阅读说明：本技术 一种判断基金财报信息预测基金调仓方法 (Method for predicting fund transfer by judging fund financial information ) 是由 张赫远 谢国亮 于 2021-09-16 设计创作，主要内容包括：本发明提供的一种判断基金财报信息预测基金调仓方法,所述调仓方法包括：查询满足前十大持仓占比大于百分之五十的基金,获取基金的持仓数据；将相邻两个月的持仓数据进行拼接,计算不同仓位的权重,计算基金的变动数据；根据基金的持仓和权重,计算持仓的每日净值数据和估值；计算所述每日净值数据和所述估值的相似度；根据所述相似度训练基金调仓模型；根据所述基金调仓模型,预测基金调仓情况。能够通过基金和前十大持仓股票的净值判断当前持仓和上季度公布的持仓是否有很大的差异,帮助基金投资者进行投资选择。(The invention provides a fund transfer method for judging fund financial report information prediction, which comprises the following steps: inquiring the fund meeting the condition that the first ten taken positions are more than fifty percent, and acquiring the position data of the fund; splicing the position taken data of two adjacent months, calculating the weight of different positions, and calculating the change data of the fund; calculating daily net worth data and valuation of position taken according to position taken and weight of the fund; calculating a similarity of the daily net worth data and the valuations; training a fund transfer model according to the similarity; and predicting the fund transfer bin condition according to the fund transfer bin model. The net value of the fund and the top ten large position holding stocks can be used for judging whether the current position holding and the position holding published in the last quarter are different greatly, so that the fund investor is helped to make investment selection.)

1. A fund transfer method for judging fund financial report information prediction is characterized by comprising the following steps:

inquiring the fund meeting the condition that the first ten taken positions are more than fifty percent, and acquiring the position data of the fund;

splicing the position taken data of two adjacent months, calculating the weight of different positions, and calculating the change data of the fund;

calculating daily net worth data and valuation of position taken according to position taken and weight of the fund;

calculating a similarity of the daily net worth data and the valuations;

training a fund transfer model according to the similarity;

and predicting the fund transfer bin condition according to the fund transfer bin model.

2. The method for judging fund transfer according to claim 1, wherein the querying of funds meeting the condition that the top ten taken positions are greater than fifty percent, and the obtaining of position data of the funds specifically comprises:

acquiring fund annual report disclosure information;

obtaining fund F according to the fund annual report disclosure information_iThe first ten best details of position holding J in quarter k_ki(ii) a The method specifically comprises the following steps: j is a function of_ki1，j_ki2，j_ki3...j_ki10；

Obtaining the first ten position information of the public fund every quarter to obtain fund F_iThe first ten weight sequences W in k quarters_kij(ii) a The method specifically comprises the following steps: w is a_ki1，w_ki2，w_ki3...w_ki10；

Calculating fund F_iSum of top ten taken weights Sum_ki

Sum Retention_kiGreater than 50% of the fund;

term acquisition, term redemption, term end gross share, term end net assets and net asset volatility of the fund are obtained for the model's supplemental data.

3. The method for judging fund transfer prediction fund transfer according to claim 1, wherein the step of splicing the position taken data of two adjacent months and calculating the weight of different positions comprises the following steps:

according to fund F_iLast two quarters of the specification of taken position S_kijAnd a weight sequence w_kijCalculating the difference of taken position D_ki；

Wherein, D is_kiIn the middle, the funds with the largest value and the first 20% are marked as the adjusted bins, and the funds with the smallest value and the first 20% are marked as the un-adjusted bins.

