CN106355503A - Method for analyzing stock trend and selecting stocks based on self-correlation sequence - Google Patents
Method for analyzing stock trend and selecting stocks based on self-correlation sequence Download PDFInfo
- Publication number
- CN106355503A CN106355503A CN201611055402.7A CN201611055402A CN106355503A CN 106355503 A CN106355503 A CN 106355503A CN 201611055402 A CN201611055402 A CN 201611055402A CN 106355503 A CN106355503 A CN 106355503A
- Authority
- CN
- China
- Prior art keywords
- stock
- trend
- correlation
- auto
- stocks
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method for analyzing stock trend and selecting stocks based on a self-correlation sequence. The main thought of the method is as follows: looking for the most similar trend of each stock in self-historical trend line on the basis of the trend in the recent period, thereby obtaining the optimal matching point and matching value; after performing similar operations on all the stocks, sequencing and comparing all the matched values, thereby obtaining a strong self-correlation stock list; combining with the future trend of the matching point corresponding to the stocks on the basis of the list, thereby forming a candidate trend pool of the stocks; lastly, analyzing stock trend and selecting stocks on the basis of the strong self-correlation stock. The method is a stock-selecting method in prior accuracy. In a stock selecting process, the accuracy and the rising amplitude need be considered properly.
Description
Technical field
The present invention relates to stock certificate data digging technology field, especially relate to a kind of stock trend based on auto-correlation sequence
Analysis and share-selecting method.
Background technology
The prediction of stock price receives much concern always, and the high return of stock market has also promoted sending out of Stock Price Forecasting
Exhibition, is engraved in during investor and is concerned about stock market, analysis stock market, attempts to predict the development trend of stock market.
Stock market is a complicated nonlinear system, and market is subject to from each side such as political, social, economic, psychological
The impact in face, thus modeling is difficult to its motor behavior.But, as technical Analysis assumed " market is that have trend can follow
's；The market price reflects all；History often repeat itself ", although this also just illustrates complexity, market still implies some
Regular.
The historical track form of share price has important predictive value to future price trend particularly short-term trend.Tradition
Morphological analyses theory is often more mechanical, and all stocks apply mechanically some identical shape mode, or in identification share price sequence shape
State often has certain difficulty, the phenomenon varying with each individual usually, and variety of problems all have impact on the effect of investment.
Conventional technical Analysis method is except traditional k lineation opinion, Wave Theory, Moving Average theory and technology index
Analysis etc., the also method of data mining and artificial intelligence, such as time series analysis, multivariate regression models, artificial neural network,
Genetic algorithm etc..
Content of the invention
The invention discloses a kind of stock trend analysiss based on auto-correlation sequence and share-selecting method.The main thought of method
It is that every stock finds most like tendency with the tendency of recent a period of time in the historical trend line of itself, optimal to obtain
Match point and matching value；After carrying out similar operation to all stocks, all of matching value is ranked up comparing, obtains strong
Auto-correlation stock list；It is based on this list afterwards, in conjunction with tendency after the corresponding match point of stock, form candidate's trend of stock
Pond；It is finally based on strong auto-correlation stock to carry out trend analysiss and select stocks.Method is a kind of preferential share-selecting method of accuracy rate, selects stocks
When need carry out between accuracy and amount of increase one compromise.
Every stock has the characteristic of stock of its own, carries out mating and find similar tendency section, institute in the historical data of itself
The characteristic of stock of acquisition trend is also consistent with this stock, and the tendency after the stock auto-correlation date can be very good to be somebody's turn to do as investigating
One kind reference of stock later stage tendency.
The inventive method is a kind of preferential stock selection of accuracy and trend analysis, can be user's short operation
Select stocks offer decision support.
Hypothesis stock list is s, s=[s1, s2,…,si,…,sn], n is the quantity of stock in stock pond, such as in China
The quantity of city's stock or the quantity of listed stock of the U.S..The step of the inventive method is as follows:
(1) load stock certificate data and carry out pretreatment；
(2) data sectional is carried out to stock to be predicted；
(3) Mobile state Time alignment of matching section range format being gone forward side by side mates；
(4) all stocks in stock pond are carried out with similar process, obtains optimal match point and the matching value of every stock；
(5) obtain strong auto-correlation stock list, and pre-loaded structure candidate's trend pond；
(6) comprehensive analysis are carried out to stock and select stocks.
