TW200926039A - Method of predicting the high/low points of TAIEX based on grey and Markov theories - Google Patents

Abstract

The invention is to provide a kind of method of based on grey and Markov theories to predict the High/Low points of TAIEX, can trace and predict the variety of TAIEX, and make use of the method of artificial intelligence, predict the next turning point of TAIEX, to be used as the basis that the investor passes in and out the stock market. In better implement in the invention example, the history data of TAIEX is used as the primitive data, then make use of the method of the grey relational analysis (GRA) finds out the technical index of maximum relational degree with the TAIEX, immediately after then pick out the investigative data from the selected technical index, after then make use of the grey predict model GM(1, 1), grey metabolizing predict model GMeM(1, 1) and the hybrid predict model HGMeM(1, 1) to find out the optimal predict model, finally, the Markov model is developed to promote the prediction accuracy, to be used as the basis that the investor passes in and out the stock market.

Description

200926039 IX. Description of invention: [Technical field to which the invention belongs] Invention ί = An information prediction method' More specifically, this method is also used in the intelligent stock trend forecasting system to track the change of the method s 6' Analysis and forecasting to assess market transactions [prior art] ❹ ❹ 技术 技术 ' 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 评 可以 可以 可以 可以 可以 可以 可以 可以 可以 可以匕 市 市 = = = = = = = : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 2. Estimate the net surplus and dividends of the stock. 3. Estimate the stock (four) shape and #运 outlook and financial overview. Nanxun II. Estimating the future growth of the stock' Whether the P/E ratio is reasonable, the dividends are issued, and I, the other _ type targets on the market are compared. The data or description on the root t can be used as the basis for investors to judge the investment. § However, the basic analysis not only considers other external factors, such as interviews, 思思^素, but also the factors that can finally be decided: in the second-class, you must integrate all kinds of factors~ you will find other factors that will cause the market to miss the market. Therefore, because of the month b, when investing in the basic analysis, the investor 5 200926039 must have relevant financial knowledge to analyze the results, which is a big obstacle for the average investor. Generally speaking, the company's relevant information usually lags behind the market reaction. If the investor waits until the company announces the information, it will analyze and predict it. Usually, the market has already completed the reaction, and the result may cause chasing and killing, even Changes in economic system structure or unanticipated changes may lead to imbalances in market supply and demand, affecting price forecasts. 2. Technical analysis ❹ Technical analysis mainly uses graphical or quantitative technical indicators to judge the popularity or supply and demand of market commodity transactions. And then decide when to buy and sell. Because technical analysis can not only predict the compensation and judge the market investment Machine, and can be used as a warning signal, more importantly, the basic analysis is simple and easy to understand, so it is widely used by the general public. The following will introduce the methods of technical analysis commonly used in the market: 1. Dow theory (pow Theory The Dow theory is based on Trends and Patterns. The trend is composed of support lines and pressure lines, while the shape ❷ trend is divided into Continuation Patterns and Reversal Patterns. Two. The Bullish Market, the Bearish Market, the Support, the Resistance, the Head and Shoulders Formation, etc., which are well known to the public today, are all made by Dow. The theory develops. After understanding the Dow theoretical analysis method, combined with statistics, the results are derived from a variety of valence technical analysis methods. 2. Wave Theory was proposed by RN Elliott in 1943. This theory recognizes that the price of the market is cyclical and volatility, and that there is a total of eight waves in a complete cycle, including five main bands and A back-shift wave, a theory that combines trend lines, target area predictions, and natural laws, is one of the most commonly used analytical tools in today's market. 3. Moving Average Moving Average (ΜΑ) is the most widely used. The main reason for the technical analysis tool is that it is easy to calculate and easy to calculate. Since it calculates the average price of the stock over a period of time by statistical methods, and then connects the points into a smooth curve of φ, the moving average has smoothness. Characteristics such as stability and trend. The moving average is a reference for determining the timing of buying and selling. The eight rules proposed by American investment expert Joseph Granville are set for the timing of the trading of moving averages. 4. Volume (V0L) Use the volume moving average to determine whether the trend is up or down. When the amount of turnover increases, it means that the trend is upward, and when the volume is decreased, it means that the trend is downward. When stocks are going to rise, they must match the volume. If the stock price rises but the trading volume shrinks and the price is deviated, the stock price may only be a short-term rebound. Note that when the volume of trading increases, it does not necessarily guarantee that the stock price will rise. It is necessary to cooperate with other factors and technical indicators to judge whether the price will rise or fall.

