CN103236013A - Stock market data analysis method based on key stock set identification - Google Patents

Abstract

The invention discloses a stock market data analysis method based on key stock set identification. The method comprises the following steps that (1) data is collected, the association relationship among the stock trading volume is calculated, and a stock association network is built; (2) in the stock association network, a key stock set is identified by a search algorithm in an iteration mode; and (3) the market trend expectation is calculated by using the trading volume as the weight according to the key stock price trend. The method has the advantages that the association relationship among the stock trading volume is fully excavated, stocks which are in an active state and have an influence effect can be accurately judged according to the trading condition of the stock market, and the accuracy of the stock market trend prediction is improved. The calculation is simple, timeliness, flexibility and expansibility are realized, the historical data processing requirement can be regulated, and the method is suitable for conditions with great stock market data quantity and frequent stock trading changes.

Description

A kind of stock market deep bid data analysing method based on crucial stock collection identification

Technical field

The present invention relates to a kind of Forecasting Methodology, particularly in the stock market, accurately predict the method for views on broad market movements; This method is identified current active state and the influential crucial stock of being in based on the incidence relation between the stock volume quantity, finishes the views on broad market movements prediction according to this.

Background technology

The views on broad market movements of stock market is reflected as the variation of stock price index usually.For calculating stock price index, current method is to choose representational one group of stock, be weighted on average or simple arithmetic mean according to the transaction value to these stocks such as trading volume or share price total value, on the basis of selected historical price baseline, obtain with respect to the percentage of baseline total value by calculating current market capitalization.Common stock price index comprises Dow Jones index, Standard ﹠ Poor's Index, Hang Seng Index, Shanghai Stock Exchange's 180 indexes, Shanghai and Shenzhen 300 indexes etc.

Investment in stocks need be made reasonable estimation to direction and the possibility of stock future trend.A kind of feasible strategy is to do things by following nature, and namely buys in the rise stage of deep bid, and the stage that is about to drop at deep bid is sold, and this strategy can significantly improve the chance of success of investment in stocks.In addition, along with the introducing of stock price index futures in recent years, views on broad market movements prediction accuracy and quantification are required obviously to improve.Existing deep bid data analysing method mainly relies on stock price index, by analyzing historical data and the trend curve of stock price index, estimates the stock price index variation in future.Means commonly used comprise for analyzing stock exponential wave moving GARCH and OHLC technology, be used for excavating neural network, Markov chain and the support vector machine technology of stock index change rule, and based on the fractal theory of stock index trend curve style characteristic, various land and negative line forecasting techniques etc.

Predict that from stock price index itself there is certain risk in views on broad market movements merely.What the calculating of at first various stock index was adopted all is standard stocks, can not adjust according to the current stateful transaction in stock market, does not therefore have ageing; Adopt standard stocks to calculate stock index, often deep bid is gone up or downward tendency slow in reacting, obviously change appears in stock price index by the time, the fact, miss best opportunity of investment in stocks.In addition, the views on broad market movements changeable situation of stock market, be subjected to multiple factor affecting, along with the symmetric raising of both parties' information, stock price index itself trends towards a fair game (being defined as halter strap), it is characterized in that stock index equals the last observed reading in the value expectation of next period.This point is similar to coin and throws, and all faces up even throw 99 times in history, and heads expected probability still equals the probability of the 99th throwing during the 100th throwing.If there is fair game, then only according to historical data and the trend curve of stock price index, can not estimate accurately that change the future of stock index.

Summary of the invention

Technical matters to be solved by this invention provides a kind of stock market deep bid data analysing method based on crucial stock collection identification, this method is fully excavated the incidence relation between the stock volume quantity, can identify efficiently and in real time to be in active state and influential stock; Calculate simply simultaneously, have ageing and extendability, can improve stock market views on broad market movements prediction accuracy, and adaptation stock market data amount is big and stock exchange changes characteristics frequently.

