Stock transaction behavior analysis method and system driven by big data mining

Abstract

The invention relates to the technical field of financial big data analysis, and provides a big data mining-driven stock transaction behavior analysis method and system, and the method comprises the steps: obtaining stock price data of a period of time; dividing to obtain a plurality of periods and generating a Boforest belt; dividing the period into a plurality of random stages, determining the price oscillation degree of each random stage based on the number of times that stock price data in the same random stage breaks through the upper and lower rails of the Boforest belt and the amplitude and duration of each breakthrough, and screening price oscillation stages; acquiring a boarding and alighting confirmation rail for boarding and alighting rails corresponding to each price oscillation stage according to the price oscillation degree; analyzing the duration difference of each breakthrough in the same price oscillation stage, and determining the breakthrough price stability degree of each price oscillation stage; and judging the stock price data based on the Boforest belt. The objective of the invention is to solve the problem of misjudgment of the Bucforest belt on the transaction behavior under short-term high-frequency oscillation of the stock transaction price.

Description

Technical Field

[0001] The present invention relates to the technical field of financial big data analysis, and particularly to a method and system for analyzing stock trading behaviors driven by big data mining. Background Art

[0002] In the field of stock trading, "driven by big data mining" means that through in-depth analysis of massive historical data, real-time data, and multi-dimensional data, valuable information is extracted using big data mining technology to provide support and guidance for stock trading; and the core role of big data mining driven is to extract valuable information by deeply analyzing massive market data to help achieve intelligent trading and conduct effective risk management. These analyses and mining can cover multiple aspects such as the historical behaviors of stocks, market sentiment, capital flows, technical indicators, etc., so as to improve the speed of investors' response to the market. In the field of stock trading, the fluctuations of stock prices often contain the most market information, and the Bollinger Band judges the range of price fluctuations through the standard deviation of stock prices and sends signals of overbought or oversold in the market, thereby achieving the purpose of analyzing trading behaviors and judging risk behaviors during the trading process.

[0003] During the process of judging stock trading data behaviors through the Bollinger Band, due to objective reasons such as the market economic cycle or market policies, when the stock price is in a short-term shock state, the upper and lower tracks of the Bollinger Band will frequently show "overbought" or "oversold" signals, leading to misleading judgments of trading behaviors; and under short-term shocks, the fluctuations of stock prices do not form a clear upward or downward trend, and the price fluctuations are limited to a certain range. Therefore, short-term shocks will cause the Bollinger Band to be unable to accurately judge the breakthrough points of stock prices, resulting in misjudgments of trading behaviors. Summary of the Invention

[0004] The present invention provides a method and system for analyzing stock trading behaviors driven by big data mining to solve the problem that the Bollinger Band misjudges trading behaviors under short-term high-frequency shocks of existing stock trading prices. The specific technical solutions adopted are as follows:

[0005] The present invention proposes a method for analyzing stock trading behaviors driven by big data mining, which includes the following steps:

[0006] Obtain stock price data for a period of time;

[0007] Divide the period of time to obtain several cycles and generate their Bollinger Bands; divide the cycles into several random stages, and based on the number of times the stock price data in the same random stage breaks through the upper and lower tracks of the corresponding cycle, as well as the amplitude and duration of each breakthrough, determine the price shock degree of each random stage, and screen the price shock stages;

[0008] For the upper and lower tracks of the Bollinger Bands corresponding to each price oscillation stage, according to the degree of price oscillation, obtain the upper and lower confirmation tracks of each price oscillation stage; analyze the duration differences of each breakthrough in the same price oscillation stage to determine the stability of the breakthrough price in each price oscillation stage.

[0009] Based on each breakthrough in the random stage and the stability of the breakthrough price in the price oscillation stage, combined with its upper and lower confirmation tracks, make a judgment on the stock price data based on the Bollinger Bands.

[0010] Optionally, the method for obtaining several periods and generating their Bollinger Bands specifically includes:

[0011] Divide the period of time into several periods according to the period length;

[0012] For the stock price data within any period, obtain the middle track of this period by the moving average method according to the time sequence;

[0013] Calculate the standard deviation of all the stock price data within this period, and take the curve obtained by adding K times the standard deviation to the middle track as the upper track of this period, and take the curve obtained by subtracting K times the standard deviation from the middle track as the lower track of this period. The middle track, upper track and lower track of this period together constitute the Bollinger Bands of this period.