5. RSI RSI uses the price movements of a certain period of time to predict the trend of future price changes. The basic principle is that in a normal stock market, the strength of both sides of the stock market must be balanced and the stock price will be stable. RSI is the ratio of the average value of the stock price increase to the average of the total increase and decrease in a certain period of time. Take the 6-day RSI as an example. If the value is 80, it will be overbought on 200926039, 90 or more will be the selling point; below 20 will be oversold, 10 or below will be the buying point; RSI will peak before the stock price changes. Or the bottom, can reflect the ups and downs of the stock price in advance, can be regarded as the leading indicator of the trend of the market index; if the RSI on the 6th through the 12th RSI from the bottom up, it can be regarded as a buy point; otherwise, when the RSI on the 6th goes from the top to the bottom When the 12th RSI is broken, it can be regarded as a selling point; the trend of the RSI line and the trend of the broader market index will indicate that the market is about to reverse.

6. KD φ combines the advantages of the RSI indicator, the moving average and the concept of quantity and energy. If the market is an obvious downturn, it will drive the squall line and the D line to rise. If the rally starts to be sluggish, it will reflect the devaluation and D value, so that the devaluation falls below the D value. Established; when the Κ line fell from the top to break the D line, and the D value is above 80 (overbought area), is the sell signal; when the Κ line breaks the D line from the bottom up, and the value is below 20 (oversold area) ) appears as a buy signal; when the Κ value is greater than 80, the D value is greater than 70, indicating that the closing price of the day is in the high price of the price area, that is, the overbought condition; when the K value is less than 20, the D value is less than 30, It means that when the closing price is in the low compensation zone, it is oversold; the price is high or low, and KD does not have this phenomenon. This is a divergence phenomenon, which is an important precursor to possible reversal. 7.MACD (DIF) uses the fast and slow two exponential smoothing moving averages (EMA) to calculate the difference between the two (DIF), and then use the difference and the difference mean (DEM) The signs of convergence (crossing) and divergence (separation) are used to judge the timing of stock market buying or selling. If both MACD and DIF are positive, they can be regarded as long market; MACD and DIF are negative, 8 200926039 can be regarded as the s head market; DIF breaks through the MACD and buys the signal; (4) sells the signal to the MACD of the decline; the value of the coffee changes from negative to positive, and crosses the MACD beacon signal; the mF value changes from positive to negative, and breaks through When the signal is sold, the DIF and the broader market index deviate from the trend, if the stock repays a continuous low of innovation' and the DIF value does not reach a new low point, this is the "positive deviation" trend, which is a feed machine; When the new high point, and the service value is not innovative, this is the "negative divergence" trend, which is the time to sell. ❹ 3 ίI There is a lot of information in the field to analyze 'of course, the related technology is two „„, here are just a few of the more commonly used technical indicators. ^It still has daily, weekly, monthly, financing, financing Volume...etc. How to deal with so much information? Although the computer can report to the valley. Various technical indicators, but the analysis of the simple technical indicators combines: the situation of post-discipline and indicator passivation occurs, and there is no self-description; : > positive this force' so the results of its analysis and prediction are very inaccurate. I hope to have artificial intelligence (Al), so that not only the ability to correct the 'can adapt to the various activities of the stock market' and then can be more Accurately predict the future trend of the stock market. The lack of surface gray system and Markov prediction are two types of applications. Both methods are suitable for time series prediction. Each chain f is just combined with gray (10) (10) prediction model = Markov chain Predictive method of pre-type advantage. 较 More suitable for data data and #合“ so-called “generalized m prediction, only a small amount of data (more than four) can be modeled as two = fit function is just – smooth index function The curve, for the random rut: the data fit is poor, and the prediction accuracy is also low. Markov 200926039 is like a dynamic system with random changes, the theoretical basis of which is the Markov process. An n-order Markov chain is determined by n state sets and a set of transfer probability =, the process can only be in one state at any time, and the future changes of the system can be predicted according to the transition probability between states. In the time series, the process is required to be stable, and the transition probability reflects the influence of various random factors. Therefore, the Markov chain is suitable for predictions with large randomness of data. Most = time series are non-stationary stochastic processes that exhibit a certain trend with time. The process will always be biased or oscillated around a certain β-winding trend with various random factors, so if different GM (1, O' model to obtain the trend of change in the process, and then use the Markov chain for the transfer profile, the combination of these two methods will improve the prediction accuracy of GM (U)' and expand The scope of application of GM(1,1). 【Abstract】 The flute has the ability of artificial intelligence, which can be used by investors to make predictions and use the gray correlation analysis and gray prediction in the grey theory. 