4. The method for forecasting fund transfers according to the judgment fund financial report information of claim 1, wherein the calculating daily net worth data and valuations of taken positions according to the taken positions and weights of the fund specifically comprises:

obtaining the daily fluctuation range of the stock according to fund F_iThe top ten warehouse details j in k quarter_kiObtaining the stock price fluctuation range sequence C of each stock_kdjWherein d represents the number of days, there being about more than 60 trading days c per quarter_k1j，c_k2j，c_k3j…c_k60j；

Calculating the fluctuation range of the fund valuation according to the fund F_iLast two quarters of the specification of taken position S_kijAnd a weight sequence w_kijAnd the stock fluctuation range sequence C_kdjCalculating the fund F_iEstimated sequence A in k quarters_kiWherein the estimate of day d is changed to

Obtaining the estimated fluctuation range sequence A of each fund_kdjWherein d represents the number of days, there being about more than 60 trading days a per quarter_k1j，a_k2j，a_k3j…a_k60j。

5. The method of claim 1, wherein said calculating a similarity between said daily net worth data and said estimate comprises:

acquiring daily fluctuation range of the fund according to the fund F_iObtaining the net value fluctuation amplitude sequence B in k quarter_kdjWherein d represents the number of days, there being about 60 trading days per quarter

b_k1j，b_k2j，b_k3j...b_k60j

Calculating fund F_iEvaluation sequence A_kdjAnd net worth sequence B_kdjThe time window N is 10

To obtain a gold F_iCorrelation sequence R in k quarters_kiAnd calculating the correlation sequence R_kiAverage, maximum, minimum and variance of;

using estimated fluctuation sequence A_kdjSum net amplitude fluctuation sequence B_kdjCalculating to obtain a difference sequence DIFF_kdj

diff_kdj＝a_kdj-b_kdj

And calculates a difference sequence DIFF_kdjMean, maximum, minimum and variance.

6. The method for judging fund transfer prediction fund transfer bin according to the fund financial report information, according to the claim 1, wherein the training of the fund transfer bin model according to the similarity specifically comprises:

calculating a correlation sequence R_kiAverage, maximum, minimum and variance of;

calculating difference sequence DIFF_kdjAverage, maximum, minimum and variance of;

acquiring and integrating term procurement, term redemption, term end total share, term end net assets and net asset fluctuation rate of the fund;

the data are divided into a training set and a testing set, and the testing set data are trained through a python calling optimized gradient lifting decision tree model.

Technical Field

The invention relates to the field of finance, in particular to a fund transfer method for judging fund financial and newspaper information prediction.

Background

With the development of society and the progress of economy, the concept of managing wealth of residents is increasing day by day, more and more people choose to buy fund, the essence of fund is that you are replaced by professional teams to manage assets, and unlike the direct buying of stocks, professional investment knowledge is not needed, certain risks can be shared, and meanwhile, stable benefits are brought. According to data published by the national securities fund association, the management scale of public fund and the management scale of private fund keep a steady growth situation from 2015 to 2020 within two quarters. By two seasons in 2020, the management scale of public fund reaches 16.9 trillion yuan, and the management scale of private fund reaches 14.9 trillion yuan.

The fund transfers the storehouse including adding the storehouse and subtracting the storehouse, is embodied as that some stocks in the handle are sold or a part of them is sold, and the stock changed into other stocks is called transfers the storehouse, shows that the position of the storehouse has been adjusted. Every fund has its own position proportion, the stock market has rise and fall, the fund manager predicts the advance or risk of the position of the stock income, and then sells the stock to buy other stocks, which is the normal work of fund operation. Under general conditions, when a large plate enters an uplink channel, stock positions of the expanding varieties need to be enlarged slightly, and even the stock positions are full; when the large disk enters the downlink channel, the positions of all stocks need to be adjusted down, and stock varieties need to be changed in time according to market hotspots to obtain higher benefits.

The fund transfer bin belongs to non-public information and is not published externally. The position of fund is disclosed in the season newspaper and the annual newspaper, and the season newspaper is generally released in the middle and the last of the next month after the season is over and has certain hysteresis. And the adjustment of the fund can be generally judged, the fund valuation provided by a third-party platform can be observed, and if the difference between the valuation and the income of the day is overlarge, the adjustment of the fund is possibly carried out.