For achieving the above object, the present invention adopts following step:

1) collecting the stock exchange data, calculate the incidence relation between the stock volume quantity, is that node, incidence relation are that the limit makes up the stock related network with stock;

2) in the stock related network, with the crucial stock set of the mode application searches algorithm identified of iteration;

3) price trend recent according to crucial stock is weight with the trading volume, calculates the views on broad market movements expectation, estimates that deep bid goes up or downward tendency.

Above-mentioned steps 1) detailed process is in: at first obtain the stock market transaction data of all stocks at no distant date; To each stock, be unit with the sky then, record the trading volume of each day of trade in chronological order, form the transaction vector; Next setting-up time is poor, and the transaction vector according to last period of stock and one period of back calculates the incidence relation weight between any two stock volume quantities; Be incidence relation weight setting threshold value, keep the incidence relation that weight is not less than threshold value; Be that node, incidence relation are that directed edge makes up the stock related network at last with stock, form a digraph.

Wherein, the method for calculating the incidence relation weight of two stocks is: be unit with the sky, obtain the trading volume of every day of a certain stock si in the continuous l day of trade, form the vectorial ξ of transaction of si _i=＜x _{I, 1}, x _{I, 2}..., x _{I, l}; Element in the vector is the sampling stock volume quantity of day, and l is vector length; Setting-up time difference t is unit equally with the sky then, and parameter l and t adjust according to the deep bid fluctuation situation of stock market, when the fluctuation of stock market deep bid violent, the investment in stocks time cycle in short-term, the corresponding minimizing with the t value of l, otherwise then corresponding increase; Next to any two stock si and sj, according to the transaction vector of both different periods, calculate the incidence relation weight w from si to sj _{＜i, j 〉}, formula is as follows:

$w_{<i,j>}=\frac{{\mathrm{\&Sigma;}}_{k=1}^l(x_{i,k}-{\stackrel{\&OverBar;}{x}}_i)(x_{j,k}-{\stackrel{\&OverBar;}{x}}_j)}{(l-1)\&CenterDot;{\mathrm{\&sigma;}}_i\&CenterDot;{\mathrm{\&sigma;}}_j};$

The transaction of si vector ξ wherein _i=＜x _{I, 1}, x _{I, 2}..., x _{I, l}And the transaction of sj vector ξ _j=＜x _{J, 1}, x _{J, 2}..., x _{J, l}Take from the different periods, first vector element xj of sj, 1 date collected are than first vector element xi of si, and 1 a late t day of trade, the trading volume of reflection stock si changes the degree that stock sj is exerted an influence; With Represent the vectorial ξ of transaction respectively _iAnd ξ _jAverage, σ i and σ j be the standard deviation of expression vector respectively then; Transaction vector with stock si is example,

As follows with the computing formula of σ i:

${\stackrel{\&OverBar;}{x}}_i=\frac{1}{l}{\mathrm{\&Sigma;}}_{k=1}^l{x}_{i,k}$

${\mathrm{\&sigma;}}_i=\sqrt{\frac{1}{l-1}{\mathrm{\&Sigma;}}_{k=1}^l{(x_{i,k}-{\stackrel{\&OverBar;}{x}}_i)}^2};$

According to computing formula, the span of incidence relation weight is-1≤w _{＜i, j 〉}≤ 1; Set the threshold value w of incidence relation weight accordingly, keep the incidence relation that weight is not less than threshold value w.

Step 2) detailed process is in: at first determine the scale n of crucial stock set, the n value can be along with the current transaction size adjustment in stock market, and transaction size is more big, the corresponding increase of n, on the contrary then reduce; In the mode of iteration, application searches algorithm in the stock related network selects m to come foremost stock at every turn then; Add selected stock to crucial stock set, the selected stock of deletion and incidence edge from the stock related network enter the next round iteration again; Up to the crucial stock of selecting sufficient amount.