Wherein, the loading stock certificate data of step (1) carry out pretreatment, particularly as follows: for every stock sm,m=1,…,n
, using certain time point as starting point, such as 2005-01-01, intercept from this time point to come data now.Make initial time
The closing price of point on the basis of 1, enter line translation according to amount of increase and amount of decrease and obtain by the closing price of its subsequent point in time, the ups and downs of such as second day
Width is change2, then the benchmark closing price of second day be: 1* (1+change2/100);The amount of increase and amount of decrease of the 3rd day is change3,
Then the benchmark closing price of the 3rd day is: 1* (1+change2/100) (1+change3/100);By that analogy.Eventually form one
Benchmark closing price array myclose.
Wherein, step (2) data sectional is carried out to stock to be predicted, particularly as follows: by benchmark closing price array myclose
It is divided into two sections, one section is as-is data to be predicted, another section is historical data to be matched.As-is data to be predicted is array
The benchmark closing price data of the nearly l day of trade in myclose, forms an array, is designated as a, a=[a1,a2,…,ai,…al]；
Historical data to be matched is the part after the data removing the nearly l day of trade, forms another array, is designated as b, b=[b1,
b2,…,bj,…bk].Record b simultaneouslyjTrade date, be designated as another array c, c=[c1,c2,…,cj,…ck].
Wherein, the Mobile state Time alignment coupling that matching section range format is gone forward side by side of step (3), particularly as follows: to array a
Mated with b, the length of a is l, the length of b is k, k=l.Matching process first carries out [0-1] interval formatting to a,
I.e. ai= (ai- min (a))/(max (a)-min (a)), and i=1 ... l, after conversion, result is designated as a '；Then with step-length for 1, circulation from
Obtain the data of length l in b, and be also carried out [0-1] interval formatting, every section of result is designated as b '；Then a ' and b ' is with dynamic
The method of Time alignment (dtw) is mated, and the window size of coupling is set to 2, obtains a matching value；Finally each is walked
The long matching value obtaining is compared, and obtains smallest match and is worth the corresponding coupling date, that is, corresponding to last element of b '
Trade date, as itself history point correlation time.
Wherein, the best match all stocks in stock pond being carried out with similar process, obtaining every stock of step (4)
Point and matching value, particularly as follows: all other stock in stock pond is carried out with the similar process of step (1) ~ (3), ultimately form
Best match two-dimensional array g, g=[gi],i=1…n;Wherein, gi= [siStock code, the auto-correlation date, matching value], represent
Every stock optimal match point under itself historical data environment, i.e. auto-correlation time point.
Wherein, the acquisition strong auto-correlation stock list of step (5), and pre-loaded structure candidate's trend pond, particularly as follows: right
Matching value matchvalue in two-dimensional array g sorts from small to large, obtains the front n stock of minimum, these stocks after drained sequence
Ticket constitutes strong auto-correlation list h.Based on strong auto-correlation stock list h, for each record h in hj,
hj=[ sjStock code, the auto-correlation date, matching value], j=[1 ... n]
Obtain stock code and auto-correlation date, then inquire about data base, the tendency of 30 days after the direct access auto-correlation date
Data, including daily amount of increase and amount of decrease, opening price, closing price, highest price, lowest price and exchange hand, and these data outputs is k
Line chart chart.All these tendency chart forms candidate's trend pond.
Wherein, comprehensive analysis being carried out to stock and select stocks, particularly as follows: sequentially examining from small to large by matching value of step (6)
Examine strong auto-correlation stock list, immediately show or the later stage tendency after the corresponding auto-correlation date is provided, provide hand-off simultaneously
Rate, exchange hand, the basic quotation information such as circulation is for reference, and user is according to the possible ups and downs of the stock later stage tendency checked
Situation and the position Synthesis in strong auto-correlation stock list consider to select stocks.
Brief description
The flow chart of the stock trend analysiss that Fig. 1 is the present invention to be sorted based on auto-correlation and share-selecting method.
Fig. 2 is the strong auto-correlation stock tabulating result based on the inventive method output.It is specially on November 1st, 2016 base
Strong auto-correlation stock list in the inventive method output.Based on this list, stock contains ocean science and technology taking first record as a example
(603703) the auto-correlation date is on June 5th, 2015 that is to say, that by current trading day on November 1st, 2016 and auto-correlation
June 5 2015 date carries out registration, and before two times, the tendency of 20 ~ 30 days of trade is more close, and on June 5th, 2015
Tendency afterwards can be used as the prediction of following following day of trade on November 1st, 2016.
