio strategies, which has become a better choice for many investors.

The choice of the Taiwan index is based on the index option of Taiwan's futures. It is one of the emerging products in the Taiwan trading market in recent years. Various market positions can be optimally compensated by a combination of trading strategies with different options. Therefore, Investors can make different strategies for different trends to get the best profit. However, precisely because of the variability of its trading strategy, investors are easily confused, which in turn leads to capital investment and investment efficiency. In addition, because investors are vulnerable to various external environmental factors that affect their psychological emotions, they may suffer irrational investment and suffer losses. For example, investors may hear from the wrong news, and invest too much money may fall. The subject matter leads to a loss.

Program trading refers to the use of technical indicators commonly used in the market, using computer software to write them into the system, and using the program to calculate the trading opportunities for trading. Because program trading can replace investors for complex calculations, And to overcome the emotional or psychological reaction of investors to avoid irrational trading behavior, so more and more investors tend to use program trading to invest. The common common transaction strategy generators are composed of a single logical judgment, for example, a single technical indicator, multiple sets of technical indicators, and simultaneously based on chip and technical aspects to form a single group of option transaction decisions. Although the trading decisions generated by the prior art may be completed by multiple sets of technical indicators or different trends and patterns, the decisions they generate still account for 100% of the weight and do not conform to the concept of risk dispersion. In other words, when a decision made by a prior art program transaction is misjudged, the investor's investment in accordance with his instructions will suffer all losses.

Therefore, it is necessary to invent a system or method that can generate an option trading strategy with a diversified risk effect to help investors conduct option trading in a risk-disaggregated situation.

One aspect of the present invention is to provide a system that can generate an option transaction strategy with a diversified risk effect to solve the problems of the prior art.

According to a specific embodiment, the distributed option transaction strategy generation system of the present invention comprises a database, an indicator module, a genetic algorithm module, a gray correlation analysis module, and a decision tree module, wherein the indicator module is connected to the database, and the gene The calculus module and the gray correlation analysis module are respectively connected to the indicator module and the decision tree module. A plurality of option transaction data can be stored in the database, and the indicator module can calculate a plurality of technical indicators, volatility and moving averages based on the transaction data of the options. The gene calculus module can receive technical indicators from the indicator module and perform genetic calculation on the gene to obtain at least one first selection. Trading strategy. In addition, the gray correlation analysis module can receive the volatility and the moving average, and perform gray correlation analysis on the two to obtain at least one second option trading strategy. Finally, the decision tree module can receive the first option transaction strategy and the second option transaction strategy from the genetic algorithm module and the gray correlation analysis module, and separately apply the first option transaction strategy and the second option transaction strategy respectively. Assign weights to export decentralized trading strategies.

In the specific embodiment, the genetic algorithm can save the selection transaction strategy with good performance and eliminate the poorly performing option transaction strategy. In addition, the gray correlation analysis method can also generate the performance transaction with good performance. Strategy. The decision tree module can respectively give the above-mentioned option trading strategy weights, so each option trading strategy can be combined with its weight to form a decentralized option trading strategy to provide investors with the effect of diversifying investment risks.

Another aspect of the present invention is to provide a method for generating an option transaction strategy with a risk-spreading effect that solves the problems of the prior art.

According to another specific embodiment, the method for generating a distributed option transaction strategy of the present invention comprises the steps of: first, collecting a plurality of option transaction data; and then, calculating a plurality of technical indicators, volatility, and Moving average; performing genetic calculation on the above technical indicators to obtain at least one first option trading strategy; performing gray correlation analysis on the volatility and the moving average to obtain at least one second option trading strategy; and, An option trading strategy and a second option trading strategy respectively assign weights, and then output a distributed option trading strategy.

In the method of the specific embodiment, performing genetic calculation on technical indicators The law can generate a well-executed option trading strategy and eliminate the poorly performing option trading strategy. In addition, gray correlation analysis of volatility and moving averages can also produce a well-performing option trading strategy. The above-mentioned option trading strategy is assigned its weight, and each option trading strategy is combined with its weight to generate a decentralized option trading strategy, and this decentralized option trading strategy can provide investors with the effect of diversifying investment risk.

The advantages and spirit of the present invention will be further understood from the following detailed description of the invention.