Wherein, the application searches algorithm is searched the method for crucial stock and is: adopt the PageRank algorithm to come searched key stock; Given current stock related network, namely digraph G=(S R), sets up the adjacency matrix A of G, and matrix A is one | S| * | the 0-1 matrix of S|, wherein | S| represents the scale of stock node set S, and R is directed edge set; Element aij is defined as follows in the matrix A:

$\mathrm{aij}=\left\{\begin{array}{cc}1& <s_i,s_j>\&Element;R\\ 0& <s_i,s_j>\&NotElement;R\end{array}\right.;$

With the input as the PageRank algorithm of the transposed matrix of matrix A; In addition, the ratio of damping that the PageRank algorithm is set is α=0.3.

The process of calculating the views on broad market movements expectation according to crucial stock set in the step 3) is: divide two steps; The first step is calculated its price trend with regard to each crucial stock; Given crucial stock si at first obtains the si closing price of nearest 6 day of trade, accounting price tendency vector △ _i=＜δ _{I, 1}, δ _{I, 2}..., δ _{I, 5}, element δ i wherein, k(1≤k≤5) computing formula as follows:

Calculate the vectorial average of Δ i then Represent the price trend of stock si, formula is as follows:

${\stackrel{\&OverBar;}{\mathrm{\&delta;}}}_i=\frac{1}{5}\underset{k=1}{\overset{5}{\mathrm{\&Sigma;}}}{\mathrm{\&delta;}}_{i,k};$

Second step is according to the price trend of every crucial stock, is weight with the total volume of nearest 6 day of trade, adopts the tendency expectation of calculated with weighted average method stock market deep bid; The total volume that makes stock si is Qi, and the quantity of crucial stock is m, and formula is as follows:

The inventive method is used the incidence relation between the stock volume quantity, makes up the stock related network; Use ripe searching algorithm in the mode of iteration, identify efficiently and in real time and be in active state and influential crucial stock set in the stock market; By calculating the price trend of crucial stock, the views on broad market movements of quantitative estimation and prediction stock market.Theoretical analysis and real data check show, use the inventive method and can effectively avoid adopting fixing standard stocks when calculating stock price index, can not in time reflect the problem of the current stateful transaction in stock market; Avoid simultaneously in the prediction of stock indices may problem, the problem includes: the fair game problem; Effectively improve rationality and the accuracy of stock market views on broad market movements prediction.The inventive method is calculated simple, have ageing, dirigibility and extendability, can adjust and control the processing demands to the stock market historical data, be applicable to that the big and stock exchange of stock market data amount changes situation frequently, can finish the prediction of stock market views on broad market movements quantitatively, accurately and in time.

Description of drawings

Fig. 1 is based on the overall framework of the stock market deep bid data analysing method of crucial stock collection identification;

Fig. 2 analyzes association between the stock volume quantity to make up the treatment scheme of stock related network;

Fig. 3 is an example that makes up the stock related network according to 8 stock certificate datas;

Fig. 4 is the treatment scheme of the crucial stock set of identification in the stock related network;

Fig. 5 is the treatment scheme of calculating the views on broad market movements expectation according to the price trend of crucial stock.

Embodiment

Figure 1 shows that the general technical framework based on the stock market deep bid data analysing method of crucial stock collection identification.The input of method is the stock market transaction data of all stocks at no distant date, and the output of method is the quantitative forecast of stock market deep bid future trend.The inventive method comprises three processing modules: at first according to the stock market transaction data, calculating the incidence relation between the stock volume quantity, is that node, incidence relation are that the limit makes up the stock related network with stock; In the stock related network, use ripe searching algorithm in the mode of iteration then, identify crucial stock set; The last price trend recent according to crucial stock is weight with the trading volume, calculates the views on broad market movements expectation, estimates that deep bid goes up or downward tendency.

