JP2005071212A - Market impact estimation system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a market impact estimation system which can perform estimation of exact market impact. <P>SOLUTION: In the market impact estimation system 2, transaction data of securities (stocks) memorized in a transaction data memory section 8 are memorized for each brand, data indicating excess demand in stock exchange in a day are memorized for each brand in an excess demand data memory section 10, data indicating price change rate of share price in market place in a day computed based on transaction data are memorized in a price change rate data memory section 12. Parameters used for estimation of market impact computed based on cumulative values of excess demand for the predetermined period and the price change rate of the same predetermined period, and average spread rate computed based on the transaction data are memorized in a data memory section 14. Estimated values of market impact computed based on parameters and average spread rate are displayed in a display section 18. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a market impact estimation system that estimates a market impact that is a price change that occurs in a market based on a transaction of securities.

  Market impact means that in a trading market where securities, for example, stocks are traded, the trading price of the market is moved in an unfavorable direction by one's own trading behavior, that is, a change in the stock price caused by one's own trading. Point to. For example, if a stock currently traded at 100 yen is raised to 105 yen by its own transaction, a change in price is expressed as “a market impact of 5 yen has occurred”.

  Conventionally, estimation of market impact has been performed to predict in advance how much a price change will occur when a large amount of stock is traded. The estimation of the market impact was based on the average number of shares of sellers and buyers waiting at the best price (the most concessional price) in the trading market for buying and selling stocks. That is, in the market, for example, an investor who thinks that a stock traded at 100 yen may sell for 101 yen or more is waiting at a price that he / she can sell. Investors who are willing to buy below are also waiting at their respective prices. In such a case, the best concession price among sellers and buyers, for example, 101 yen for sellers and 99 yen for buyers is called the best price. The estimation of the conventional market impact is based on the average number of shares waiting at the best price. For example, if the past average value of the number of shares waiting at the best price is 5,000, 5,000 It was estimated that the stock price would change by 1 yen in the trading of stocks, and trading 50,000 shares would produce a market impact of 10 yen as 50,000 / 5,000.

  Note that there is a market impact model using a plate interval for estimating the market impact (see, for example, Non-Patent Document 1). In this market impact model, the market impact is estimated using the difference between the best price and the second best price as an index representing the “board interval”, that is, the “density” of the limit price on the board. Therefore, an accurate market impact can be estimated for a stock with low liquidity, and a market impact of a stock with high liquidity can be accurately estimated by reflecting the relationship between the bid price and the price level.

http://www.nqi.co.jp/report/0011/nakatuka.pdf

  By the way, for example, when trying to purchase a large amount of stock, if the number of stocks waiting at the best price is very small compared to the number of stocks planned to be purchased, for example, about 1%, usually time Stock transactions are conducted while paying attention to avoiding price changes as much as possible. That is, by conducting transactions over time, it is possible to purchase shares at a generally better price than attracting new sellers and purchasing all of the planned number of shares to be purchased at once. However, since the conventional market impact estimation is performed assuming an instantaneous transaction, that is, a transaction in which a large amount of stock is purchased at once, accurate market impact estimation cannot be performed.

  In general, it is necessary to estimate the market impact when the transaction volume is large. For example, like an institutional investor, trade in stocks several times as large as the daily volume of the stock. In such cases, estimation of market impact is particularly important. In such a case where the transaction volume is large, even if the market impact is estimated based on the small number of stocks waiting for the best quotation, assuming that the transaction is instantaneous, Was not reflected, and a reliable estimate could not be obtained.

  An object of the present invention is to provide a market impact estimation system capable of accurately estimating a market impact by reflecting an actual transaction of securities.

  The market impact estimation system according to claim 1 includes transaction data storage means for storing transaction data for each brand of securities, and a predetermined period of a predetermined brand based on the transaction data stored in the transaction data storage means. The excess demand accumulated value calculating means for calculating the accumulated value of excess demand in the transaction and the price change rate within the predetermined period of the predetermined brand based on the transaction data stored in the transaction data storage means Price change rate calculating means, sample data creating means for creating a plurality of sample data with a set of accumulated value of excess demand within the predetermined period and price change rate within the predetermined period, and the sample data creating means A parameter calculator that calculates the parameters used to estimate market impact based on multiple sets of sample data created by And an average spread ratio calculating means for calculating an average spread ratio of the predetermined brand based on transaction data stored in the transaction data storage means, a parameter calculated by the parameter calculating means, and the average spread ratio Market impact estimated value calculating means for calculating an estimated value of market impact based on the average spread ratio calculated by the calculating means is provided.

