Predicting stock prices for Chinese performing arts companies using genetic algorithm-based backpropagation neural networks

Abstract

Due to China’s thriving economy and culture, the performing arts sector has grown remarkably. To study its development, this study has examined the closing prices of performing arts companies. The GA-BPN model was used to analyze the daily closing prices of Funshine Culture (ticker: 300860) and Sanxiang Impression (ticker: 000863) for the predictions of their future daily closing prices. Next, the study compared the predicted prices with the actual closing prices. By comparing four models, namely GA, 7-4-1, 7-4-4-1, and 7-4-4-4-1, the GA-BPN model has a mean square error (MSE) of 2472.580273 and a root mean square error (RMSE) of 49.72504674, which is the smallest value and the smallest error among the four assessment metrics, it was determined that the GA-BPN model yielded the most accurate prediction results, so it was suitable for forecasting stock closing prices.

Keywords

GA-BPN model performing arts companies business strategies closing prices

1. Introduction

The performing arts industry is an important part of global economic and cultural development, and it plays a positive role in economic development. The development of performing arts can create a large number of employment opportunities, such as, cultural and creative enterprises, artists, art managers, etc.; performing arts enterprises can also promote the development of tourism, art as a form of culture can attract tourists from all over the world to come to appreciate and experience, and promote the development of tourism, but also promote the cultural interactions of various places; performing arts industry also brings great economic growth potential. The creation, production, exhibition and sale of art works involve the participation of many related industries and drive the prosperity of related industries. As the performing arts industry tends to rely more on artists, creators and cultural and creative industries than on traditional manufacturing or conventional services; at the same time, the market for this industry is highly influenced by audience preferences, trends, policies and cultural factors, and is therefore subject to a certain degree of uncertainty. Therefore, this paper examines stocks in the performing arts industry.

The stock market is a key component of the market economy. The stock market is favored by numerous investors as it will reward them with high returns in exchange for additional risks they take in holding stocks. By the end of 2022, there were a total of 4,917 listed companies on the Shanghai Stock Exchange and the Shenzhen Stock Exchange. The thriving stock market greatly facilitates the development of China’s market economy. Under China’s comprehensive deepening of reforms, the financial sector is undergoing significant changes. China’s stock market has experienced remarkable growth since the founding of the Shenzhen Stock Exchange and the Shanghai Stock Exchange in the 1990s, but it also faces many new challenges. Volatile stock prices, for instance, not only adversely impact the national economy. The stock market crash may harm investor’s well-being, cause psychological stress, and affect their health. For an investor with a low risk tolerance who favors investments that maintain his or her original investment, abnormal stock prices will harm their interests. Effective investment theories and methods are required to help investors grasp the patterns of the stock market and enhance their investment returns. This is beneficial to both the stock market and the financial reform.

As investing theories keep upgrading, they are often tested in real life. More and more quantitative finance practitioners try to apply them to the stock market. They use a variety of approaches, including statistical and machine learning to predict long-term or short-term stock prices [1]. Wen Dechun; Zhao Tianlong; Fang Lexin; Zhang Caiming; Li Xuemei used a multi-level wavelet decomposition interaction network (MWDINet) to design a depth difference module (DIF Block) for correcting autocorrelation errors [2]. The article considers the correlation between the error of the current time step and the error of the previous time step, but does not use specific examples for wind analysis. Chauhan Akshat; S.J. Shivaprakash; H. Sabireen; Md. Abdul Quadir; Venkataraman Neelanarayanan explored Recurrent Neural Networks (RNNs) and will also investigate the application of metaheuristic algorithms to improve the results of prediction models [3]. This article improved the accuracy of the model, but failed to make predictions, and this model is too complex. Although sometimes these methods can make successful predictions, most methods’ performances are less than our expectations. This is because the stock market is a very complicated system, which is influenced by numerous factors that interact with each another. We need a large database to predict the stock market, and powerful algorithms should underpin the prediction process.