[0014] Optionally, the method for dividing the period into several random stages and determining the degree of price oscillation of each random stage based on the number of times the stock price data breaks through the upper and lower tracks of the Bollinger Bands corresponding to the period within the same random stage, as well as the amplitude and duration of each breakthrough, specifically includes:

[0015] Perform random stage division on each period respectively, and obtain several breakthroughs of the stock price data in each random stage with respect to the upper and lower tracks of the Bollinger Bands corresponding to the random stage;

[0016] Based on the number of breakthroughs and the duration of each breakthrough within the same random stage, obtain the breakthrough randomness of each random stage;

[0017] On the basis of the breakthrough randomness, combined with the amplitude of each breakthrough within the same random stage, obtain the degree of price oscillation of each random stage.

[0018] Optionally, the method for specifically obtaining several breakthroughs of the stock price data in each random stage with respect to the upper and lower tracks of the Bollinger Bands corresponding to the random stage is as follows:

[0019] Based on the upper and lower tracks of the Bollinger Bands of any period, if any stock price data in any random stage of this period is greater than the upper track of the Bollinger Bands of this period or less than the lower track of the Bollinger Bands of this period, record this stock price data as a breakthrough;

[0020] If multiple consecutive stock price data break through the upper track or the lower track, the multiple consecutive stock price data are jointly recorded as one breakthrough, and the product of the number of the stock price data that break through and the update frequency duration is recorded as the duration of this breakthrough.

[0021] Optionally, the specific method for obtaining the breakthrough randomness of each random stage includes:

[0022] Obtain the average value of the durations of all breakthroughs in any random stage within any period;

[0023] Based on the ratio of the number of breakthroughs in this random stage to the average value of the durations of all breakthroughs, obtain the breakthrough randomness of this random stage.

[0024] Optionally, the specific method for obtaining the price oscillation degree of each random stage includes:

[0025] Obtain the absolute value of the difference between each stock price data of each breakthrough in any breakthrough within any random stage in any period and the upper track or the lower track of the Bollinger Bands of this period as the deviation amount of each stock price data of the breakthrough, and take the average value of all the deviation amounts in this breakthrough as the amplitude of this breakthrough;

[0026] Based on the amplitudes of all breakthroughs in this random stage and in combination with the breakthrough randomness of this random stage, obtain the price oscillation degree of this random stage.

[0027] Optionally, the specific method for obtaining the upper and lower confirmation tracks of each price oscillation stage is as follows:

[0028] For any price oscillation stage in any period, take the sum of 1 and the price oscillation degree of this price oscillation stage as the upper track adjustment coefficient of this price oscillation stage; take the difference obtained by subtracting the price oscillation degree of this price oscillation stage from 1 as the lower track adjustment coefficient of this price oscillation stage;

[0029] Take a new curve formed by multiplying the values corresponding to each time stamp in this price oscillation stage of the upper Bollinger Band of this period by the upper track adjustment coefficient respectively as the upper confirmation track of this price oscillation stage;

[0030] Take a new curve formed by multiplying the values corresponding to each time stamp in this price oscillation stage of the lower Bollinger Band of this period by the lower track adjustment coefficient respectively as the lower confirmation track of this price oscillation stage.

[0031] Optionally, the specific method for analyzing the duration differences of each breakthrough in the same price oscillation stage and determining the breakthrough price stability degree of each price oscillation stage includes:

[0032] Obtain the difference between the duration of each breakthrough in any price oscillation stage of any period and the breakthrough duration threshold, and obtain the breakthrough price stability degree of this price oscillation stage based on the difference.

[0033] Optionally, based on each breakthrough in the random stage and the breakthrough price stability degree of the price oscillation stage, combined with its upper and lower confirmation tracks, judge the stock price data based on the Bollinger Bands. The specific method included is as follows:

[0034] For any breakthrough in any random stage that is not a price oscillation stage in any period, send an "overbought" or "oversold" signal.

[0035] For any price oscillation stage in any period, if the breakthrough price stability degree of this price oscillation stage is greater than or equal to the breakthrough price stability threshold, eliminate the upper and lower confirmation tracks of this price oscillation stage. Each breakthrough in this price oscillation stage is judged based on the corresponding upper and lower tracks, and each breakthrough is beyond the range of the Bollinger Bands, then send an "overbought" or "oversold" signal.

[0036] If the breakthrough price stability degree of this price oscillation stage is less than the breakthrough price stability threshold, for each breakthrough in this price oscillation stage, judge it with the upper and lower confirmation tracks of this price oscillation stage. If it is greater than the upper confirmation track or less than the lower confirmation track, send an "overbought" or "oversold" signal for the corresponding breakthrough. If it is beyond the range of the upper and lower confirmation tracks, do not send an "overbought" or "oversold" signal.