  According to the market impact estimation system according to claim 1, the market impact is calculated based on a plurality of sets of sample data created by setting a cumulative value of excess demand within a predetermined period and a price change rate within a predetermined period as one set. A parameter used for estimation is calculated, and an estimated value of market impact is calculated based on the calculated parameter and the average spread rate. Therefore, unlike traditional market impact estimation, which assumes that market trading is performed at the same time, assuming that trading transactions for securities, for example, stocks are traded at once, a trading mode in which stock trading transactions are performed over a predetermined time is assumed. The market impact can be estimated. Therefore, it is possible to estimate an accurate market impact that appropriately reflects an actual transaction mode.

  The market impact estimation system according to claim 2, wherein the excess demand accumulated value calculation means accumulates the value of excess demand for each day within the predetermined period, thereby obtaining the accumulated value of excess demand in the predetermined period. It is characterized by calculating.

  According to the market impact estimation system according to the second aspect, since the value of excess demand for each day within a predetermined period is accumulated, the volume of the volume used for estimation can be increased. Accordingly, it is possible to estimate the market impact by appropriately reflecting the actual situation of a large amount of transactions performed by institutional investors.

  The market impact estimation system according to claim 3, wherein the parameter calculation unit calculates the parameter by performing regression analysis based on a plurality of sets of sample data created by the sample data creation unit. And

  According to the market impact estimation system according to claim 3, the parameter is calculated by regression analysis based on a plurality of sets of sample data in which a cumulative value of excess demand and a price change rate within a predetermined period are set as one set. The market impact can be estimated assuming a transaction mode close to the actual transaction mode.

  The market impact estimation system according to claim 4, wherein the average spread ratio calculation means calculates the average spread ratio by weighting each of the plurality of spread ratios corresponding to the period during which each spread ratio was indicated. It is characterized by doing.

  According to the market impact estimation system according to claim 4, the average spread rate, that is, the spread rate indicating the distance between the buyer who makes the most concessions in price (the best bid price) and the seller (the best bid price), The average spread rate is calculated by weighting corresponding to the period. In other words, since the best bid price and the best bid price change from moment to moment, the spread rate also changes from moment to moment, so the average spread rate is calculated by weighting corresponding to the time when the spread rate value was constant. By doing so, it is possible to estimate the market impact appropriately reflecting price fluctuations.

  According to this invention, since the parameter used when estimating the market impact is calculated using the relationship between the excess demand and the price change rate in the predetermined period, a transaction mode close to the actual transaction mode is assumed. Market impact can be estimated. In other words, excess demand does not occur instantaneously, but occurs while buying and selling within a certain period of time. By calculating the parameters, it is possible to estimate the market impact assuming a transaction mode that takes time. Therefore, it is possible to accurately estimate how much market impact will occur when the planned transaction is actually performed, and it is estimated as useful data for conducting appropriate securities transactions. Market impact values can be used.

  The best mode for carrying out a market impact estimation system according to the present invention will be described below with reference to the drawings. In the following, an example will be described in which the market impact caused by trading stocks as securities is estimated.

  FIG. 1 is a block configuration diagram of a market impact estimation system according to an embodiment. As shown in FIG. 1, the market impact estimation system 2 includes a data control unit 4 that controls data processing in the market impact estimation system 2. The data control unit 4 includes a communication control unit 6 that controls communication with a stock exchange system (not shown) installed in the stock exchange, and a communication control unit 6 from the stock exchange system. A transaction data storage unit 8 is connected to store transaction data relating to the acquired stock sales transaction. Further, the data control unit 4 stores an excess demand data storage unit 10 for storing data indicating excess demand for each stock brand and data indicating a price change rate of the stock price for each stock brand every day. A price change rate data storage unit 12 is connected. Further, the data control unit 4 includes a data storage unit 14 for storing estimated market impact values, a keyboard 16 used when inputting predetermined data to the data control unit 4, and the calculated market impact. A display unit 18 for displaying estimated values and the like is connected.