6 know that 'partially unknown system is called gray stock market towel has a lot of unknown information, the wealth is often suitable, and the application of the mm (u) model based on the gray theory (four) The often used method has different practices from the same application; because the stock market has a non-stationary stochastic process that exhibits a certain trend, so the optimized GM(u) model and the horse are used. The chain correction method to carry out the stock market The main purpose of the two-in-one is to provide a method for tracking and forecasting changes in stock market information to help investors invest easily and reduce the investment risk and investment efficiency of 200926039. ^The other purpose is to provide a method to make the prediction model construction method more adaptive. ^When the book will be sent to the target market in the stock market to predict the future stock market trend to evaluate the market" The second method is to raise the two profitability and reduce the investment risk. In the above information, we can understand that the stock market has more than two phases at the same time and can track all of them in the long run. The physical test, the law on the miscellaneous line ί f for investors can immediately make the right vote (bei, shell or close), and no longer need to spend time to study various related information and reports. In the market, there is a lot of information in the market: the color system is suitable for the gray theory method. In this issue:: =: technology and the market refers to the heart = _ material, the second? H〇Me: 0^ 槎刑^ Find good predictions The model, and finally the combination of Markov's stock market, to improve the accuracy of the forecast, for investors to enter 200926039. [Embodiment] The following description of the preferred embodiment and the accompanying drawings will be fully described, but described herein. It is to be understood that a person skilled in the art can revise the embodiments described herein, and at the same time achieve the same effect of the invention. Therefore, it should be understood that the following description is a broad disclosure of those skilled in the art and The content is not intended to limit the present invention. The following is an extended description of the technology related to the background of the present invention. Although there is a good experience and knowledge of reading @者 can choose to skip only or not read the following background information, but understand this information The specific implementation of the invention can be further understood and the recommendations should be read in detail. Referring first to Figure 1, there is shown a flow chart of the method of the present invention. In the embodiment of the method of the present invention, the historical data of the moving average (ηΜΑΡ) of the market index 2 and the price of different days in the stock market is first used as the raw data 11. In the case of inputting data in this embodiment, the information is input using the market index and different daily average lines, such as the 24-day moving average (24ΜΑΡ), the 72-day moving average (72ΜΑΡ), the 144-day moving average (144ΜΑΡ), and the 288-day moving average ( 288ΜΑΡ), etc., the present invention can be divided into three parts to add ® to discuss, and a part of it is gray correlation analysis 12 (refer to Figure 3 for detailed process). In the original data 11, the preferred embodiment of the present invention tests the Taiex and the different daily averages (ηΜΑΡ). In the preferred embodiment, the daily average takes 24 ΜΑΡ χ, 72 ΜΑΡ χ, 144 ΜΑΡ Χ, 288 ΜΑΡ χ χ 均. The η in the above ηΜΑΡ_χ indicates that the η-day moving average 'X indicates the length of the time interval, and the time interval is 6, 12, 24, 72 days. Using the gray correlation analysis method, the grey relational degree (GRG) of the reference sequence (ηΜΑΡ_χ) and the comparison series (the market index) can be divided into 12, 200926039, which can be regarded as the maximum correlation between the 4 test series and the comparison series, that is, (10) The party's input information. Calculate the formula of GRG in this (4). It is known to use the distance, slope, area, etc. to calculate the difference, but the / order m, the result of the calculation method will be different. The distance from the GRG is not, so this is the case. The invention directly relates to the gray correlation analysis: Step J: The following 3:: Programs 31, 32. ❹ (x〇w) ^ , .012 "'(4)) factor is determined, where yx and value, .''.··'" 'w), x'w • respectively represent the mathematical column at I point and after comparing the series The number of columns must be used as the method of front-end association generation, so that the reference number column has the same physical meaning or unit of measure as the comparison series. 3·^ After the pre-treatment, 'you can find the gray correlation coefficient of the pair's 'f ey R, nal CWfieient, GRC) 'The formula is the process r of Fig. 3. No, "Where, ζ is the resolution coefficient (Distinguishing 7T e ^ ζ = 〇·5, Δ^ type and different calculation methods. 4. After finding the gray correlation coefficient, the gray relational grade (GRG) between the series can be obtained. χ is all information The sequence r ^ set ' ζ · = 0,1,2,; take its td 'training ice' as the value at point k. If it satisfies the procedure as in Figure 3, the order '32' is done., \) For χ, the gray correlation degree of _ to X. The other part is the analysis of different prediction models 13, 14, and 15. For each prediction 13 200926039 杈, the input data is the optimized comparison series (24MAP) 21 High = the number of days generated by points 22, 23. In the GM (U) prediction model η y, after obtaining the relevant data from the ancient lows of the Republic of China from May 86 to October 1993 relative to the first day of the first day, The sequence of the GM(U) prediction model is as follows. The following is the 13th of the icon number of the dynasty 1: 1. Obtain the data needed for the calculation, and let X〇=(^o (lXx (2),....,x〇(A:),,χ.(8)) (J) 2. Perform the 1-AGO operation on it to get the following expression ^〇) =(^^(1) ,^(2),^(3),.·.··,(8)) (2) where ki) — ΣΊ =1, 2ν··, \ μ {ό) Lu 3· by confrontation (1 ) as the meam generating operation, get (3), ....., z%)) m where () 4. The original equation of GM(1,1) modeling is as follows (household ^