During the past decade, artificial neural network have been extensively used in many business applications, such as pattern recognition, intelligent robotics, automatic control, forecasting, biomedicine, and economics. They have successfully addressed many challenging problems that cannot be solved by traditional methods. They belong to artificial intelligence that teaches computers to process data in a way that is inspired by the human brain. The advantages of artificial neural networks include: first, they have learning abilities. For instance, in image recognition, they can gradually learn to identify similar images by analyzing example images that have been manually labeled. Successive adjustments will cause the neural network to produce output that is increasingly similar to the target output. This learning abilities are particularly useful for predictions. Artificial neural network is a potential tool to be applied in economic forecasts, market predictions, effectiveness predictions. Secondly, they have the function of associative memory. For example, an artificial neural network will store the set of patterns as memories when the associative memory is being presented with a key pattern, it responds by producing one of the stored pattern which closely resembles or relates to the key pattern. Thirdly, they can rapidly search for the optimal solution. Finding an optimal solution for a complex problem often requires significant computational power. The recurrent networks are designed for special problems. They can fully leverage the ability of computers to process data and perform complex calculations at high speeds, and quickly identify the optimal solutions. Considering the advantages and capabilities of artificial neural networks, it is of great significance to use it in stock market predictions. As a result of the New Crown Pneumonia outbreak, events, venues, markets and arts access have been closed and performances have been canceled on a large scale, leading to a drop in the stock prices of performing arts companies, which in turn affects the market capitalization of the companies and the confidence and sentiment of investors, which in turn reduces their investment in the companies. In order to recover the performing arts market, we will forecast the stock prices of performing arts companies, thereby managing risks in advance and adjusting business strategies in time for better operations. This article uses an optimized genetic evolutionary neural network (GA-BPN) to predict the closing price of stocks, which is more accurate than general artificial neural network predictions, and this model is more innovative. Can help entertainment companies predict future stock prices and develop sound business strategies to achieve good operations.

This paper will use ANN to analyze common stocks from the cultural and creative industries. Its main subjects are Funshine Culture and Sanxiang Impression. It has studied the daily historical closing stock prices of the two companies from 2018 to 2022. The accuracy of the artificial neural network can be evaluated by making predictions of the two companies’ daily closing prices and comparing them with the actual ones. In this way, we will know whether ANN is able to forecast the stock market in the long run.

2. Literature review

The stock market in China has developed for many years, during which many scholars, from home and abroad, have discussed various stock prediction methods. Among the methods, most popular ones are fundamental analysis and technical analysis [1, 2, 3, 4]. However, these two methods are insufficient to predict a much more complicated and volatile stock market. More precise and stable methods, such as neural networks, are needed for stock price forecasting and research.

In 2019, Chu Wenhua [5] established a three-layer BP neural network and analyzed its convergence speed. The study concluded that when selecting reasonable and high-quality data, the fitting accuracy would improve, and thus BP neural networks was a feasible tool for stock price prediction. During the same year, Sun Boyuan [6] conducted theoretical research on LSTM neural networks. After organizing and analyzing the data, Sun reviewed previous neural network prediction systems and optimized the stock prediction model by combining the LSTM algorithm with linear algebra. As many network models may have difficulties in feature extraction of stocks, their accuracy remains low in stock price prediction. To address these issues, Qiao Ruoyu [7] compared many algorithm-based models in 2019 to find the best ones. The related deep learning algorithms included the Multilayer Perceptron (MLP), the Convolutional Neural Network (CNN), the Recurrent Neural Network (RNN), the Long Short-Term Memory (LSTM), and the Gated Recurrent Unit (GRU). His comparative experiments, which were based on data of the Shanghai Composite Index, successfully determined the impacts of different important variables. Qiao’s research offered a direction for the establishment of stock prediction models in the future. In 2021, Jiang Bolin et al. [8] introduced the LSTM model to avoid various errors in the recurrent neural network. The new model, which was based on recurrent neural networks, could effectively addressed problems like data forgetting and gradient explosion. They also applied its computational analysis capability to predict stocks. In 2021, Zhang Rumeng et al. [9] investigated the functions of BP neural networks and ARMA-GARCH models in predicting the closing prices of SAIC Motor Corporation Limited. According to their comparative analysis, BP neural networks demonstrated a higher accuracy in long-term predictions, while ARMA-GARCH models had a slight advantage in short-term predictions. In 2022, Yan Dongmei et al. [10] addressed the nonlinear and non-stationary characteristics of stock prices and proposed a Generative Adversarial Network model, called SAR-GAN, which combined self-attention mechanisms with residual networks. To validate the model’s excellent generalization ability, they selected leading stocks in the market to test the prediction abilities of the model. The results showed that the SAR-GAN model had less errors than LSTM, GRU, CNN-LSTM, and CNN-GRU models. In 2023, Huang Mingxing et al. [11] presented a prediction model which adopted PSO-BP neural networks to forecast the closing prices of Ping An Bank. When the number of hidden layer nodes was set to be 7, the PSO-BP neural network model achieved the best prediction performance, with RMSE to be 0.026615 and MAE to be 0.0371. Therefore, Huang concluded that the PSO-BP neural network model was reliable and can be used to forecast Ping An Bank’s closing prices. Shile Chen and Changjun Zhou (2021) This paper use Genetic Algorithm (GA) for feature selection and use the combination of optimal factors and LSTM model for stock prediction [12]. Omer Berat Sezera, c, Murat Ozbayoglua1, Erdogan Dogdub (2017) propose a stock trading system based on optimized technical analysis parameters for creating buy-sell points using genetic algorithms. Then passed to a deep MLP neural network for buy-sell-hold predictions [13]. Parshv Chhajer, Manan Shah, Ameya Kshirsagar (2022) discuss the strengths and weaknesses of machine learning for stock market prediction and provide some insight into the opportunities and threats in applying advanced technologies for stock market prediction. and further study the applications of three machine learning technologies in the stock market prediction, including artificial neural networks, support vector machines, and long-short term memory [14]. M. Qiu, Yu Song (2016). The main contribution of this study is the ability to predict the direction of the next day’s price of the Japanese stock market index by using an optimized artificial neural network (ANN) model. To improve the prediction accuracy of the trend of the stock market index in the future, we optimize the ANN model using genetic algorithms (GA). We demonstrate and verify the predictability of stock price direction by using the hybrid GA-ANN model and then compare the performance with prior studies [15]. Kim [16] used genetic algorithm to optimize neural network model for stock price prediction and found that the model optimized after genetic algorithm was more accurate in predicting the stock price. Vasiani VD and Handari [17] used preferred index method and genetic algorithm to optimize the stock and the results showed that the average return was higher with the use of genetic algorithm than without it. Persi [18] and others used Recurrent Neural Networks (RNN), Long and Short Term Memory Neural Networks (LSTM) and Gated Recurrent Neural Networks (GRU) to predict Google’s stock history data respectively and the results showed that Long and Short Term Memory Neural Networks were more accurate in their predictive ability.