First module of the inventive method is to make up the stock related network, and implementation as shown in Figure 2.At first obtain the stock market transaction data of all stocks at no distant date; To each stock, be unit with the sky then, record the trading volume of each day of trade in chronological order, form the transaction vector; Next setting-up time is poor, and the transaction vector according to last period of stock and one period of back calculates the incidence relation weight between two stock volume quantities; Be incidence relation weight setting threshold value, keep the incidence relation that weight is not less than threshold value; Being that node, incidence relation are that directed edge makes up the stock related network at last with stock, is a digraph.

For calculating the incidence relation weight of two stocks, be unit with the sky at first, obtain the trading volume of every day of a certain stock si in a continuous l day of trade, form the transaction vector ξ of si _i=＜x _{I, 1}, x _{I, 2}..., x _{I, l}; Element in the vector is the sampling stock volume quantity of day, and l is vector length, can be set at l=15.Setting-up time difference t is unit equally with the sky then, can be set at t=3.Parameter l and t can adjust according to the deep bid fluctuation situation of stock market, and what the inventive method provided is reference value; General when the fluctuation of stock market deep bid violent, the investment in stocks time cycle in short-term, the corresponding minimizing with the t value of l, otherwise then corresponding increase.Next to any two stock si and sj, according to the transaction vector of both different periods, adopt the incidence relation weight w of crossing dependency calculating from si to sj _{＜i, j 〉}, as shown in Equation (1):

$w_{<i,j>}=\frac{{\mathrm{\&Sigma;}}_{k=1}^l(x_{i,k}-{\stackrel{\&OverBar;}{x}}_i)(x_{j,k}-{\stackrel{\&OverBar;}{x}}_j)}{(l-1)\&CenterDot;{\mathrm{\&sigma;}}_i\&CenterDot;{\mathrm{\&sigma;}}_j}---\left(1\right)$

The transaction of si vector ξ wherein _i=＜x _{I, 1}, x _{I, 2}..., x _{I, l}And the transaction vector of sj

ξ _j=＜x _{J, 1}, x _{J, 2}..., x _{J, l}Take from the different periods, first vector element xj of sj, 1 date collected are than first vector element xi of si, and 1 a late t day of trade, the trading volume of reflection stock si changes the degree that stock sj is exerted an influence;

With

Represent the vectorial ξ of transaction respectively _iAnd ξ _jAverage, σ i and σ j be the standard deviation of expression vector respectively then; Transaction vector with stock si is example,

With the calculating of σ i as shown in Equation (2):

${\stackrel{\&OverBar;}{x}}_i=\frac{1}{l}{\mathrm{\&Sigma;}}_{k=1}^l{x}_{i,k}$

${\mathrm{\&sigma;}}_i=\sqrt{\frac{1}{l-1}{\mathrm{\&Sigma;}}_{k=1}^l{(x_{i,k}-{\stackrel{\&OverBar;}{x}}_i)}^2}---\left(2\right)$

Incidence relation between the stock volume quantity is oriented, the incidence relation＜si from stock si to stock sj, sj〉be different from the incidence relation＜sj from stock sj to stock si, si 〉; What both adopted when calculating weight is different transaction vectors.

According to computing formula, the span of incidence relation weight is-1≤w _{＜i, j 〉}≤ 1; Set the threshold value w of incidence relation weight at last, can be set at w=0.6, keep the incidence relation that weight is not less than threshold value w, be used for making up the stock related network.

The transaction data that Fig. 3 samples in 18 day of trade with 8 stocks is example, explains the building process of stock related network.Stock exchange shown in the table 1 reflects the trading volume of each stock in continuous 18 day of trade data from Yahoo's finance and economics.For making up the stock related network, each parameter is set at: transaction vector length l=15, mistiming t=3, weight threshold w=0.6; Table 2 is depicted as the calculated value of incidence relation weight between stock in twos, and wherein row represents the initial stock of incidence relation, and row represent the target stock of incidence relation.Be example with stock s1 to the incidence relation of stock s2, the transaction vector of s1 is made up of 15 trading volume data of 1～15 day of trade sampling, and 15 trading volume data that the transaction vector of s2 was sampled by 4～18 days of trade are formed; Calculate weight w＜1,2 according to formula (1) and (2) 〉, the result is 0.763; In like manner calculate stock s2 to incidence relation weight w＜2,1 of stock s1 〉, the result is-0.177, obviously both are also inequality.Based on weight threshold, remove the incidence relation of weight deficiency, constructed stock related network is as shown in Figure 3.