  The transaction data storage unit 8 stores data such as tick-by-tick stock prices (stock prices every moment) of stocks to be traded for a predetermined period, for example, one year. The tick-by-tick stock price includes data indicating how many shares are desired to be sold or purchased at which price range at any time as shown in Figure 2 below. It is. Further, the transaction data is stored in the transaction data storage unit 8 for each date corresponding to the stock brand.

  Further, the excess demand data storage unit 10 subtracts the demand volume of the seller-led transaction from the volume of the buyer-led transaction in the daily volume, that is, the excess demand data indicating the daily excess demand for the stock brand. The data indicating the value is stored for a predetermined period, for example, one year. Here, the buyer-initiated transaction or the seller-initiated transaction is determined based on which of the buyer and the seller led the sale transaction. That is, in the stock market, investors are waiting at a price range that they want to sell or purchase, respectively, and these waiting investors are waiting for another investor to buy (sell). It is in a passive state. In this state, for example, when an active buyer appears and purchases at the desired selling price of the waiting seller, the buyer leads the transaction. In addition, when an active seller appears and sells at the desired purchase price of the waiting buyer, the seller leads the transaction. Therefore, first, based on the tick-by-tick stock price included in the transaction data stored in the transaction data storage unit 8, the contents of the completed transaction and the investors who are waiting in each price range are shown. By matching the data, all of the daily transactions are classified as buyer-initiated or seller-initiated transactions. Then, the excess demand in any one day is calculated by subtracting the total value of the volume of the seller-led transaction from the total volume of the volume of the buyer-led transaction. The excess demand has a value of “+” when the buyer is dominant, and a value of “−” when the seller is dominant. The data indicating excess demand is stored in the excess demand data storage unit 10 every day in correspondence with the stock brand to be traded.

  FIG. 2 is a diagram schematically showing sellers and buyers waiting in a desired price range for brands traded at 100 yen in the trading market. As shown in FIG. 2, with respect to stocks traded at 100 yen, the seller wants to sell at a slightly higher price, and is waiting in each price range of 101 yen to 105 yen. On the other hand, the buyer is waiting between 99 yen and 95 yen in order to purchase at a low price. In such a state, the most concessional price among sellers and buyers, that is, as shown in FIG. 2, the best price is 101 yen for the seller and 99 yen for the buyer. Here, as shown in FIG. 2, another investor who can purchase at 101 yen despite the buyer's best price (99%) is the price range of 101 yen. When a sales transaction is concluded by meeting a seller who is waiting at, the transaction becomes a buyer-initiated transaction. In addition, another investor who can sell for 99 yen despite meeting the seller's best price (the best price) is 101 yen meets a buyer who is waiting at a price range of 99 yen. When a deal is closed, the deal becomes a seller-initiated deal. Therefore, by comparing the data indicating the sellers and buyers waiting in each price range with the transaction contents included in the transaction data, the transaction of the day can be made into a seller-initiated transaction and a buyer-initiated transaction. The excess demand per day can be calculated. In addition, the block which shows the seller and the buyer who are waiting in each price range shown in FIG. 2 shows the number of stocks, and shows that, for example, 1,000 stocks are waiting by one block. Therefore, 5,000 shares are waiting for 101 yen and 4,000 shares are waiting for 102 yen.

  Further, the price change rate data storage unit 12 stores price change rate data, that is, data indicating the price change rate of the stock price for one day for a predetermined period, for example, for one year. Here, when the stock price has an upward trend, the price change rate is calculated with a value of “+”, and when the stock price has a downward trend, the price change rate is calculated with a value of “−”. The price change rate data is stored in the price change rate data storage unit 12 every day corresponding to the stock brand.

  Further, the data storage unit 14 stores data indicating an estimated value of market impact, that is, a price given to the market by performing a sales transaction when a predetermined number of shares are sold or purchased for an arbitrary stock issue. Data indicating an estimated value of how much the change rate of the data will be stored.

  Next, processing in the market impact estimation system according to the embodiment of the present invention will be described with reference to the flowchart of FIG. In addition, the process demonstrated below is performed in the data control part 4 of a market impact estimation system.