Z.(1) No 5. The solution for generating the parameters α and δ is the root 摅Θ.γ, the method of the 崎 W W flat { {ττγγΒτΥΝ where dx{l) dt αχ (1)_ (6) 14 (7) 1 (〇)(A:) + az(l)(A:) = 6, k = 2,3,'·. „ -(2) 200926039

B 7(]) -(2)

v(3), (8) 6. After the operation of equation (7), the value of the parameter "^ is obtained as follows JL ri k=2

(〇) W (9) 6=Σ«Σ«2-Σ>ϊ)1>((μ -Λ=2 Δ (10) where 厶= i=2 (11) Ο will be the value of a, b The final prediction produced by the shadow equation (6) (Moxibustion +1) = [/.) (1) _ A] 〆 + A (12) aavy where 'α) = χ (〇) (1) (π) 8. Yes, ( λ) ("ι" do 1-IAGO last available predictive value 〇 t) is ^\k) = x(l\k)-xm{k-\) = (14) a Value χ(.)(9) and predicted value p(error analysis between illusions 15 200926039 xi〇)(k)-x(0\k)~^\k) ~xl00%, k = 2,3,4, , n. (15) ❹ In the GMeM (l, l) gray metabolism prediction model 14 analysis, after obtaining the relevant data of the high and low points of the Republic of China from May of the 86th year to the first day of the year of 1993, relative to the first day, The date of occurrence of each high and low point minus each day, the number of the data section can be modeled. The modeling data is 3~Ν·1, N is the maximum amount of data selected, from which Calculate the optimal number of points (data amount) for predicting the next high/low point for use in the HGMeM(U) hybrid prediction mode. The process of establishing the (10)^(3) model is as follows: (16) L order W is the original sequence where x = 1,.·· ·, "} Metabolic sequence 2. Remember ~ (.) is the first in x (°) & +1 ^ (〇) 3 · When (17) (18) ,) (1 ί + 丨) = ^ (2, Pot K = const (19) K = el is called. (〇) O+1(0),v,ei 16 (20) 200926039 is the metabolism, called {x,(0)|z-6i} = {x , (0)} (21) is the metabolic subfamily of χ(.) 4. Modeling the metabolic subfamily of X(9) (22) 〇GM 〇JGO:工'(.)-^(0)(„, +1 ) ^+1 = 1 +j, 4<j^nl where iw(«, +l) is the predicted value ❹ 5. Use the residual test formula q{0\i) = χ(〇)〇+ /)~^(〇)〇' + /)