Specific embodiment
Below in conjunction with the accompanying drawings and example, the present invention is described in detail.
Every stock has the characteristic of stock of its own, carries out mating and find similar tendency section, institute in the historical data of itself
The characteristic of stock of acquisition trend is also consistent with this stock, and the tendency after the stock auto-correlation date can be very good to be somebody's turn to do as investigating
One kind reference of stock later stage tendency.
The inventive method is a kind of preferential stock selection of accuracy and trend analysis, can be user's short operation
Select stocks offer decision support.
Hypothesis stock list is s, s=[s1, s2,…,si,…,sn], n is the quantity of stock in stock pond, such as in China
The quantity of city's stock or the quantity of listed stock of the U.S..The comprising the following steps that of the inventive method.
First, load stock certificate data and carry out pretreatment.
For every stock sm, m=1 ..., n, using certain time point as starting point, such as 2005-01-01, intercept from this when
Between put to come data now., on the basis of 1, the closing price of its subsequent point in time is according to rising for the closing price making start time point
Drop range is entered line translation and is obtained, and the amount of increase and amount of decrease of such as second day is change2, then the benchmark closing price of second day be: 1* (1+
change2/100);The amount of increase and amount of decrease of the 3rd day is change3, then the benchmark closing price of the 3rd day be: 1* (1+change2/100)
(1+change3/100);By that analogy.Eventually form benchmark closing price array myclose.
2nd, data sectional is carried out to stock to be predicted.
The new benchmark data that above-mentioned steps are obtained is divided into two sections, and one section is as-is data to be predicted, and another section is to treat
Join historical data.
The acquisition process of 2.1 as-is datas to be predicted is: obtains the benchmark closing quotation of the nearly l day of trade in array myclose
Valence mumber evidence, forms an array, is designated as a, a=[a1,a2,…,ai,…al], wherein, aiThe benchmark representing the day of trade of nearly l-i is received
Disk valency.a0Represent the benchmark closing price of current trading day.The value of l should not be too little, and generally 20 or 30.
The acquisition process of 2.2 historical datas to be matched is: to the remaining data part in myclose, that is, removes nearly l
Part after the data of the day of trade, as historical data to be matched, forms another array, is designated as b, b=[b1,b2,…,bj,…
bk], wherein, bjRepresent the benchmark closing price of the j position corresponding day of trade.Record b simultaneouslyjTrade date, be designated as another array c,
c=[c1,c2,…,cj,…ck], wherein, cjRepresent the corresponding trade date in j position of stock.
3rd, Mobile state Time alignment of matching section range format being gone forward side by side mates.
This step mainly completes the matching process between as-is data to be predicted and historical data to be matched, and obtains itself
Point correlation time of history.Matching process launches in a and b, and the length of a is l, and the length of b is k, k >=l, specifically mated
Journey is as follows.
3.1 setting matching value matchvalue=are just infinite, match time point matchdate=starting time point；
3.2 couples of a carry out [0-1] interval formatting, ai= (ai- min (a))/(max (a)-min (a)), and i=1 ... l, conversion
Result is designated as a ' afterwards；
3.3 with step-length for 1, and circulation obtains the data of length l from b, is designated as btemp；
3.4 couples of btempCarry out [0,1] interval formatting, result is designated as b '；
3.5 a ' are mated using dynamic time consolidation (dtw) method with b ', and window size is set to 2.The matching value obtaining
It is designated as tempvalue；
Tempvalue is compared by 3.6 with matchvalue, if tempvalue < matchvalue,
matchvalue = tempvalue；Matchdate is set to b simultaneouslytempThe coupling date corresponding to last element；
3.7 execute 3.3 ~ 3.6 steps repeatedly, and the coupling date matchdate finally obtaining is history point correlation time；
3.8 record stock smAutocorrelation result gm, gm=[ smStock code, the auto-correlation date, matching value], wherein, gmFor
One-dimension array, the auto-correlation date goes up the matchdate in step, and matching value is matchvalue, and the auto-correlation date is as
The coupling date corresponding to little matching value.