  First, based on the excess demand data stored in the excess demand data storage unit 10, a cumulative value of excess demand for a given stock brand in a predetermined period is calculated (step S10). That is, first, based on the transaction data stored in the transaction data storage unit 8, every day within a predetermined period, all transactions on that day are classified into buyer-initiated transactions and seller-initiated transactions. Next, the excess demand is calculated by subtracting the volume of the seller-initiated transaction from the volume of the buyer-initiated transaction, and the excess daily demand is stored in the excess demand data storage unit 10 in correspondence with the date. Remember. Then, the accumulated excess demand for 20 days is calculated by accumulating the excess demand for each day stored in the excess demand data storage unit 10 for a predetermined period, for example, for 20 days.

  The accumulated value of excess demand for 20 days indicates how much the balance between the seller and the buyer in 20 days is deviated, that is, whether the seller is dominant or the buyer is dominant. Note that the calculated accumulated value of excess demand for 20 days is stored in the data storage unit 14 in correspondence with the period for each stock brand, that is, the date indicating which period is the accumulated value of excess demand. .

  Next, the price change rate within a predetermined period is calculated based on the price change rate data stored in the price change rate data storage unit 12 for the stock brand for which the cumulative value of excess demand has been calculated (step S11). That is, based on the transaction data stored in the transaction data storage unit 8, the daily price change rate within a predetermined period is calculated, and the calculated price change rate is associated with the date, and the price change rate data storage unit 12 to store. Based on the daily price change rate data stored in the price change rate data storage unit 12, the price change rate in the same predetermined period as the cumulative value of excess demand in step S10 is calculated. To do. For example, based on the daily price change rate from the beginning to the end of the 20 days, the price change rate for 20 days indicating how the stock price changed in 20 days is calculated. The calculated price change rate for 20 days is stored in the data storage unit 14 in association with the accumulated value of excess demand for 20 days calculated in step S10.

  Next, sample data is created by combining the accumulated value of excess demand in a predetermined period and the price change rate in the same predetermined period (step S12). That is, sample data is created in which the accumulated value of excess demand and the price change rate are set as one set.

  Next, it is determined whether or not a predetermined number of sets of sample data have been created (step S13). If a predetermined number of sets of sample data has not been created, a predetermined number of sets of sample data has been created. Until it is done, the calculation of the accumulated value of excess demand in a predetermined period (step S10), the calculation of the price change rate of the stock price in the same predetermined period (step S11), and the creation of sample data (step S12) are repeated. For example, sample data of cumulative value of excess demand and price change rate for 20 days from June 27 to May 30 (20 business days excluding weekends and holidays and stock days when stock exchanges are closed), 5 For example, 10 sets of sample data of different predetermined periods based on transaction data for the past one year, such as sample data of the accumulated value of excess demand and price change rate for 20 days from March 29 to April 30 Step S10 to Step S12 are repeated until is calculated. In addition, a peculiar day means the day when it is known experimentally that a specific transaction will be performed very actively based on the past transaction history etc.

When a predetermined number of sample data has been created (step S13), parameters used for estimating market impact are calculated based on the created sets of sample data (step S14). That is, let Δv _{i be} the cumulative value of excess demand for 20 days, Δp _i be the rate of price change during the same period,
Δp _i = λΔv _i + μ + ε _i (Equation 1)
Define Then, by applying a plurality of sets of sample data created in step S12 to this (Equation 1), and performing regression analysis using the accumulated excess demand value Δv _i and the price change rate Δp _i as variables, parameters λ and Calculate the value of μ. Note that ε _i in (Equation 1) is a term representing an error. Further, the calculated values of the parameters λ and μ are stored in the data storage unit 14 in association with the stock brand. The number of sample data to be created can be set in advance by inputting an arbitrary value via the keyboard 16, for example.

  FIG. 4 is a diagram for explaining a process for calculating a parameter. As shown in FIG. 4, the value of the sample data created in step S12 is plotted on the coordinates with the accumulated value Δv of excess demand in a predetermined period on the horizontal axis and the price change rate Δp in the predetermined period on the vertical axis. Then, based on the values of the sampled data plotted, as shown in FIG. 4, a graph in which the parameter λ is inclined and the parameter μ is an intercept is shown. The calculated parameters λ and μ are calculated for each stock brand to be traded and stored in the data storage unit 14 in correspondence with the stock brand.