Table 1

Unit: ten thousand strands

Table 2

Second module of the inventive method is the crucial stock set of identification in the stock related network, and implementation as shown in Figure 4.At first determine the scale n of crucial stock set, be set at n=60, the n value can be along with the current transaction size adjustment in stock market, and transaction size is more big, the corresponding increase of n.In the mode of iteration, use ripe searching algorithm in the stock related network then, select m to come foremost stock at every turn, parameter m can be set at m=3; Add selected stock to crucial stock set, the selected stock of deletion and incidence edge from the stock related network enter the next round iteration again; So processing can reduce interactive degree between the crucial stock, increases independence and the representativeness of selected crucial stock; After selecting the crucial stock of sufficient amount, export crucial stock set at last.

According to Fig. 4, in iteration each time, can adopt the PageRank algorithm to come searched key stock.Given current stock related network (digraph) G=(S R), at first sets up the adjacency matrix A of G, is one | S| * | the 0-1 matrix of S|, wherein | S| represents the scale of stock node set S, and R is the directed edge set; The definition of element aij is as shown in Equation (3) in the matrix A:

$\mathrm{aij}=\left\{\begin{array}{cc}1& <s_i,s_j>\&Element;R\\ 0& <s_i,s_j>\&NotElement;R\end{array}\right.---\left(3\right)$

Because the PageRank algorithm is gone into the criticality that limit quantity (being in-degree) determines node according to chain, and for the stock related network, therefore the criticality of stock node is decided by that the chain of node goes out limit quantity (being out-degree), with the input as the PageRank algorithm of the transposed matrix of matrix A; In addition, the PageRank algorithm needs a ratio of damping α, original meaning represents the probability that page viewers browse any webpage at random, be applied to the stock related network, represent External Stochastic Factor to the influence degree of stock market, consider that the stock market is subjected to the influence of all kinds of external factor bigger, can be set to α=0.3.

The 3rd module of the inventive method is the price trend according to crucial stock, calculates the views on broad market movements expectation, and implementation as shown in Figure 5.Divide two steps, the first step is calculated its price trend with regard to each crucial stock.Given crucial stock si at first obtains the si closing price of nearest 6 day of trade, accounting price tendency vector △ _i=＜δ _{I, 1}, δ _{I, 2}..., δ _{I, 5}, element δ i wherein, k(1≤k≤5) calculating as shown in Equation (4):

Calculate the vectorial average of Δ i then

Represent the price trend of stock si, as shown in Equation (5):

${\stackrel{\&OverBar;}{\mathrm{\&delta;}}}_i=\frac{1}{5}\underset{k=1}{\overset{5}{\mathrm{\&Sigma;}}}{\mathrm{\&delta;}}_{i,k}---\left(5\right)$

Second step is according to the price trend of every crucial stock, is weight with the total volume of nearest 6 day of trade, adopts the tendency expectation of calculated with weighted average method stock market deep bid; The total volume that makes stock si is Qi, and the quantity of crucial stock is m, and formula is as follows:

Table 3 is depicted as an example calculating the views on broad market movements expectation:

Table 3:

As shown in table 3, to simplify the process, suppose to have selected stock s1～s6 as crucial stock.Be example with stock s1, the closing price of its 6 day of trade as shown in Table, by formula (4) calculate the price trend vector Δ 1 of s1, can get: Δ 1=＜-0.110 ,-1.211 ,-2.371,0.424 ,-2.889 〉; (5) compute vector average more by formula

Can get

Other stocks are with identical method compute vector average.At last based on the total volume of each 6 day of trade of stock vectorial average is got weighted mean, by formula calculate (6), can get the tendency expectation of deep bid for-0.299.From the deviation situation of each stock vector average, tend towards stability substantially in current stock market, shows downward tendency slightly.