Referring to FIG. 1, FIG. 1 is a schematic diagram of a distributed option transaction strategy generation system 1 according to an embodiment of the present invention. As shown in FIG. 1 , the distributed option transaction strategy generation system 1 of the specific embodiment includes a database 10 , an indicator module 12 , a gene calculation module 14 , a gray correlation analysis module 16 , and a decision tree module 18 . The indicator module 12 is connected to the database 10, and the gene calculation module 14 and the gray correlation analysis module 16 are respectively connected to the indicator module 12 and the decision tree module 18. The decentralized option trading strategy generation system 1 can be used to generate a decentralized option trading strategy for the user or investor to make the option transaction and provide the effect of diversifying the investment risk.

In this embodiment, the database 10 may further include a collection unit 100 and a storage unit 102. The collection unit 100 can be used to continuously collect the latest selection market information published around the world, and the storage unit 102 can store the data therein. The indicator module 12 is connected to the database 10 to receive the option market data stored in the storage unit 102, and Based on the calculation of a number of technical indicators, volatility and moving average. The technical indicators may include technical indicators commonly used in the prior art, such as MA, AR, BIAS, BR, KD, MACD, MTM, OBV, OSC, PSY, RSI, TRIX, WMS, and WRSI. The volatility can be divided into historical volatility and implied volatility. In the specific embodiment, the metric volatility is calculated by the indicator module 12, but it is not limited to this in practice.

The indicator module 12 is connected to the genetic algorithm module 14 and the gray correlation analysis module 16 respectively, so that the gene calculation module 14 can receive the technical indicators calculated by the indicator module 12, and the gray correlation analysis module 16 can be The volatility and moving average calculated by the indicator module 12 are received. The gene calculus module 14 can perform gene calculation according to the received technical indicators, and the gray correlation analysis module 16 can perform gray correlation analysis on the volatility and the moving average.

In the specific embodiment, the gene calculus module 14 further includes a copy unit 140, a mating unit 142, an abrupt unit 144, and a first policy generating unit 146. Before performing the genetic calculation on the above technical indicators, it is necessary to randomly generate a plurality of chromosomes as the initial ethnic group, and start the subsequent genetic behavior with the initial ethnic group. The gene calculus module 14 can randomly select a certain number of technical indicators to form chromosomes of these initial groups. For example, 10 technical indicators are randomly selected from the above various technical indicators to form a chromosome, and repeated 10 times. A total of 10 groups of chromosomes. Each technical indicator is a gene in the initial ethnic chromosome and can be used for genetic calculation.

In this embodiment, one technical indicator can be composed of 10 Bit codes, and the chromosome can be composed of 10 technical indicators (genes). The color body can be encoded by 100 Bits. Please note that the number of technical indicators contained in a chromosome can be determined according to the needs of the user or the designer. Similarly, the number of bits used in each technical indicator can also be determined according to the needs of the user or the designer, and is not limited. In this specific embodiment. Since the number of bits used in the technical indicators in each chromosome is the same, when genetic behavior is performed, the technical indicators in each chromosome can be directly exchanged.

The replication unit 140, the mating unit 142, and the mutation unit 144 of the gene calculus module 14 can be used to perform genetic manipulation of chromosomes. The copy unit 140 is used for copying, and its purpose is to retain the better-performing features to the children, so as to avoid the evolution of the advantages and disappear over time. The copying unit 140 copies N chromosomes in the initial population and replaces the other N chromosomes, wherein the copied chromosome is a chromosome that performs better, and the replaced chromosome is the one with the worst profit. The profitability of the chromosome can be calculated by an external computing platform, or directly by the genetic algorithm module 14. Taking the chromosomes of the above 10 groups of initial groups as an example, after calculating the profit performance of each chromosome, two groups of chromosomes with the best profit and two groups with the worst profitable chromosomes can be found, and 2 After the best profitable chromosomes were copied, the two groups were replaced with the worst profitable chromosomes. Since the copying behavior is to select the best performing individual to replace the worst performing individual, this behavior can also be called "choice." The copying unit 140 performs not only the copying operation on the chromosomes of the initial population, but also the chromosomes of the respective generations.

The mating unit 142 can be used to perform the mating behavior between chromosomes, which randomly pairs the chromosomes of the mother pair two-two, and selects two-point mating to exchange the genes corresponding to the positions of the two chromosomes (technical index), and then shape Into two new progeny chromosome individuals, that is, complete the mating process. In addition, the mutation unit 144 determines that a plurality of genes in one generation will be mutated according to a predetermined mutation rate. For example, if the mutation rate is set to 0.01, 10 of the 10 sets of chromosomes can be calculated to have mutations, so the mutation unit 144 can randomly select a technical index for coding conversion as a mutation event.