[0037] The present invention also proposes a stock trading behavior analysis system driven by big data mining. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the above method.

[0038] The beneficial effects of the present invention are as follows: In the process of using the Bollinger Bands of stock price data to judge stock trading behaviors, by analyzing the characteristics of high-frequency and short-duration stock price breakouts through the upper and lower tracks of the Bollinger Bands under short-term oscillations, the price oscillation stage is determined and the upper and lower confirmation tracks are amplified and obtained; among them, after dividing the stock price data into cycles and obtaining the Bollinger Bands respectively, the breakouts of the stock price data in each random stage of each cycle to the upper and lower tracks of the Bollinger Bands are quantified. The greater the breakout amplitude and the shorter the duration, and the more frequent the multiple breakouts, showing the random fluctuations of the stock price, the more in line with the characteristics of short-term oscillations. Based on this, the price oscillation stage in the random stage is quantified, providing a basis for subsequently adding the upper and lower confirmation tracks to eliminate the influence of short-term oscillations; after amplifying the upper and lower tracks based on the price oscillation degree in the price oscillation stage to obtain the upper and lower confirmation tracks, analyze the duration differences and changes of each breakout in the same price oscillation stage. The more breakouts with longer durations make the price oscillation stage less in line with short-term oscillations, and vice versa, the upper and lower confirmation tracks need to be eliminated to avoid misjudging the stock price fluctuations under non-short-term oscillations; thereby improving the accuracy of using the upper and lower tracks of the Bollinger Bands to judge stock price fluctuations, and thus correctly sending "overbought" or "oversold" signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0040] Figure 1 It is a schematic flowchart of a method for analyzing stock trading behaviors driven by big data mining provided by an embodiment of the present invention;

[0041] Figure 2 It is a schematic diagram of stock price trends and their Bollinger Bands. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0043] Please refer to Figure 1 , which shows a flowchart of a method for analyzing stock trading behaviors driven by big data mining provided by an embodiment of the present invention. The method includes the following steps:

[0044] Step S001: Obtain stock price data for a period of time.

[0045] The purpose of this embodiment is to adjust the Bollinger Bands generated based on stock prices in the field of stock trading to improve their robustness; when the upper and lower tracks of the Bollinger Bands are broken by the stock price, "overbought" or "oversold" signals will appear, and the short-term fluctuations of the stock price will cause such signals to appear frequently, affecting the judgment of stock trading. Therefore, it is necessary to eliminate the influence of short-term fluctuations of the stock price on the judgment of the Bollinger Bands, so it is necessary to obtain stock price data for a period of time and quantify its data changes.

[0046] Specifically, obtain the real-time trading data of stock prices through the API interface, where the update frequency is set to 5 minutes. The stock price data includes the opening price, closing price, highest price, lowest price of the stock, and the stock price at each update; at the same time, it is necessary to obtain stock price data for a period of time to quantify the changes in stock prices. In this embodiment, the stock price data for the past 180 days is obtained through the API interface, and the update frequency of the stock price is the same as the update frequency for obtaining real-time trading data; then the stock price data for the past 180 days and the real-time obtained stock price data, through preprocessing, including removing outliers, filling in missing data, and unifying the corresponding timestamp format, together constitute the stock price data for a period of time and are stored in the database; the preprocessing method is a well-known technology and will not be elaborated in this embodiment.

[0047] Step S002: Divide the said period of time to obtain several cycles and generate their Bollinger Bands; divide the cycle into several random stages, and based on the number of times the stock price data breaks through the upper and lower tracks of the corresponding cycle in the same random stage, as well as the amplitude and duration of each breakthrough, determine the price oscillation degree of each random stage, and screen the price oscillation stages.

[0048] Preferably, in an embodiment of the present invention, dividing the said period of time to obtain several cycles and generate their Bollinger Bands includes the following specific methods:

[0049] It should be noted that the calculation of the Bollinger Bands is based on the stock price data within a specific cycle. If the cycle parameter is too long, the Bollinger Bands cannot quickly respond to the short-term fluctuations of the stock price, resulting in a lag in trading signals. Therefore, it is necessary to set appropriate cycle parameters and obtain the Bollinger Bands for each cycle by dividing the stock price by cycle; that is, it ensures that the short-term fluctuations of the stock price can be monitored in a timely manner, and at the same time, it is also convenient to identify and quantify the influence of short-term oscillations of the stock price on the "overbought" or "oversold" signals of the Bollinger Bands.