  Next, the average spread rate is calculated based on the transaction data stored in the transaction data storage unit 8 (step S15). That is, first, the spread rate is calculated based on the tick-by-tick stock price included in the transaction data stored in the transaction data storage unit 8. The spread rate is, for example, the distance between the buyer of 99 yen that is the best bid concessioned in terms of price and the seller of 101 yen that is the best bid shown in FIG. Defined.

Spread ratio = (best bid price-best bid price) / (best bid price + best bid price) (Formula 2)
In this case, since the best bid price is 99 yen and the best bid price is 101 yen, the spread rate is “1%” from (Formula 2). Here, since the tick-by-tick stock price is constantly changing, the spread rate is also constantly changing. Therefore, it is determined based on the transaction data stored in the transaction data storage unit 8 how long the spread rate was, and the average spread rate is calculated while weighting the period (time). . Since the average value of the spread rate is performed independently of the calculation of the parameter λ, it is calculated based on transaction data for an arbitrary period, for example, 10 days. That is, for each day of 10 days, the average value of the spread rate for the day is calculated while weighting with time, and the average value of the spread rate for each day for 10 days is calculated as the average value of the spread rate for 10 days (average (Spread rate). Further, the calculated average spread rate is stored in the data storage unit 14 in association with the stock brand.

Next, an estimated value of market impact is calculated based on the parameters and average spread rate stored in the data storage unit 14 (step S16). That is, for any stock issue, define (equation 3) using the stock issue parameters and average spread rate,
(Output) = λ (Input) + average spread ratio (Formula 3)
Based on this (Formula 3), an estimated value of market impact is calculated. Here, “(Input)” is the number of shares scheduled to be bought and sold, and the value input via the keyboard 16 is used. Further, “(Output)” is an estimated value of the rate of change of the price that occurs when the number of stocks input via the keyboard 16 is traded, that is, the market impact. Therefore, the estimated value of the market impact when the planned transaction is performed is calculated by applying the parameters of the stock issues scheduled for the sales transaction, the average spread rate, and the number of shares scheduled for sale to (Formula 3). be able to. The estimated market impact value is stored in the data storage unit 14 in association with the stock brand.

  Here, it is known from experience that it is always affected by the spread rate regardless of what transactions are made, so the average spread rate is added on the right side of (Formula 3). . Note that the parameter μ corresponding to the intercept of the graph shown in FIG. 4 calculated in step S14 is generally not a significant value, and therefore only λ is used as a parameter without considering the parameter μ (Equation 3). Based on the above, an estimated value of market impact is calculated.

  Next, the calculated estimated value of the market impact is displayed on the display unit 18 (step S17). For example, the estimated value of the market impact stored in the data storage unit 14 is displayed on the display unit 18 constituted by a liquid crystal display.

  According to the market impact estimation system according to the embodiment of the present invention, since the parameter used when estimating the market impact is calculated using the cumulative value of excess demand and the price change rate in a predetermined period, It is possible to estimate the exact market impact that reflects the actual situation. In other words, excess demand does not occur instantaneously, but occurs while buying and selling are intermingled within a certain period of time. By creating and calculating the parameters, it is possible to estimate the market impact based on the transaction mode over time. Therefore, it is possible to accurately estimate the market impact that occurs when a planned transaction is performed assuming a mode close to an actual transaction mode.

  In addition, according to the market impact estimation system according to the embodiment of the present invention, since the parameters used when estimating the market impact are calculated using the accumulated value of excess demand over a predetermined period, a large number of shares The market impact when trading can be accurately estimated. That is, since a plurality of accumulated values of excess demand for a plurality of days are used, it is possible to estimate the market impact using a quantity several times the daily volume. Accordingly, when the number of stocks that are actually traded is large, it is possible to accurately estimate the market impact caused by conducting the trade.

  Further, according to the market impact estimation system according to the embodiment of the present invention, since the market impact is estimated in consideration of the average spread rate, the accurate market impact is estimated appropriately reflecting the price fluctuation of the stock price. A value can be calculated.