The replication unit 140, the mating unit 142, and the mutation unit 144 can be controlled by the first strategy generation unit 146 to generate chromosomes of the next generation after at least one replication, mating, and mutation of each chromosome, respectively. After a certain number of cycles, for example, 1000 or 1500 cycles, the resulting final generation chromosome may be output by the first policy generation unit 146 as a first option transaction strategy. Taking the chromosomes of 10 groups of initial groups as an example, the first strategy generating unit 146 can output 10 sets of final generation chromosomes generated after 100 times of replication, mating and mutation, as 10 first option trading strategies.

Referring to FIG. 1 again, the gray correlation analysis module 16 further includes a normalization unit 160, a difference calculation unit 162, an association coefficient calculation unit 164, an association degree calculation unit 166, and a second policy generation unit 168. The normalization unit 160 can normalize the volatility and the moving average received by the gray correlation analysis module 16 from the indicator module 12, in detail, it is found in the data column of the volatility and the moving average. The reference point is normalized, and all the data is normalized according to the reference point, and the normalized value is between 0 and 1. The normalization unit 160 may normalize the volatility and the moving average to form a first normal sequence and a second normal sequence, respectively.

The difference calculation unit 162 can be configured for the first normal number column and the second normal number column Calculating the corresponding difference value of the corresponding data point, and then the correlation coefficient calculation unit 164 may calculate the correlation coefficient of the corresponding point in the first regular number column and the second normal number column according to the corresponding difference value calculated by the difference calculating unit 162. Moreover, the correlation degree calculation unit 166 may obtain an average value of each correlation coefficient, and the average value is the degree of association between the two normalized series. The second policy generating unit 168 may generate a second option transaction strategy according to the degree of association and each of the correlation coefficients. For example, if the correlation coefficient of the day (the corresponding point of the first normal number column and the second normal number column) is high, In the case of the degree of association, it is possible to enter the field with the trend; if the correlation coefficient of the day is lower than the correlation degree, it is not suitable for the approach operation.

As described above, the genetic algorithm module 14 and the gray correlation analysis module 16 respectively generate one or more first option transaction strategies and second option transaction strategies, and then the decision tree module 18 can be self-generating models. The group 14 and the gray correlation analysis module 16 respectively receive the first option transaction strategy and the second option transaction strategy, and respectively give weights. The decision tree module 18 may calculate the weights of the first option transaction strategy and the second option transaction strategy according to the actual profit and loss (Actual Gain, AG) and the Trading Profit Percent (TPP), and assign weights according to the weights. As a result, each of the first option transaction strategy and the second option transaction strategy are combined to output a decentralized option transaction strategy. Therefore, the decentralized option transaction strategy output by the distributed option transaction generation system 1 of the specific embodiment is composed of various option transaction strategies with better profit performance, and each option transaction strategy is It also has a weight distribution, which allows investors to make choice investment according to the weight ratio, which has the effect of diversifying risks.

Please refer to FIG. 2 , which illustrates a specific embodiment of the present invention. A flow chart of the steps of the method for generating a distributed option transaction strategy, please note that the steps of the method of the specific embodiment can be performed by the distributed option transaction policy generation system 1 of the above specific embodiment. The decentralized option trading strategy generation method can generate a decentralized option trading strategy, which not only has good profit performance, but also provides investors with the effect of diversifying risks.

As shown in FIG. 2, the method for generating a distributed option transaction strategy in the specific embodiment includes the following steps: in step S20, collecting a plurality of option transaction data; in step S22, calculating a plurality of pieces according to the option transaction data. a technical indicator, a volatility, and a moving average; in step S24, genetically calculating a technical indicator to obtain at least one first option trading strategy; and in step S26, performing gray relational analysis on the volatility and the moving average, Obtaining at least one second option transaction strategy; and, in step S28, using the decision tree to assign weights to each of the first option transaction strategy and the second option transaction strategy, respectively, to output a decentralized option transaction strategy.

In the specific embodiment, steps S20 to S28 can be performed through the database 10, the indicator module 12, the genetic algorithm module 14, the gray correlation analysis module 16, and the decision tree module 18 of the specific embodiment, respectively. The detailed execution process has been described above, and thus will not be described herein. Therefore, the decentralized option trading strategy generation method of the specific embodiment can generate a plurality of option trading strategies with good profit performance by using a genetic algorithm and a gray correlation analysis method, and then using a decision tree method. These option trading strategies assign weights, which in turn output a decentralized option trading strategy with a diversified risk function.