Pot I

Xi〇)0 + J) -Σ 丨?(0)(7) I xlOO% (23) l€l I = {1,2,····,«-;'} Detectable GMeM(l,l) model does it reach the requirement. In the HGMeM (l, l) gray hybrid prediction model i5 analysis, it combines the GMeM (l, l) model with the gmg, ;!) model, the modeling of the two process can refer to the aforementioned establishment of GM (1,1) ) Detailed steps with the GMeM(l,l) model. The GMeM(U) predictive model can be used to find the best modeling point. Then we can use the GM(1,1) model to obtain more accurate prediction results. The two-dimensional verification method uses the HGMeM(l,l) prediction model to obtain an excellent prediction accuracy as a basis for investors to buy or sell. According to a preferred embodiment of the present invention, the target predicts that the market will obtain historical data of the information index to be predicted in the period between the Republic of China/13th and the high/low point of the Republic of China on August 18, 1993, assuming that the Republic of China knows . The value of the market change from H to the country of April 29, 1993 is the number of days that have been compared with the number of comparisons (four) 24MAP all the high and low points of production 17 200926039 (refer to the above description), so the establishment of GM (1,1) model The required information is 24 MAP between the Republic of China on May 3, 1986, and the Republic of China on April 29, a total of 20 high turning points, and between June 5, 1986, and Republic of China, April 15, 1993. A total of 19 strokes resulted in a low turning point. After obtaining the modeling data, these series are substituted into different prediction models, namely GM (1, 1) prediction model 13, GMeM (l, l) prediction model 14 and HGMeM (l, l) prediction model 15. Then, through the residual test rule, see equations (15) and (23), and finally the best φ prediction model and high/low point modeling points are HGMeM(l,l) and 5/4 points, respectively. After the three models of prediction models 13, 14, and 15 are tested, the best prediction model is HGMeM(l,l). Although this model can improve the accuracy of prediction, the stock market has time to change. A non-stationary stochastic process of a certain trend is not a smooth exponential function curve, so it is necessary to systematically correct the predicted value to improve the prediction accuracy. The present invention is based on the general characteristics of the stock market, so the gray is adopted. The Markov chain method is used to correct the shortcomings of the HGMeM(l,l) ® predictive model modeling. The other part is the GMaM (l, l) gray Markov prediction model. Since the Markov process has a statistical concept, it is necessary to use the full segment data and then analyze the predicted values generated by HGMeM(l,l) modeling. After obtaining the predicted values of the data in all the sections, the Markov chain method can be used to improve the prediction results of the prediction model to expand the application of different gray prediction models GM(l,l), GMeM(l,l), and the like. The steps for establishing the gray Markov model GMaM(l,l) are as follows, refer to the 4th icon number 41, 42 , 43 , 44 , 45 : 18 200926039 1. The optimization prediction model is established 41 Optimized prediction The model is established by one of the GM(1,1) prediction model 13, the GMeM(l,l) prediction model 14 and the HGMeM(l,l) prediction model15, and the current optimized prediction model. For the HGMeM (l, l) prediction model, of course, it may be one of the other two prediction models, depending on the experimental results, please refer to the detailed steps mentioned above for the method of establishing each prediction model. 2. Division of state 42 g Divide each moment in the process into s states; to be the moment: the center point of each state, and take the appropriate percentage (P%) as the upper and lower bounds of each state at each moment, If & represents the jth state at time i, then ΕνΕ[Αν, Βνυ = \, 2,..., 5 (24) where is the lower and upper bounds of the jth state at time i. E.g

3. Establish state transition probability matrix 43 If the state transition probability is recorded as Ρίπ,) =

(25) where /Γ denotes the probability of transitioning from state i to state j via m steps; 19 200926039 is the number of times state i has transitioned to state by m steps. Then, the state transition probability matrix can be recorded, and it is rolled into the state i. (2ϋ ·"^ ? ? 9 o )) ^--1 fl 4·Making the prediction table 44 Selecting the r profit moments closest to the prediction time, Close and far, 瘅斟 你 你 amp amp amp amp amp amp amp amp amp amp amp amp amp amp amp 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵 矩阵The state corresponding to the column I;: and the new probability matrix 'for the direction of the new probability matrix' and the corresponding state of the larger is the predicted state. example 1-