  In the market impact estimation system according to the above-described embodiment, the predetermined period for calculating the accumulated value of excess demand is 20 days. However, the accumulated value with 20 days or less or 20 days or more as the predetermined period. May be calculated. That is, the cumulative value may be calculated by setting an arbitrary number of days such as 3 days, 5 days, 7 days, 10 days, 12 days, 15 days, or 25 days as a predetermined period.

  In the market impact estimation system according to the above-described embodiment, 10 sets of sample data are created, but the number of sample data created may be 10 sets or more or 10 sets or less. For example, the number of sample data to be created may be adjusted based on the amount of transaction data stored in the transaction data storage unit.

  In addition, in the market impact estimation system according to the above-described embodiment, a value obtained by subtracting the volume of the seller-led transaction from the volume of the buyer-led transaction is defined as excess demand. By dividing this by an appropriate value, You may use the normalized value as an excess demand. In other words, the parameter used to estimate the market impact may be calculated using the value obtained by subtracting the seller-led volume from the buyer-led volume divided by the number of listed shares or the total volume in the market as excess demand. Good.

  Further, in the market impact estimation system according to the above-described embodiment, the period for calculating the average spread rate is 10 days, but the average spread rate for the number of days of 10 days or less or the number of days of 10 days or more is calculated. May be. That is, you may make it calculate the average spread rate of arbitrary periods, such as 5 days, 15 days, and 20 days.

  Further, in the market impact estimation system according to the above-described embodiment, the number of shares scheduled to be traded is input via the keyboard, but the number of shares scheduled to be traded may be input via the communication control unit. Good. That is, for example, when a user terminal possessed by a trader or the like and a market impact estimation system are connected via a network, the number of shares scheduled to be traded input by the trader via the user terminal is obtained via the communication control unit. You may make it receive. In this case, the estimated market impact value may be transmitted to the user terminal via the communication control unit, and the estimated market impact value may be displayed on the user terminal.

It is a block block diagram of the market impact estimation system which concerns on embodiment of this invention. It is the figure which showed typically the seller and buyer who wait in each price range which concern on embodiment of this invention. It is a flowchart for demonstrating the process by the market impact estimation system which concerns on embodiment of this invention. It is a figure for demonstrating the process which calculates the parameter which concerns on embodiment of this invention.

Explanation of symbols

  2 ... Market impact estimation system, 4 ... Data control unit, 6 ... Communication control unit, 8 ... Transaction data storage unit, 10 ... Excess demand data storage unit, 12 ... Price change rate data storage unit, 14 ... Data storage unit, 16 ... Keyboard, 18 ... Display section.

Claims (4)

Transaction data storage means for storing transaction data for each brand of securities,
Based on the transaction data stored in the transaction data storage means, excess demand accumulated value calculating means for calculating a cumulative value of excess demand within a predetermined period of a predetermined brand;
Price change rate calculating means for calculating a price change rate within the predetermined period of the predetermined brand based on the transaction data stored in the transaction data storage means;
Sample data creating means for creating a plurality of sets of sample data in which a cumulative value of excess demand within the predetermined period and a price change rate within the predetermined period are set as one set;
Based on a plurality of sets of sample data created by the sample data creation means, parameter calculation means for calculating parameters used when estimating market impact;
Based on the transaction data stored in the transaction data storage means, an average spread ratio calculation means for calculating an average spread ratio of the predetermined brand;
Market impact estimated value calculating means for calculating an estimated value of market impact based on the parameter calculated by the parameter calculating means and the average spread ratio calculated by the average spread ratio calculating means. Impact estimation system. The excess demand accumulated value calculating means is:
2. The market impact estimation system according to claim 1, wherein a cumulative value of excess demand in the predetermined period is calculated by accumulating excess demand values for each day within the predetermined period. The parameter calculation means includes
3. The market impact estimation system according to claim 1, wherein the parameter is calculated by performing regression analysis based on a plurality of sets of sample data created by the sample data creation means. The average spread rate calculating means is:
4. The average spread ratio is calculated by weighting each of the plurality of spread ratios corresponding to a period in which each of the spread ratios was indicated. 5. Market impact estimation system.

Images