Please refer to FIG. 3, and FIG. 3 is a detailed step of step S24 of FIG. flow chart. As shown in FIG. 3, step S24 of this embodiment further includes the following steps: in step S240, a predetermined number of technical indicators are selected by random numbers to form a plurality of chromosomes; in step S242, the chromosomes are copied. a first chromosome, and the first chromosome after replication is used to replace the second chromosome of the chromosomes; in step S244, the chromosomes are mated to each other to generate a plurality of progeny chromosomes; in step S246, the chromosome is induced to undergo mutation; In step S248, the chromosome is repeatedly copied, mated, and mutated to obtain at least one final generation chromosome as the first option trading strategy.

In the above step S242, the first staining system is obtained by calculating the profit performance of each chromosome and selecting the N chromosomes having the best profit performance. Similarly, the second dyeing system calculates each chromosome by respectively. Profit performance, and select the N chromosomes with the worst profit performance, so the N first chromosomes can be replicated in this step to replace the N second chromosomes, respectively, to complete the replication in the gene algorithm. action. After the genetic algorithm, step S24 will produce a plurality of first option trading strategies (final generation chromosomes) with good profitability.

Please refer to FIG. 4, which is a flow chart showing the detailed steps of step S26 of FIG. As shown in FIG. 4, step S26 further includes the following steps: in step S260, normalizing the volatility and the moving average to respectively generate the first normalized sequence and the second normalized sequence; in step S262, calculating Corresponding at least one corresponding difference between a normal sequence and a second normal sequence; in step S264, calculating, according to the at least one corresponding difference, respectively, the association between the corresponding points in the first regular sequence and the second normal sequence a coefficient; in step S266, calculating a degree of association between the first normalized sequence and the second normalized sequence according to each correlation coefficient; and, in step S268, according to The degree of association and the correlation coefficients are used to generate a second option trading strategy. Similarly, step S26 is capable of generating a second option trading strategy with good profitability through gray correlation analysis.

In summary, the distributed option trading strategy generating system and method of the present invention can process the technical indicators, the volatility and the moving average calculated by the option transaction data through the genetic algorithm and the gray correlation analysis method, and In order to obtain a plurality of option trading strategies with good profit performance. In addition, the distributed option trading strategy generating system and method of the present invention utilizes a decision tree method to assign weights to each of the above-mentioned option trading strategies, and outputs a distributed option trading strategy according to the result of the weighting allocation. . In addition to various strategies with good profitability, this decentralized option trading strategy can also weight these strategies and further provide investors with the effect of risk diversification.

The features and spirit of the present invention will be more apparent from the detailed description of the preferred embodiments. On the contrary, the intention is to cover various modifications and equivalents within the scope of the invention as claimed. Therefore, the scope of the patented scope of the invention should be construed as broadly construed in the

1‧‧‧Distributed option trading strategy generation system

10‧‧‧Database

12‧‧‧ indicator module

14‧‧‧Genetic calculation module

16‧‧‧ Gray Correlation Analysis Module

18‧‧‧Decision Tree Module

100‧‧‧Collection unit

102‧‧‧ storage unit

140‧‧‧Replication unit

142‧‧‧ mating unit

144‧‧‧mutation unit

146‧‧‧First Strategy Generation Unit

160‧‧‧Formalization unit

162‧‧‧ difference calculation unit

164‧‧‧correlation coefficient calculation unit

166‧‧‧Affinity calculation unit

168‧‧‧Second Strategy Generation Unit

S20~S28, S240~S248, S260~S268‧‧‧ Process steps

1 is a schematic diagram showing a decentralized option transaction strategy generation system in accordance with an embodiment of the present invention.

2 is a flow chart showing the steps of a method for generating a distributed option transaction strategy according to an embodiment of the present invention.

FIG. 3 is a flow chart showing the detailed steps of step S24 of FIG.

FIG. 4 is a flow chart showing the detailed steps of step S26 of FIG.