If the stock market is considered as a power system, then adopt the inventive method, the crucial stock of identifying is equivalent to drive source, can drive trading volume and the tendency of other stocks; The price trend vector average of calculating is equivalent to the average velocity of every crucial balloon or drop, and the total amount of transactions of stock is equivalent to quality and the power of every stock; By the views on broad market movements expectation of this calculating, be equivalent to give a momentum of stock market, can quantize to infer the tendency of stock market deep bid in future according to this.

The inventive method is used the incidence relation between the stock volume quantity, makes up the stock related network; Use ripe searching algorithm in the mode of iteration, identify efficiently and in real time and be in active state and influential crucial stock set in the stock market; By calculating the price trend of crucial stock, the views on broad market movements of quantitative estimation and prediction stock market.Theoretical analysis and real data check show, use the inventive method and can effectively avoid adopting fixing standard stocks when calculating stock price index, can not in time reflect the problem of the current stateful transaction in stock market; Avoid simultaneously in the prediction of stock indices may problem, the problem includes: the fair game problem; Effectively improve rationality and the accuracy of stock market views on broad market movements prediction.The inventive method is calculated simple, have ageing, dirigibility and extendability, can adjust and control the processing demands to the stock market historical data, be applicable to that the big and stock exchange of stock market data amount changes situation frequently, can finish the prediction of stock market views on broad market movements quantitatively, accurately and in time.

Claims (6)

1. stock market deep bid data analysing method based on the identification of crucial stock collection is characterized in that comprising following steps:

1) collecting the stock exchange data, calculate the incidence relation between the stock volume quantity, is that node, incidence relation are that the limit makes up the stock related network with stock;

2) in the stock related network, with the crucial stock set of the mode application searches algorithm identified of iteration;

3) price trend recent according to crucial stock is weight with the trading volume, calculates the views on broad market movements expectation, estimates that deep bid goes up or downward tendency.

2. the stock market deep bid data analysing method based on crucial stock collection identification according to claim 1 is characterized in that detailed process is in the step 1): at first obtain the stock market transaction data of all stocks at no distant date; To each stock, be unit with the sky then, record the trading volume of each day of trade in chronological order, form the transaction vector; Next setting-up time is poor, and the transaction vector according to last period of stock and one period of back calculates the incidence relation weight between any two stock volume quantities; Be incidence relation weight setting threshold value, keep the incidence relation that weight is not less than threshold value; Be that node, incidence relation are that directed edge makes up the stock related network at last with stock, form a digraph.

3. the stock market deep bid data analysing method based on the identification of crucial stock collection according to claim 2, it is characterized in that, the method of calculating the incidence relation weight of two stocks is: be unit with the sky, obtain the trading volume of every day of a certain stock si in a continuous l day of trade, form the transaction vector ξ of si _i=＜x _{I, 1}, x _{I, 2}..., x _{I, l}; Element in the vector is the sampling stock volume quantity of day, and l is vector length; Setting-up time difference t is unit equally with the sky then, and parameter l and t adjust according to the deep bid fluctuation situation of stock market, when the fluctuation of stock market deep bid violent, the investment in stocks time cycle in short-term, the corresponding minimizing with the t value of l, otherwise then corresponding increase; Next to any two stock si and sj, according to the transaction vector of both different periods, calculate the incidence relation weight w from si to sj _{＜i, j 〉}, formula is as follows:

$w_{<i,j>}=\frac{{\mathrm{\&Sigma;}}_{k=1}^l(x_{i,k}-{\stackrel{\&OverBar;}{x}}_i)(x_{j,k}-{\stackrel{\&OverBar;}{x}}_j)}{(l-1)\&CenterDot;{\mathrm{\&sigma;}}_i\&CenterDot;{\mathrm{\&sigma;}}_j};$