Although many scholars have studied the stock predictive effects of neural networks, few of them focused on artificial neural networks (ANN), especially in the past three years. This study endeavors to investigate the performance of artificial neural networks in stock price prediction, focusing on ANN model’s prediction accuracy and stability. In addition, the previous research focuses more on some traditional industries, but this paper chooses to carry out a forecasting research on the stock price of performing arts companies, which is innovative in terms of the theme, and we use the optimized GA-BPN model to make predictions, and the predicted values made have a high degree of fit with the actual values.

3. Indicator analysis

Based on the understanding of stock closing price and the organization of its indicators, this paper will select the following six indicators to analyze it.

3.1 Selection of indicators and their definitions

Media are the communication tools used to store and deliver information or data. Being a global phenomenon, it involves political and economic factors, as well as technological innovations. The growing market economy and social advance drive the cultural industry to be more diverse, refined and differentiated. To facilitate the faster and better development of China’s cultural and media industry and support the formulation of investment strategies, According to Chien-Jen Huang, Chien-Jen Huang, Peng-Wen Chen and Wen-Tsao Pan [17] this article, this paper selected six commonly used technical analysis indicators for comprehensive evaluation. The indicators include: KDJ stochastic indicator, relative strength index (RSI), moving average convergence divergence (MACD), BIAS deviation ratio, WR Williams %R indicator, and the closing price. These indicators describe the stock price changes from different dimensions. They will help us better understand price movements and make corresponding investment strategies.

KDJ stochastic indicator is a technical analysis indicator calculated based on the highest price, lowest price, and closing price. It is used to analyze and predict changes in stock trends and price patterns in a traded asset for the identification of best buying or selling opportunities.

The RSI (relative strength index) measures the ratio of up-moves to down-moves, and normalizes the calculation so that the index is expressed in a range of 0–100. If the RSI is 70 or greater, the instrument is assumed to be overbought (a situation whereby prices have risen more than market expectations). An RSI of 30 or less is taken as a signal that the instrument may be oversold (a situation in which prices have fallen more than the market expectations).

The MACD (moving average convergence/divergence) is a statistical analysis that turns two trend-following indicators, moving averages, into a momentum oscillator by subtracting the longer moving average from the shorter one. The MACD indicator is designed to help investors identify price trends and measure the trend momentum.

The BIAS (deviation rate) is a technical analysis indicator used to measure the deviation degree between the stock price and the moving average line in the process of fluctuation. It provides a better understanding of the price trends.

The WR (Williams%R) indicator is a technical analysis tool that shows the current closing price in relation to the highest price and the lowest price, It helps investors determine the degree of price fluctuation.

KDJ stochastic indicator is calculated based on the highest price, lowest price and closing price within a certain period of time, it can reflect the overbought and oversold status of the price within a certain period of time, and has a certain correlation with the trend of the stock price. RSI indicator is a relative strength indicator, by comparing the rate of increase and decrease within a period of time, it can judge the overbought and oversold situation of the stock market, and has a high correlation with the short-term trend of the stock price. MACD The indicator is plotted by calculating the difference between two moving averages and the difference between their moving averages. It can reflect the rapid upward and downward trend of stock prices and has a high correlation with the medium and long-term trend of stock prices.