[0050] Specifically, preset the cycle length. In this embodiment, the cycle length is described as 20 days, and then divide the said period of time into several cycles through the cycle length.

[0051] It should be further noted that the Bollinger Bands consist of three lines, namely the middle band, the upper band and the lower band of the Bollinger Bands. The middle band is generated by calculating the moving average of the stock price data within a period, while the upper band and the lower band are generated by adding or subtracting K times the standard deviation of the stock price data to / from the middle band respectively.

[0052] Specifically, taking any period as an example, the middle band of this period is obtained by the moving average method according to the time sequence of the stock price data within this period; the standard deviation of all the stock price data within this period is calculated. In this embodiment, the value of K is described as 2. Then, the curve obtained by adding 2 times the standard deviation to the middle band is used as the upper band of this period, and the curve obtained by subtracting 2 times the standard deviation from the middle band is used as the lower band of this period. The middle band, the upper band and the lower band of this period together constitute the Bollinger Bands of this period.

[0053] Furthermore, for the period that is updated in real time, that is, the period containing the currently real-time collected stock price data, its Bollinger Bands need to be updated in real time every time the stock price data is updated and obtained; then the Bollinger Bands of each period are obtained according to the above method; as Figure 2 shown, it shows the trend of the stock price changing with time, as well as the schematic diagram of the corresponding Bollinger Bands.

[0054] It should be noted that the fluctuations presented by the short-term fluctuations of the stock price will cause frequent breakouts of the upper and lower tracks of the Bollinger Bands corresponding to the corresponding period, and then return to the normal price fluctuation changes after the short-term fluctuations. Furthermore, due to the frequent breakouts of the upper and lower tracks, "overbought" or "oversold" signals are frequently sent. However, the actual short-term stock price fluctuations usually have a certain degree of randomness, and the trading signals sent by them have a certain degree of misleading to the stock price fluctuations. Therefore, it is necessary to quantify the randomness to eliminate the influence of short-term fluctuations; at the same time, in the process of analyzing each breakout, the larger the cumulative duration of the breakout, the more it indicates that the breakout continues. The duration can reflect the stability of the stock price to a certain extent. The larger the duration, the more stable the corresponding breakout, and the less random it is. It is necessary to avoid misjudgment of such stable breakouts while eliminating short-term fluctuations.

[0055] Preferably, in an embodiment of the present invention, the period is divided into several random stages. Based on the number of times the stock price data in the same random stage breaks through the upper and lower tracks of the Bollinger Bands corresponding to the period, as well as the amplitude and duration of each breakout, the degree of price fluctuation of each random stage is determined, and the price fluctuation stages are screened. The specific method included is as follows:

[0056] Each period is divided into random stages respectively, and several breakouts of the stock price data in each random stage through the upper and lower tracks of the Bollinger Bands corresponding to the random stage are obtained;

[0057] Based on the number of breakouts and the duration of each breakout in the same random phase, the breakout randomness of each random phase is obtained;

[0058] Based on the breakout randomness, combined with the amplitude of each breakout in the same random phase, the price oscillation degree of each random phase is obtained;

[0059] Based on the price oscillation degree, several price oscillation phases are screened from the random phases.

[0060] As an example, the random phases are divided for each cycle respectively, and several breakouts of the stock price data in each random phase with respect to the upper and lower Bollinger Bands of the cycle corresponding to the random phase are obtained. The specific method includes:

[0061] A preset phase length is set. In this embodiment, the phase length is described as 2 hours. Since the update frequency of the stock price data is once every 5 minutes, any cycle is divided into several random phases by the phase length, and each random phase contains 24 stock price data arranged in time sequence.

[0062] Furthermore, based on the upper and lower Bollinger Bands of the cycle, if any stock price data in any random phase of the cycle is greater than the upper Bollinger Band of the cycle or less than the lower Bollinger Band of the cycle, then this stock price data is recorded as a breakout; if multiple consecutive stock price data all break through the upper band or all break through the lower band, then the multiple consecutive stock price data are jointly recorded as a breakout, and the product of the number of the breakout stock price data and the update frequency duration is recorded as the duration of this breakout. If a breakout only contains one breakout stock price data, then its duration is 5 minutes.