ο j2〇211 Total 5. Calculate the predicted value 45 After determining the future transition state, it can be determined that the top/lower bound of the status area is the most likely predicted value, which can be regarded as the midpoint of the interval. The average of the bounds and the lower bounds. 20 200926039 =0.5 χ (4+^) or (ρ丨+Pj) 93/=ϋ is better, 'the target predicts that the market is high in the Republic of China Ιί(3]νί,ιν^8/18 appears high/low At the point of time, after completing the relevant value of all the materials obtained by the relevant capital construction, the four calculation steps of 42-, 43, 44, and 45 of the advance-repair number are used to improve the accuracy of the prediction, that is, to produce the best. Pre-point: im: 'When there is a market, there will be a lower-high/low transition; second, the unwillingness to buy or sell the futures index early, and get the return on investment from the quarterly return. · Can pass through 3 = Master The preferred embodiment of the 'Gray Markov Model G ride (U) is corrected by the predictive value of the formula and the ratio of the steam, the number of states, and the number of transition steps m, ^: ^ upper and lower bounds = To produce = pre-:: type: the rolling model of the measured value =, each main = upper and lower bounds, the ratio of the number of states i and the invariant of the heart-turning will follow the data amount of the prediction model: :: fixed, the prediction model GMaM (u) can automatically generate the best pre-audit fruit according to the data volume of the follower's text. 17 Pre-type is used as shown in Figure 2 for the large-scale κ line graph , Ganzi time line (5/10/20.../18〇 line line ^ = axis is different and all are used 曰 κ line graph 曰 all = lose = points: in this ^ month the average line combination as a forecast Judging the conditions. Dongtie, the use of the average line of observation can also use other time-sharing, 曰,: one, the tendency of the tribute to break the index, the judgment method is also like the ^ ^ ^ ^ to judge 21 200926039 in this In the preferred embodiment of the invention, the Taiex and the different averages (ηΜΑΡ) are selected as the source of the original data, and all the analysis is for the two items. The investor can change it to other technical indicators, such as MACD, Technical indicators such as RSI, or other averages or indicators added to the analysis, may lead to better prediction results. Of course, this is verified by experiments. In addition, in the preferred embodiment of the present invention, the selection factor (Taiex) is selected. Different daily averages (nMAP) are tested separately. 'Investors can change them to other factors, such as financing or other technical indicators, or add other factors such as the average and technical indicators in the analysis. 'The rest of the practice with the aforementioned stream In the preferred embodiment of the present invention, the gray correlation analysis (GRA) is used to find the 24MAP moving average with the highest degree of association with the market, and then via GM(l,l)/GMeM(l,l)/H GMeM(l, l) Produce the best predictive model' and then use the gray Markov method to correct the best predictive model. The final predicted result is better than the non-invention method. Therefore, the present invention enables investors to Knowing the buy/sell signal early, providing investors with a low-risk and highly profitable investment method. In the preferred embodiment of the invention, because the forecast of the market index turning point θ is long-term rather than short-term or long-term The market changes are not a idling' so the timeline of the moving average should not be too short or too long. If: 迅月巨 is being bullied by short-term fluctuations, and too long, it will not be able to respond quickly to _Z, so this The length of the four groups of moving averages as the rounds of the invention is 6Λ2/24/72 days. Bessie's and the average of the oysters + f Ming, a more common example, the C process refers to the 4th 42nd, 43, 44, 45) 'Using the Markov chain method to correct the best 22 200926039 type, and finally the predicted results Among them, the optimal upper and lower boundary percentage P values, the number of states i, and the number of transition steps m are 4%, 3, and 3 steps, respectively. The embodiments of the present invention have been described in detail in the foregoing embodiments, and those skilled in the art can make modifications and modifications of the present invention as needed without departing from the spirit and scope of the invention. The embodiment is also included in the scope of the patent application of the present invention. Based on the above, the present invention is based on the method of gray and Markov theory for predicting the turning point of the stock market, that is, using artificial intelligence to predict the future stock market turn point, not only allows investors to understand the future stock market trend and grasp the investment first. Learning can further reduce investment risks and losses. It can be understood from the above that the present invention has many excellent characteristics, not only can solve the lack and inconvenience of some practical applications, but also provides a cost-effective solution, and allows investors to obtain relevant investment information in a short time, Avoiding the influence of the external environment and human factors, and thus effectively controlling the risk of investment, it has already met the application requirements of the invention patents, and the bureau can be given a detailed review and patent protection to provide benefits to the people. 23 200926039 [Simple description of the diagram] Fig. 1 is a flow chart of the analysis and prediction method of the present invention. Figure 2 shows the K-line chart of the Republic of China 85/5~87/4. Figure 3 is a flow chart of a conventional gray correlation analysis. Figure 4 is a flow chart for the establishment of the gray Markov prediction model. [Main component symbol description] Large-cap index ~2; Raw data ~11; ® Gray correlation analysis (GRA)~12; Gray prediction model GM(1,1)~13; Gray metabolism prediction model GMeM(l,l)~14 Gray mixed prediction model HGMeM(l,l)~15; gray Markov prediction model GMaM(l,l)~16; best prediction result~17; 24MAP moving average ~21; 24MAP high point~22; 24MAP low point ~ 23; gray gray correlation coefficient formula ~ 31; gray correlation degree formula ~ 32; best predictive model establishment ~ 41; state division ~ 42; establish state transition probability matrix ~ 43; make prediction table ~ 44 ; Calculate the predicted value ~45. twenty four