4th, all stocks in stock pond are carried out with similar process, obtains optimal match point and the matching value of every stock.
All other stock in stock pond is carried out with the similar process of step (1) ~ (3), ultimately forms best match two
Dimension group g, g=[gi],i=1…n;Wherein, gi= [siStock code, the auto-correlation date, matching value], represent every stock and exist
Optimal match point under itself historical data environment, i.e. auto-correlation time point.
5th, strong auto-correlation stock list pre-loaded structure candidate's trend pond are obtained.
5.1 auto-correlations refer to that the currently recent tendency of stock and the historical time sequence fragment of itself have preferable
Degree of joining, that is, tendency form is more similar.Matching value matchvalue in two-dimensional array g is sorted from small to large, drained
The front n stock of minimum is obtained, these stocks constitute strong auto-correlation list h after sequence.
5.2 obtain the strong later stage tendency mated after the auto-correlation date of every stock in stock list, and sectional drawing output.
Particularly as follows: being based on strong auto-correlation stock list h, for each record h in hj,
hj=[ sjStock code, the auto-correlation date, matching value], j=[1 ... n],
Obtain stock code and auto-correlation date, then inquire about data base, the tendency of 30 days after the direct access auto-correlation date
Data, including daily amount of increase and amount of decrease, opening price, closing price, highest price, lowest price and exchange hand, and these data outputs is k
Line chart chart.All these tendency chart forms candidate's trend pond.
6th, comprehensive analysis are carried out to stock and select stocks.
The comprehensive analysis of this step and process of selecting stocks are completed by artificial participation.Due to the stock in strong auto-correlation stock list
Matching degree, very well that is to say, that the tendency after the auto-correlation date all has higher accuracy, specifically carries out comprehensive analysis simultaneously
The process selected stocks is: sequentially investigates strong auto-correlation stock list from small to large by matching value, immediately shows or provide accordingly certainly
Later stage tendency after relevant date, provides turnover rate, exchange hand, the basic quotation information such as circulation is for reference simultaneously,
The position Synthesis according to the possible ups and downs situation of the stock later stage tendency checked and in strong auto-correlation stock list for the user consider
Select stocks.Position in strong auto-correlation stock list be actually a kind of embodiment of accuracy it is therefore desirable to later stage amount of increase with
Carry out a compromise between accuracy, select the stock of high-quality.
In sum, the invention discloses a kind of stock trend analysiss based on auto-correlation sequence and share-selecting method.Method
The historical trend of itself stock is mated, is obtained smallest match value and the maximally related time point of itself history；Exist afterwards
In all stocks, matching value is ranked up comparing, obtains correlation highest, be i.e. the later stage tendency of accuracy good similar section,
Carry out comprehensive analysis based on strong auto-correlation list and its later stage tendency to select stocks.Method is a kind of preferential share-selecting method of accuracy rate,
Need when selecting stocks to carry out a compromise between accuracy and amount of increase.
The inventive method is similarly applied to the data that security class has time serieses feature, such as fund, futures etc..Cause
This, although disclosing the specific embodiments and the drawings of the present invention for the purpose of illustration, its object is to help understand that the present invention's is interior
Hold and implement according to this, but it will be appreciated by those skilled in the art that: in the essence without departing from the present invention and appended claim
In god and scope, various replacements, to change and modifications be all impossible.Therefore, the present invention should not be limited to most preferred embodiment and
Accompanying drawing disclosure of that.Presently disclosed embodiment should be understood illustrative in all respects rather than it be claimed
Scope restriction.
Claims (4)
1. a kind of stock trend analysiss based on auto-correlation sequence with share-selecting method it is characterised in that methods described includes walking as follows
Rapid:
(1) load stock certificate data and carry out pretreatment；
(2) data sectional is carried out to stock to be predicted；
(3) Mobile state Time alignment of matching section range format being gone forward side by side mates；
(4) all stocks in stock pond are carried out with similar process, obtains optimal match point and the matching value of every stock；
(5) obtain strong auto-correlation stock list, and pre-loaded structure candidate's trend pond；
(6) comprehensive analysis are carried out to stock and select stocks.
2. the stock trend analysiss based on auto-correlation sequence according to claim 1 with share-selecting method it is characterised in that examining
Consider characteristic of stock and the feature of stock itself, every stock carries out the search coupling of similartrend only in the historical data of itself.