[0063] As an example, based on the number of breakouts and the duration of each breakout in the same random phase, the breakout randomness of each random phase is obtained. The specific method includes:

[0064] Taking the nth random phase in the cycle as an example, the breakout randomness r n of the nth random phase is calculated as follows:

[0065]

[0066] where q n represents the number of breakouts in the nth random phase of the cycle, t n,m represents the duration of the mth breakout in the nth random phase; norm() represents a linear normalization function, and the normalization object is all random phases in the cycle Obtain the breakthrough randomness of each random stage in each cycle according to the above method. In particular, it should be noted that the length of the random stage containing real-time updated stock price data may be less than two hours. If it is less than two hours, it is directly used as a random stage for the above processing.

[0067] It should be noted that the more times the stock price data breaks through the upper and lower tracks of the Bollinger Bands in the random stage, the greater the randomness of the stock price data in this random stage, that is, frequent breakthroughs of the upper and lower tracks; and after the breakthrough, if the duration of each breakthrough is relatively large and accumulates to a large duration, the larger the duration indicates a stable stock price after the breakthrough, and then there is a continuous breakthrough of the upper or lower track. At this time, the stock price fluctuation is not random and does not conform to short-term oscillations, so the breakthrough randomness needs to be reduced.

[0068] As an example, on the basis of the breakthrough randomness, combined with the amplitude of each breakthrough in the same random stage, obtain the price oscillation degree of each random stage. The specific method included is:

[0069] It should be noted that the breakthrough randomness reflects the random degree of the stock price oscillation in the random stage, that is, whether the stock price conforms to the characteristics of short-term oscillations after the breakthrough. On the basis of randomness, it is also necessary to quantify the amplitude of the stock price breaking through the upper and lower tracks. The greater the breakthrough amplitude, the greater the deviation of the short-term oscillation from the upper and lower tracks, and the greater the short-term oscillation fluctuation. On the basis of relatively large randomness, a greater degree is required for subsequent confirmation of the track to eliminate the influence of short-term oscillations on the Bollinger Bands judgment.

[0070] Specifically, for any breakthrough in the nth random stage of any cycle, calculate the absolute value of the difference between the stock price data of each breakthrough in this breakthrough and the upper or lower track of the Bollinger Bands of this cycle (calculate with the upper track if breaking through the upper track, and calculate with the lower track if breaking through the lower track) as the deviation amount of the stock price data of each breakthrough. Take the average value of all the deviation amounts in this breakthrough as the amplitude of this breakthrough. If a breakthrough only contains the stock price data of one breakthrough, then its deviation amount is the amplitude of the breakthrough.

[0071] Furthermore, take the product of the average value of the amplitudes of all breakthroughs in the nth random stage and the breakthrough randomness of the nth random stage as the price oscillation coefficient of the nth random stage; perform linear normalization on the price oscillation coefficients of all random stages in this cycle, and the obtained result is used as the price oscillation degree of each random stage.

[0072] As an example, based on the above price oscillation degree, screen several price oscillation stages from the random stages. The specific method included is:

[0073] A preset oscillation threshold. In this embodiment, the oscillation threshold is described using 0.8. For any random stage in any period, if the price oscillation degree of the random stage is greater than or equal to the oscillation threshold, the random stage is regarded as a price oscillation stage; several price oscillation stages of each period are obtained according to the above method.

[0074] So far, after dividing the stock price data into periods and obtaining Bollinger Bands respectively, the breakthroughs of the stock price data in each random stage of each period with respect to the upper and lower tracks of the Bollinger Band are quantified. The greater the breakthrough amplitude and the shorter the duration, and the more frequent the multiple breakthroughs, showing the random fluctuations of the stock price, the more in line with the characteristics of short-term oscillations. In this way, the price oscillation stages in the random stages are quantified, providing a basis for adding upper and lower confirmation tracks later to eliminate the influence of short-term oscillations.

[0075] Step S003: Obtain the upper and lower confirmation tracks of each price oscillation stage according to the price oscillation degree of the corresponding upper and lower tracks of the Bollinger Band; analyze the duration differences of each breakthrough in the same price oscillation stage to determine the stability degree of the breakthrough price of each price oscillation stage.

[0076] Preferably, in an embodiment of the present invention, obtaining the upper and lower confirmation tracks of each price oscillation stage according to the price oscillation degree of the corresponding upper and lower tracks of the Bollinger Band includes the following specific method:

[0077] It should be noted that to eliminate short-term oscillations, the upper and lower confirmation tracks of the price oscillation stage need to be amplified on the basis of the original upper and lower tracks of the Bollinger Band, that is, the upper confirmation track needs to be greater than the upper track, and the lower confirmation track needs to be less than the lower track. At the same time, since the stock trading prices are all positive, the upper and lower tracks are amplified according to the price oscillation degree. The greater the price oscillation degree corresponding to the price oscillation stage, the greater the degree of amplification of the upper and lower tracks is required to obtain the upper and lower confirmation tracks to eliminate the influence of short-term oscillations therein.