Claims (1)

200926039 X. Patent application scope: 1 · A method based on grey and Markov theory to predict the turning point of the stock market, used to evaluate the future turning point of the stock market's market index, including the following procedures: Collecting the history of different technical indicators of the stock market Data, the historical data is the various moving average values of the moving average price; using gray correlation analysis (GRA) to analyze different moving average prices, from which the best moving average price is generated as different prediction models 〇 = (1, 1), (^ say (1, 1) and shame 1 ^ ie, 1) modeling data to produce the best prediction model; and use the Markov chain method to correct the prediction of the best model described above Values to produce the best predictions. 2. The method based on the grey disk Markov = forecasting the stock market turning point as described in item 1 of the patent application scope, wherein the technical indicators collected are the market index and different moving average prices. 3. If the method of forecasting the turning point of the stock market is based on the grey and Markov theory mentioned in item 1 of the patent application scope, the range of the input area of the historical data may be determined by itself. 4. The method for predicting the turning point of the stock market based on the grey and Markifu as described in the scope of patent application ,1, wherein the moving average moving average of historical data is 6/12/2472 曰 moving average. 5. If the scope of patent application 帛1 is the basis; ^Gray and Markov 2 method for predicting the turning point of the stock market, which can be analyzed by the correlation of the gray and the average moving price of the market index A 24MAP. 6. If the gray and Markov theory is used to predict the big (four) breakpoint of the stock market as described in the scope of patent application, the high and low points of 25 200926039 24MAP in the range selected by the towel are 21 and 2 points respectively. Please fill in the paradigm (4) 1 rhyme to listen to the gray and Markov = the method of the market turning point'. The prediction model used to generate the best prediction karyotype can be the gray prediction model, the υ or the gray hybrid prediction model (1, = The method based on the gray and Markov shares in the first application of the scope of patent application! The method of the market turning point, the best prediction model is gray / ttj & t forecast weight HGMeM ( 1,1) 〇 9. If the scope of patent application帛i 盥 可: The method of theoretically predicting the turning point of the stock market, Ma Fufu's forecast value, can be the best prediction result of the model. The key method is to correct it to produce the most 10. The model of the method of predicting the turning point of the stock market in the kiln theory mentioned in the scope item is called gray horse; = Fu and Ma Ke method can automatically calculate the best upper and lower = sub-chain and transfer steps m, to produce the best prediction results. & solid number 1 The method of predicting the stock market turning point by the Keefer theory mentioned in item i of the patent scope of the patent 'The Marco chain method produces the best prediction results of each state, the husband P 4%. Take 1^ upper and lower bound percentage 13. If the patent application scope i-term theory predicts the stock market turning point, = gray, and Marco can use the Markov 26 200926039 chain method to produce the best prediction results The number of states is H. If the method of predicting the turning point of the stock market is based on the gray-ghost theory as described in the patent application scope, the number of transfer steps using the color-and-mass method to produce the best prediction result is 3 steps. ',,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,