1‧‧‧Distributed option trading strategy generation system

10‧‧‧Database

12‧‧‧ indicator module

14‧‧‧Genetic calculation module

16‧‧‧ Gray Correlation Analysis Module

18‧‧‧Decision Tree Module

100‧‧‧Collection unit

102‧‧‧ storage unit

140‧‧‧Replication unit

142‧‧‧ mating unit

144‧‧‧mutation unit

146‧‧‧First Strategy Generation Unit

160‧‧‧Formalization unit

162‧‧‧ difference calculation unit

164‧‧‧correlation coefficient calculation unit

166‧‧‧Affinity calculation unit

168‧‧‧Second Strategy Generation Unit

Claims (6)

A decentralized option transaction strategy generation system includes: a database for storing a plurality of option transaction data; an indicator module coupled to the database for receiving the option transaction data And calculating a plurality of technical indicators and a volatility and a moving average according to the option transaction data; a genetic algorithm module connected to the indicator module, the genetic algorithm module is configured to receive The technical indicators, and a predetermined number of technical indicators are selected from the technical indicators to form a plurality of chromosomes, the genetic algorithm module further comprising: a copy unit for copying the chromosomes N best profitable chromosomes, and the N best profitable chromosomes after replication replace the N worst profitable chromosomes in the chromosomes, N is a positive integer; a mating unit, Used to mate the chromosomes to each other to generate a plurality of progeny chromosomes; a mutation unit for inducing mutations in the chromosomes; and a first strategy generation a unit connected to the replication unit, the mating unit, and the mutation unit, wherein the first strategy generation unit is configured to control the replication unit, the mating unit, and the mutation unit to repeatedly perform operations such as copying, mating, and abrupt Obtaining at least one final generation chromosome as at least one first option trading strategy; a gray correlation analysis module connected to the indicator module, the gray correlation analysis module is configured to receive the volatility and the moving average And The volatility and the moving average are subjected to gray correlation analysis to obtain at least one second option trading strategy; and a decision tree module connected to the genetic algorithm module and the gray correlation analysis module, the decision The tree module is configured to respectively assign weights to the at least one first option transaction strategy and the at least one second option transaction strategy to output a decentralized option transaction strategy. The decentralized option transaction strategy generation system of claim 1, wherein the gray correlation analysis module further comprises: a normalization unit configured to separately normalize the volatility and the moving average And generating a first normalized sequence and a second normalized sequence; a difference calculating unit configured to calculate at least one corresponding difference between the first normal sequence and the second normal sequence; a correlation coefficient calculation unit, configured to calculate, according to the at least one corresponding difference, an association coefficient of each point corresponding to each other in the first regular number column and the second normal number column; The method is configured to calculate an association degree between the first normalized sequence and the second normalized sequence according to the correlation coefficients; and a second policy generating unit, configured to use the correlation degree and The correlation coefficients are used to generate the at least one second option transaction strategy. The decentralized option transaction strategy generation system according to claim 1, wherein the decision tree module is based on actual profit (Actual Gain, AG) and Trading Profit Percent (TPP). The at least one first option transaction strategy and the at least one second option transaction strategy respectively assign weights. A method for generating a distributed option trading strategy, comprising the steps of: collecting a plurality of option transaction data; calculating a plurality of technical indicators, a volatility rate, and a moving average according to the option transaction data; In the technical indicators, a predetermined number of technical indicators are selected by random numbers to form a plurality of chromosomes; the profit performance of the chromosomes is calculated to obtain N best profitable chromosomes and N worst profitable chromosomes. , wherein N is a positive integer; the N best profitable chromosomes are replicated, and the N best profitable chromosomes after replication are substituted for the N worst profitable chromosomes; the chromosomes are mated to each other Generating a plurality of progeny chromosomes; inducing the chromosomes to produce mutations; repeating the replication, mating, and mutation of the chromosomes to obtain at least one final generation chromosome as at least a first option trading strategy; the volatility and the movement The average line performs gray correlation analysis to obtain at least one second option transaction strategy; and utilizes the decision tree to First selection trading policy and the at least a second select trading strategies are assigned weights so as to output a distributed trading strategy options. The method for generating a distributed option trading strategy as described in claim 4, further comprising the steps of: normalizing the volatility and the moving average to generate a first normalized sequence and a first Two normalized series; Calculating at least one corresponding difference between the first normal number column and the second normal number column; calculating, according to the at least one corresponding difference value, points corresponding to each other in the first normal number column and the second normal number column a correlation coefficient; calculating an association degree between the first normalization sequence and the second normalization sequence according to the correlation coefficients; and generating the at least one according to the correlation degree and the correlation coefficients Two options trading strategy. The method for generating a distributed option transaction strategy as described in claim 4, further comprising the steps of: at least one based on actual profit (Actual Gain, AG) and Trading Profit Percent (TPP). The first option transaction strategy and the at least one second option transaction strategy respectively assign weights.