[0078] Specifically, for any price oscillation stage in any period, the sum of 1 and the price oscillation degree of the price oscillation stage is used as the upper track adjustment coefficient of the price oscillation stage; conversely, the difference obtained by subtracting the price oscillation degree of the price oscillation stage from 1 is used as the lower track adjustment coefficient of the price oscillation stage; a new curve formed by multiplying the values corresponding to each time stamp in the price oscillation stage of the upper Bollinger Band of the period by the upper track adjustment coefficient respectively is used as the upper confirmation track of the price oscillation stage; similarly, a new curve formed by multiplying the values corresponding to each time stamp in the price oscillation stage of the lower Bollinger Band of the period by the lower track adjustment coefficient respectively is used as the lower confirmation track of the price oscillation stage; then the upper and lower confirmation tracks of the price oscillation stage are obtained, and the upper and lower confirmation tracks of each price oscillation stage in each period are obtained respectively according to the above method.

[0079] Preferably, in an embodiment of the present invention, analyzing the duration differences of each breakthrough in the same price oscillation stage to determine the stability degree of the breakthrough price in each price oscillation stage, the specific method includes:

[0080] It should be noted that after adding upper and lower confirmation tracks in the price oscillation stage to judge the breakthrough of stock price data and then sending an "overbought" or "oversold" signal, it can, to a certain extent, eliminate the influence of short-term oscillations on the judgment of the upper and lower Bollinger Bands. However, for the stock price data that continuously breaks through the upper and lower tracks, it is not random. On the contrary, it shows a stable stock price fluctuation change. Therefore, under each breakthrough in the corresponding price oscillation stage, the judgment of the upper and lower confirmation tracks needs to be eliminated, and the original upper and lower tracks are used for judgment, so as to correctly send continuous "overbought" or "oversold" signals. That is, if the price oscillation degree in the price oscillation stage is large, but there is a large cumulative duration of continuous breakthroughs, the upper and lower confirmation tracks need to be eliminated to ensure the correct judgment of its "overbought" or "oversold" signal.

[0081] Specifically, a breakthrough duration threshold is preset. In this embodiment, the breakthrough duration threshold is described as 20 minutes. Then, the calculation method of the breakthrough price stability degree G of any price oscillation stage in any cycle is:

[0082]

[0083] where Q represents the number of breakthroughs in this price oscillation stage in this cycle, T m represents the duration of the m-th breakthrough in this price oscillation stage, T 0 represents the breakthrough duration threshold; norm( ) represents a linear normalization function, and the normalization object is all breakthroughs in this price oscillation stage. Obtain the breakthrough price stability degree of each price oscillation stage in each cycle according to the above method.

[0084] It should be noted that for the duration of each breakthrough in the same price oscillation stage, the more it exceeds the breakthrough duration threshold, the less it conforms to short-term oscillations, and the more stable the price change is, and the greater the breakthrough price stability degree is. If multiple breakthroughs all exceed the threshold by a large amount, and the influence of a small number of breakthroughs with short durations is small in the mean calculation, the more breakthroughs are reflected and the greater their durations are, the more necessary it is to eliminate the upper and lower confirmation tracks for the corresponding price oscillation stage, and thus the greater the breakthrough price stability degree needs to be given.

[0085] So far, after expanding the upper and lower tracks based on the degree of price oscillation in the price oscillation stage to obtain the upper and lower confirmation tracks, analyze the differences and changes in the duration of each breakthrough in the same price oscillation stage. The more breakthroughs with a longer duration make the price oscillation stage less in line with short-term oscillation. Conversely, it is necessary to eliminate the upper and lower confirmation tracks to avoid misjudging the stock price fluctuations in non-short-term oscillation.

[0086] Step S004: Based on the stability of the breakthrough price in each breakthrough during the random stage and the breakthrough price in the price oscillation stage, combined with its upper and lower confirmation tracks, judge the stock price data based on the Bollinger Bands.