At present, many third-party platforms for fund valuation are provided by the fund, the third-party platforms are usually formed by detailed fitting of position taken in the fund quarterly report, a calculation formula is relatively single, meanwhile, a user needs to compare the net value and the valuation change condition by himself or herself, certain errors exist, the whole position adjusting process of the fund manager cannot be reflected through single-day comparison, and the judgment difficulty can be very high if data of multiple days are continuously observed.

Some fund position forecasting methods are also available in the market, and the general idea is to forecast the position condition of a fund through net value change of a period of time and published position taking information, but the method cannot visually represent the position adjusting condition of the fund, and position estimation is only used for reference.

Disclosure of Invention

In view of the above, the present invention has been developed to provide a method of determining fund transfer information prediction fund binning that overcomes or at least partially addresses the above-identified problems.

According to one aspect of the invention, a method for judging fund financial report information prediction fund transfer bin is provided, and the bin transferring method comprises the following steps:

inquiring the fund meeting the condition that the first ten taken positions are more than fifty percent, and acquiring the position data of the fund;

splicing the position taken data of two adjacent months, calculating the weight of different positions, and calculating the change data of the fund;

calculating daily net worth data and valuation of position taken according to position taken and weight of the fund;

calculating a similarity of the daily net worth data and the valuations;

training a fund transfer model according to the similarity;

and predicting the fund transfer bin condition according to the fund transfer bin model.

Inquiring the fund which meets the condition that the ratio of the first ten taken positions is more than fifty percent, and acquiring the position data of the fund specifically comprises the following steps: acquiring fund annual report disclosure information; obtaining fund F according to the fund annual report disclosure information_iThe first ten best details of position holding J in quarter k_ki(ii) a The method specifically comprises the following steps: j is a function of_ki1，j_ki2，j_ki3...j_ki10(ii) a Obtaining the first ten position information of the public fund every quarter to obtain fund F_iThe first ten weight sequences W in k quarters_kij(ii) a The method specifically comprises the following steps: w is a_ki1，w_ki2，w_ki3…w_ki10(ii) a Calculating fund F_iSum of top ten taken weights Sum_ki

Sum Retention_kiGreater than 50% of the fund; term acquisition, term redemption, term end gross share, term end net assets and net asset volatility of the fund are obtained for the model's supplemental data.

Splicing the taken position data of two adjacent months, calculating the weights of different positions, wherein the step of calculating the change data of the fund specifically comprises the following steps: according toFund F_iLast two quarters of the specification of taken position S_kijAnd a weight sequence w_kijCalculating the difference of taken position D_ki；

Wherein, D is_kiIn the middle, the funds with the largest value and the first 20% are marked as the adjusted bins, and the funds with the smallest value and the first 20% are marked as the un-adjusted bins.

Calculating daily net worth data and estimates of positions taken according to positions and weights of the fund specifically comprises: obtaining the daily fluctuation range of the stock according to fund F_iThe top ten warehouse details j in k quarter_kiObtaining the stock price fluctuation range sequence C of each stock_kdjWherein d represents the number of days, there being about more than 60 trading days c per quarter_k1j，c_k2j，c_k3j…c_k60j(ii) a Calculating the fluctuation range of the fund valuation according to the fund F_iLast two quarters of the specification of taken position S_kijAnd a weight sequence w_kijAnd the stock fluctuation range sequence C_kdjCalculating the fund F_iEstimated sequence A in k quarters_kiWherein the estimate of day d is changed to

Obtaining the estimated fluctuation range sequence A of each fund_kdjWherein d represents the number of days, there being about more than 60 trading days a per quarter_k1j，a_k2j，a_k3j…a_k60j。

Calculating the similarity between the daily net worth data and the estimate specifically includes:

acquiring daily fluctuation range of the fund according to the fund F_iObtaining the net value fluctuation amplitude sequence B in k quarter_kdjWherein d represents the number of days, there being about 60 trading days per quarter

b_k1j，b_k2j，b_k3j...b_k60j

Calculating fund F_iEvaluation sequence A_kdjAnd net worth sequence B_kdjThe time window N is 10