3. the stock trend analysiss based on auto-correlation sequence according to claim 1 with share-selecting method it is characterised in that entering
The data of row coupling have passed through the pretreatment in several stages, includes successively having obtained the amount of increase and amount of decrease data since certain time point, carries out
Benchmark closing price is changed, is carried out [0,1] range format, and these processes make result be not only coupling trend, also mates concrete
Numerical value.
4. the stock trend analysiss based on auto-correlation sequence according to claim 1 and share-selecting method it is characterised in that
During the selecting stocks of stock, it is on the premise of accuracy is preferential, sequentially investigates strong auto-correlation stock after the auto-correlation date
Tendency, carry out, in conjunction with after other factors comprehensive analysis, operation of selecting stocks.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611055402.7A CN106355503A (en) | 2016-11-26 | 2016-11-26 | Method for analyzing stock trend and selecting stocks based on self-correlation sequence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611055402.7A CN106355503A (en) | 2016-11-26 | 2016-11-26 | Method for analyzing stock trend and selecting stocks based on self-correlation sequence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106355503A true CN106355503A (en) | 2017-01-25 |
Family
ID=57862583
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611055402.7A Pending CN106355503A (en) | 2016-11-26 | 2016-11-26 | Method for analyzing stock trend and selecting stocks based on self-correlation sequence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106355503A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI661380B (en) * | 2017-12-22 | 2019-06-01 | 精誠資訊股份有限公司 | Analytical method and system that use the historical trajectory of the three-day K-line chart to predict the probability of the next day's rise and fall |
-
2016
- 2016-11-26 CN CN201611055402.7A patent/CN106355503A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI661380B (en) * | 2017-12-22 | 2019-06-01 | 精誠資訊股份有限公司 | Analytical method and system that use the historical trajectory of the three-day K-line chart to predict the probability of the next day's rise and fall |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shi et al. | Intergenerational succession and internationalisation strategy of family SMEs: Evidence from China | |
| JP6831453B2 (en) | Signal search device, method, and program | |
| CN114757432B (en) | Future execution activity and time prediction method and system based on flow log and multi-task learning | |
| CN113779260B (en) | Pre-training model-based domain map entity and relationship joint extraction method and system | |
| Goumatianos et al. | Stock selection system: building long/short portfolios using intraday patterns | |
| CN106355503A (en) | Method for analyzing stock trend and selecting stocks based on self-correlation sequence | |
| Almqvist | A comparative study between algorithms for time series forecasting on customer prediction: An investigation into the performance of ARIMA, RNN, LSTM, TCN and HMM | |
| Mishra et al. | Effect of big data analytics on improvement of corporate social/green performance | |
| Oldeweme et al. | The rhythm of teamwork: Discovering a complex temporal pattern of team processes. | |
| CN103942278B (en) | Method for conducting friend recommendation through analysis of user active friends making will | |
| CN113887471B (en) | Video time sequence positioning method based on feature decoupling and cross comparison | |
| CN115564997A (en) | Pathological section scanning and analyzing integrated method and system based on reinforcement learning | |
| CN106355500A (en) | Stock prediction method based on positive and negative related trend matching | |
| EP3739517A1 (en) | Image processing | |
| Rajabalinejad et al. | Determination of stakeholders' consensus over values of system of systems | |
| da Silva et al. | Automated Machine Learning for Time Series Prediction | |
| Fischer et al. | Data-Driven Organizations: Review, Conceptual Framework, and Empirical Illustration | |
| CN116011428B (en) | Method, device, medium and equipment for extracting, judging and predicting civil case information | |
| Frigiola | Supervised Contrastive Learning for Classification of Market Stock Series | |
| Zhang | Revolutionizing Investment Strategies: Optimizing Portfolios Through Large-Scale Language Models and Innovative Leasing Structures | |
| Liu et al. | NoxTrader: LSTM-Based Stock Return Momentum Prediction | |
| Assandri et al. | Deep Double Descent for Time Series Forecasting: Avoiding Undertrained Models | |
| Kien et al. | Applied rough set to processing data in the advisor support system | |
| Magruk | The most important stages of innovative management of the future in the foresight approach | |
| Kocher et al. | Predicting Student Smoking Likelihood: A Machine Learning Regression Analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170125 |