[0087] It should be noted that after obtaining the upper and lower confirmation tracks for the price oscillation stage, it is necessary to determine whether to make a breakthrough judgment based on the upper and lower confirmation tracks according to the stability of the breakthrough price. For the breakthroughs of the stock price data in other random stages that are not the price oscillation stage, the judgment is based on the upper and lower tracks of the original Bollinger Bands, and then send the "overbought" or "oversold" signal reflecting the stock trading behavior.

[0088] Specifically, for any breakthrough in any random stage that is not the price oscillation stage in any cycle, if this breakthrough indicates that the stock price data exceeds the range of the Bollinger Bands, send the "overbought" or "oversold" signal; preset the breakthrough price stability threshold. In this embodiment, the breakthrough price stability threshold is described as 0.68. For any price oscillation stage in any cycle, if the stability of the breakthrough price in this price oscillation stage is greater than or equal to the breakthrough price stability threshold, eliminate the upper and lower confirmation tracks of this price oscillation stage. For each breakthrough in this price oscillation stage, the judgment is based on the corresponding upper and lower tracks, that is, each breakthrough exceeds the range of the Bollinger Bands, and send the "overbought" or "oversold" signal; if the stability of the breakthrough price in this price oscillation stage is less than the breakthrough price stability threshold, then for each breakthrough in this price oscillation stage, it is necessary to judge with the upper and lower confirmation tracks of this price oscillation stage. If it is greater than the upper confirmation track or less than the lower confirmation track, the corresponding breakthrough needs to send the "overbought" or "oversold" signal. If it exceeds the range of the upper and lower confirmation tracks, it is a short-term oscillation that needs to be eliminated and does not send the "overbought" or "oversold" signal.

[0089] So far, in the process of judging the stock trading behavior based on the Bollinger Bands of the stock price data, by analyzing the characteristics of the high-frequency and short-duration breakthroughs of the stock price through the upper and lower tracks of the Bollinger Bands in short-term oscillation, determine the price oscillation stage and expand to obtain the upper and lower confirmation tracks. At the same time, to avoid misjudging the stock price fluctuations in non-short-term oscillation, further obtain the breakthrough price stability based on the difference in the duration of the breakthroughs in the same price oscillation stage, so as to improve the accuracy of judging the stock price fluctuations by the upper and lower tracks of the Bollinger Bands, and thus correctly send the "overbought" or "oversold" signal.

[0090] Another embodiment of the present invention provides a stock trading behavior analysis system driven by big data mining. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the above method steps S001 to S004 are implemented.

[0091] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for analyzing stock trading behavior driven by big data mining, characterized in that: The method comprises the following steps: Get stock price data for a period of time; Divide the period of time to obtain several cycles and generate their Bollinger Bands; divide the cycles into several random stages, determine the price volatility of each random stage based on the number of times the stock price data in the same random stage breaks through the upper and lower tracks of the Bollinger Bands of the corresponding period, as well as the amplitude and duration of each breakthrough, and screen the price volatility stage; The upper and lower Bollinger Bands corresponding to each price oscillation stage are calculated based on the price oscillation degree to obtain the upper and lower confirmation rails of each price oscillation stage; the duration difference of each breakthrough in the same price oscillation stage is analyzed to determine the stability of the breakthrough price in each price oscillation stage; Based on each breakthrough in the random stage and the stability of the breakthrough price in the price shock stage, combined with its upper and lower confirmation tracks, the stock price data is judged based on the Bollinger Bands.

2. The method for analyzing stock trading behavior driven by big data mining according to claim 1, characterized in that: The specific method of obtaining a number of cycles and generating their Bollinger Bands includes: Dividing the period of time into a number of periods by the period length; For the stock price data in any period, the middle track of the period is obtained by the moving average method according to the time series; The standard deviation is calculated for all stock price data within the period, and the curve obtained by adding K times the standard deviation to the middle track is used as the upper track of the period. The curve obtained by subtracting K times the standard deviation from the middle track is used as the lower track of the period. The middle track, upper track and lower track of the period together constitute the Bollinger Band of the period.

3. The method for analyzing stock trading behavior driven by big data mining according to claim 1, characterized in that: The division cycle is divided into several random stages. Based on the number of times the stock price data in the same random stage breaks through the upper and lower rails of the Bollinger Bands of the corresponding period, as well as the amplitude and duration of each breakthrough, the price volatility of each random stage is determined, including the specific methods as follows: Divide each period into random stages, and obtain the number of breakthroughs of the stock price data in each random stage for the upper and lower rails of the Bollinger Bands of the corresponding period of the random stage; According to the number of breakthroughs in the same random stage and the duration of each breakthrough, the breakthrough randomness of each random stage is obtained; On the basis of breakthrough randomness, combined with the amplitude of each breakthrough in the same random stage, the degree of price volatility in each random stage is obtained.