To obtain a gold F_iCorrelation sequence R in k quarters_ki，And calculating the correlation sequence R_kiAverage, maximum, minimum and variance of;

using estimated fluctuation sequence A_kdjSum net amplitude fluctuation sequence B_kdjCalculating to obtain a difference sequence DIFF_kdj

diff_kdj＝a_kdj-b_kdj

And calculates a difference sequence DIFF_kdjMean, maximum, minimum and variance.

Training the fund transfer bin model according to the similarity specifically comprises the following steps: calculating a correlation sequence R_kiAverage, maximum, minimum and variance of; calculating difference sequence DIFF_kdjAverage, maximum, minimum and variance of; acquiring and integrating term procurement, term redemption, term end total share, term end net assets and net asset fluctuation rate of the fund; the data are divided into a training set and a testing set, and the testing set data are trained through a python calling optimized gradient lifting decision tree model.

The invention provides a fund transfer method for judging fund financial report information prediction, which comprises the following steps: inquiring the fund meeting the condition that the first ten taken positions are more than fifty percent, and acquiring the position data of the fund; splicing the position taken data of two adjacent months, calculating the weight of different positions, and calculating the change data of the fund; calculating daily net worth data and valuation of position taken according to position taken and weight of the fund; calculating a similarity of the daily net worth data and the valuations; training a fund transfer model according to the similarity; and predicting the fund transfer bin condition according to the fund transfer bin model. The net value of the fund and the top ten large position holding stocks can be used for judging whether the current position holding and the position holding published in the last quarter are different greatly, so that the fund investor is helped to make investment selection.

The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a flowchart of a method for determining fund transfer prediction based on fund financial information according to an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

The terms "comprises" and "comprising," and any variations thereof, in the present description and claims and drawings are intended to cover a non-exclusive inclusion, such as a list of steps or elements.

The technical solution of the present invention is further described in detail with reference to the accompanying drawings and embodiments.

As shown in fig. 1, a method for determining fund financial report information prediction fund transfer includes:

inquiring the fund meeting the condition that the first ten taken positions are more than fifty percent, and acquiring the position data of the fund;

splicing the position taken data of two adjacent months, calculating the weight of different positions, and calculating the change data of the fund;

calculating daily net worth data and valuation of position taken according to position taken and weight of the fund;

calculating a similarity of the daily net worth data and the valuations;

training a fund transfer model according to the similarity;

predicting the condition of fund transfer according to the fund transfer model;

the fund data is first processed into the specified format of model input, including the correlation sequence R_kiOf the mean, maximum, minimum and variance, difference sequence DIFF_kdjMean, maximum, minimum and variance of (c), term of the last quarter of the fund, term redemption (parts in million), term end total share (parts in million), term end net assets (parts in million), net asset rate of change is shown in the table below.

Calling the model trained in the step 8 to obtain whether the fund transfers the label of the bin, and if the output is 0, the fund does not transfer the bin; and if the output is 1, adjusting the fund bin.

Inquiring the fund which meets the condition that the ratio of the first ten taken positions is more than fifty percent, and acquiring the position data of the fund specifically comprises the following steps:

acquiring fund annual report disclosure information;

obtaining fund F according to the fund annual report disclosure information_iThe first ten best details of position holding J in quarter k_ki(ii) a The method specifically comprises the following steps: j is a function of_ki1，j_ki2，j_ki3...j_ki10；

Obtaining the first ten position information of the public fund every quarter to obtain fund F_iThe first ten weight sequences W in k quarters_kij(ii) a The method specifically comprises the following steps: w is a_ki1，w_ki2，w_ki3…w_ki10；

Calculating fund F_iSum of top ten taken weights Sum_ki

Sum Retention_kiGreater than 50% of the fund;

term acquisition, term redemption, term end gross share, term end net assets and net asset volatility of the fund are obtained for the model's supplemental data.