The transaction of si vector ξ wherein _i=＜x _{I, 1}, x _{I, 2}..., x _{I, l}And the transaction of sj vector ξ _j=＜x _{J, 1}, x _{J, 2}..., x _{J, l}Take from the different periods, first vector element xj of sj, 1 date collected are than first vector element xi of si, and 1 a late t day of trade, the trading volume of reflection stock si changes the degree that stock sj is exerted an influence;

With

Represent the vectorial ξ of transaction respectively _iAnd ξ _jAverage, σ i and σ j be the standard deviation of expression vector respectively then; Transaction vector with stock si is example,

As follows with the computing formula of σ i:

${\stackrel{\&OverBar;}{x}}_i=\frac{1}{l}{\mathrm{\&Sigma;}}_{k=1}^l{x}_{i,k}$

${\mathrm{\&sigma;}}_i=\sqrt{\frac{1}{l-1}{\mathrm{\&Sigma;}}_{k=1}^l{(x_{i,k}-{\stackrel{\&OverBar;}{x}}_i)}^2};$

According to computing formula, the span of incidence relation weight is-1≤w _{＜i, j 〉}≤ 1; Set the threshold value w of incidence relation weight accordingly, keep the incidence relation that weight is not less than threshold value w.

4. according to claim 1,2 or 3 described stock market deep bid data analysing methods based on crucial stock collection identification, it is characterized in that, step 2) detailed process is in: the scale n that at first determines crucial stock set, the n value can be along with the current transaction size adjustment in stock market, transaction size is more big, the corresponding increase of n, on the contrary then reduce; In the mode of iteration, application searches algorithm in the stock related network selects m to come foremost stock at every turn then; Add selected stock to crucial stock set, the selected stock of deletion and incidence edge from the stock related network enter the next round iteration again; Up to the crucial stock of selecting sufficient amount.

5. the stock market deep bid data analysing method based on crucial stock collection identification according to claim 4 is characterized in that the method that the application searches algorithm is searched crucial stock is: adopt the PageRank algorithm to come searched key stock; Given current stock related network, namely digraph G=(S R), sets up the adjacency matrix A of G, and matrix A is one | S| * | the 0-1 matrix of S|, wherein | S| represents the scale of stock node set S, and R is directed edge set; Element aij is defined as follows in the matrix A:

$a_{\mathrm{ij}}=\left\{\begin{array}{cc}1& <s_i,s_j>\&Element;R\\ 0& <s_i,s_j>\&NotElement;R\end{array}\right.;$

With the input as the PageRank algorithm of the transposed matrix of matrix A; In addition, the ratio of damping that the PageRank algorithm is set is α=0.3.

6. the stock market deep bid data analysing method based on crucial stock collection identification according to claim 5 is characterized in that, the process of calculating the views on broad market movements expectation according to crucial stock set in the step 3) is: divide two steps; The first step is calculated its price trend with regard to each crucial stock; Given crucial stock si at first obtains the si closing price of nearest 6 day of trade, accounting price tendency vector △ _i=＜δ _{I, 1}, δ _{I, 2}..., δ _{I, 5}, element δ i wherein, k(1≤k≤5) computing formula as follows:

Calculate the vectorial average of Δ i then

Represent the price trend of stock si, formula is as follows:

${\stackrel{\&OverBar;}{\mathrm{\&delta;}}}_i=\frac{1}{5}\underset{k=1}{\overset{5}{\mathrm{\&Sigma;}}}{\mathrm{\&delta;}}_{i,k};$

Second step is according to the price trend of every crucial stock, is weight with the total volume of nearest 6 day of trade, adopts the tendency expectation of calculated with weighted average method stock market deep bid; The total volume that makes stock si is Qi, and the quantity of crucial stock is m, and formula is as follows:

Images