4. The method for analyzing stock trading behavior driven by big data mining according to claim 3 is characterized in that: The specific method for obtaining the stock price data in each random stage for several breakthroughs of the upper and lower rails of the Bollinger Bands of the corresponding period of the random stage is as follows: Based on the upper and lower rails of the Bollinger Bands of any period, if any stock price data in any random phase of the period is greater than the upper rail of the Bollinger Bands of the period, or less than the lower rail of the Bollinger Bands of the period, the stock price data will be recorded as a breakthrough; If multiple consecutive stock price data break through the upper track or the lower track, the multiple consecutive stock price data will be recorded as one breakthrough, and the product of the number of broken stock price data and the update frequency duration will be recorded as the duration of the breakthrough.

5. The method for analyzing stock trading behavior driven by big data mining according to claim 4 is characterized in that: The specific method for obtaining the breakthrough randomness of each random stage is as follows: Get the average duration of all breakthroughs in any random phase in any period; Based on the ratio of the number of breakthroughs in the random stage to the average duration of all the breakthroughs, the breakthrough randomness of the random stage is obtained.

6. The method for analyzing stock trading behavior driven by big data mining according to claim 4, characterized in that: The specific method for obtaining the price fluctuation degree of each random stage is as follows: Obtain the absolute value of the difference between the stock price data of each breakthrough in any random phase in any period and the upper or lower track of the Bollinger Band of the period as the deviation of the stock price data of each breakthrough, and take the average of all the deviations in the breakthrough as the amplitude of the breakthrough; Based on the amplitude of all breakthroughs in this random stage and combined with the randomness of breakthroughs in this random stage, the degree of price volatility in this random stage is obtained.

7. The method for analyzing stock trading behavior driven by big data mining according to claim 2, characterized in that: The specific method of obtaining the upper and lower confirmation rails of each price fluctuation stage is as follows: For any price fluctuation stage in any cycle, the sum of 1 and the price fluctuation degree of the price fluctuation stage is used as the upper track adjustment coefficient of the price fluctuation stage; the difference between 1 and the price fluctuation degree of the price fluctuation stage is used as the lower track adjustment coefficient of the price fluctuation stage; The new curve formed by the product of the value corresponding to each time stamp of the upper rail of the Bollinger band of the period in the price oscillation stage and the upper rail adjustment coefficient is used as the upper confirmation rail of the price oscillation stage; A new curve formed by the product of the values ​​corresponding to each time stamp of the Bollinger band lower track of the period in the price oscillation stage and the lower track adjustment coefficient is used as the lower confirmation track of the price oscillation stage.

8. The method for analyzing stock trading behavior driven by big data mining according to claim 1, characterized in that: The specific methods for analyzing the difference in duration of each breakthrough in the same price fluctuation stage and determining the stability of the breakthrough price in each price fluctuation stage include: The difference between the duration of each breakthrough in any price oscillation stage in any period and the breakthrough duration threshold is obtained, and the stability of the breakthrough price in the price oscillation stage is obtained based on the difference.

9. The method for analyzing stock trading behavior driven by big data mining according to claim 3, characterized in that: The method of judging the stock price data based on the Bollinger Bands is as follows: Send an "overbought" or "oversold" signal for any breakthrough of any random phase in any period that is not a price oscillation phase; For any price oscillation stage in any cycle, if the breakthrough price stability degree of the price oscillation stage is greater than or equal to the breakthrough price stability threshold, the upper and lower confirmation rails of the price oscillation stage are eliminated, and each breakthrough in the price oscillation stage is judged by the corresponding upper and lower rails. Each breakthrough is beyond the range of the Bollinger Bands, and an "overbought" or "oversold" signal is sent; If the breakthrough price stability degree in the price oscillation stage is less than the breakthrough price stability threshold, for each breakthrough in the price oscillation stage, it will be judged with the upper and lower confirmation rails of the price oscillation stage. If it is greater than the upper confirmation rail or less than the lower confirmation rail, the corresponding breakthrough will send an "overbought" or "oversold" signal. If it exceeds the range of the upper and lower confirmation rails, no "overbought" or "oversold" signal will be sent.

10. A stock trading behavior analysis system driven by big data mining, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the big data mining driven stock trading behavior analysis method as described in any one of claims 1-9 are implemented.

Images