Splicing the taken position data of two adjacent months, calculating the weights of different positions, wherein the step of calculating the change data of the fund specifically comprises the following steps:

according to fund F_iLast two quarters of the specification of taken position S_kijAnd a weight sequence w_kijCalculating the difference of taken position D_ki；

Wherein, D is_kiIn the middle, the funds with the largest value and the first 20% are marked as the adjusted bins, and the funds with the smallest value and the first 20% are marked as the un-adjusted bins.

Calculating daily net worth data and estimates of positions taken according to positions and weights of the fund specifically comprises:

obtaining the daily fluctuation range of the stock according to fund F_iThe top ten warehouse details j in k quarter_ki，Obtaining the stock price fluctuation range sequence C of each stock_kdjWherein d represents the number of days, there being about more than 60 trading days c per quarter_k1j，c_k2j，c_k3j…c_k60j；

Calculating the fluctuation range of the fund valuation according to the fund F_iLast two quarters of the specification of taken position S_kijAnd a weight sequence w_kijAnd the stock fluctuation range sequence C_kdjCalculating the fund F_iEstimated sequence A in k quarters_kiWherein the estimate of day d is changed to

Obtaining the estimated fluctuation range sequence A of each fund_kdjWherein d represents the number of days, there being about more than 60 trading days a per quarter_k1j，a_k2j，a_k3j…a_k60j。

Calculating the similarity between the daily net worth data and the estimate specifically includes:

acquiring daily fluctuation range of the fund according to the fund F_iObtaining the net value fluctuation amplitude sequence B in k quarter_kdjWherein d represents the number of days, there being about 60 trading days per quarter

b_k1j，b_k2j，b_k3j...b_k60j

Calculating fund F_iEvaluation sequence A_kdjAnd net worth sequence B_kdjThe time window N is 10

To obtain a gold F_iCorrelation sequence R in k quarters_ki，And calculating the correlation sequence R_kiAverage, maximum, minimum and variance of;

using estimated fluctuation sequence A_kdjSum net amplitude fluctuation sequence B_kdjCalculating to obtain a difference sequence DIFF_kdj

diff_kdj＝a_kdj-b_kdj

And calculates a difference sequence DIFF_kdjMean, maximum, minimum and variance.

Training the fund transfer bin model according to the similarity specifically comprises the following steps:

calculating a correlation sequence R_kiAverage, maximum, minimum and variance of;

calculating difference sequence DIFF_kdjAverage, maximum, minimum and variance of;

acquiring and integrating term procurement, term redemption, term end total share, term end net assets and net asset fluctuation rate of the fund;

the data are divided into a training set and a testing set, and the testing set data are trained through a python calling optimized gradient lifting decision tree model.

After the model is trained, the effect of the model needs to be evaluated, and evaluation indexes currently adopted in the industry include accuracy (Precision), Recall (Recall), F-value (F-Measure), and the like.

First, the confusion matrix needs to be calculated

TP is the number of positive classes predicted from the positive class; TN is the prediction of negative class as negative class number; FP is the number of positive classes for which negative classes are predicted; FN is the prediction of a positive class as a negative class number.

Precision calculation formula

Recall ratio recall calculation formula

Calculation formula of comprehensive evaluation index (F1-score)

By training the training set, the effect obtained on the test set is as follows

	precision	recall	F1
				0	0.84	0.80	0.82
1	0.80	0.84	0.82

。

Has the advantages that: judging whether the current position is different from the position published in the previous quarter or not according to the net value of the fund and the top ten position holding stocks, and giving corresponding prompts to help fund investors to make investment choices; without the need for the investor to have strong expertise.

The above embodiments are provided to further explain the objects, technical solutions and advantages of the present invention in detail, it should be understood that the above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the scope of the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.