Deviation BIAS is used to assess the overbought and oversold condition of a stock by calculating the difference between the current closing price and its moving average, which has a high correlation with the short-term trend of the stock price. WR Williams%R indicator is used to assess the degree of overbought and oversold condition of a stock by calculating the ratio between the high and low prices and the closing price, which has a high correlation with the short-term trend of the stock price. These indicators help us understand the price fluctuation of the listed cultural companies, so that better investment strategies can be made.

3.2 Selection of sample data for performing arts companies

Based on the growth trajectory of the performing arts sector, this study has selected data of Funshine Culture Group Co., Ltd. (“Funshine Culture”, Stock Code: 300860) and Sanxiang Impression Co., Ltd. (“Sanxiang Impression”, Stock Code: 000863) to be sample data, as can be shown in the Table 1.

Table 1
Introduction of performing arts companies

Enterprise abbreviation	Founded in	Location
Funshine Culture	2002	Beijing
Sanxiang Impression	1996	Shanghai

3.3 Data sources and descriptive statistics

The experimental data of this paper is the SSE index from October 17, 2019 to December 30, 2022, with a total of 567 valid data from wind database. The data is real and effective, this experiment uses the first 500 data as the training set and the remaining 67 data as the test set. The model was inputted six indicators to predict the stock’s closing price of the next day and the indicators included KDJ stochastic indicator, relative strength index (RSI), moving average convergence/divergence (MACD), the deviation rate BIAS, WR Williams %R indicator, and the closing price. The GA-BPN model used in this paper can better help acting companies to predict the future trend of the company’s closing price. Enhance its operating ability.

The analysis of the descriptive statistics of Funshine culture is presented in Table 2 and the descriptive statistics of Sanshou impressions are analyzed in Table 3.

Table 2
Descriptive statistics of Funshine Culture

Variables	Sample size	Max.		Min.		Mean		Standard deviation
KDJ (K)	567	89	.85	1	.39	38	.714	22.383
KDJ (D)	567	84	.088	3	.077	38	.597	19.479
Closing price	567	168	.17	36	.01	73	.392	40.668
BIAS	567	57	.482	$-$ 47	.922	$-$ 0	.97	9.355
KDJ (J)	567	126	.153	$-$ 17	.455	38	.948	35.173
WR	567	100		0		61	.932	28.65
MACD	567	10	.68	$-$ 20	.68	$-$ 1	.792	5.065
RSI	567	95	.152	7	.333	44	.753	18.703

Table 3

Descriptive statistics of Sanxiang Impression

Variables	Sample size	Max.		Min.		Mean		Standard deviation
KDJ (K)	567	93	.908	6	.037	46	.008	22.558
KDJ (D)	567	87	.036	8	.287	46	.103	19.879
Closing price	567	117	.255	$-$ 15	.502	45	.818	34.461
BIAS	567	5	.26	2	.68	4	.033	0.62
KDJ (J)	567	100		0		54	.761	28.349
WR	567	33	.938	$-$ 19	.799	$-$ 0	.254	4.869
MACD	567	0	.318	$-$ 0	.245	$-$ 0	.015	0.105
RSI	567	92	.189	7	.119	46	.722	19.235

This study analyzes several indicators of stock prices of performing arts companies in terms of mean and standard deviation, including KDJ Stochastic, Relative Strength Index (RSI), Moving Average Convergence/Divergence (MACD), Deviation BIAS, WR Williams%R indicator, and closing price.

By analyzing the data of Frontier Culture, we found that the closing price has a mean value of 73.392 and a standard deviation of 40.668, which shows a large fluctuation. While the mean value of the J index is 38.948 with a standard deviation of 35.173, indicating a high degree of variability. For the MACD indicator, the mean value is $-$ 1.792 and the standard deviation is 5.065, indicating the volatility of the market condition.

In the data analysis of Sanshou Impression, the mean value of the closing price is 45.818 with a standard deviation of 34.461, which is relatively small volatility. the mean value of the J-index is 54.761 with a standard deviation of 28.349, which indicates a higher level of volatility in the market. For the MACD indicator, the mean value is $-$ 0.015 with a standard deviation of 0.105, indicating relatively stable market conditions.

In summary, the closing price is the main factor that affects the stock price of performing arts companies with high volatility. This is followed by the influence of the J-index, which also has a high degree of variability. In contrast, the MACD indicator has less influence and the market state changes are relatively stable. By studying and analyzing these indicators, the Performing Arts Company can better understand the market situation and thus formulate a decision-making basis for adjusting its operating strategies.

4. Empirical analysis

We make the following analysis based on the model results.

The Artificial Neural Networks (ANN) is comprised of three components: the input layer (the first layer), the hidden layers (the intermediate layers between input and output layer), and the output layer (the final layer). Each layer can have multiple neurons. The number of neurons in the input layer is equal to the number of features in the data. Usually, people use one hidden layer for simple tasks, but nowadays research in deep neural network architectures show that many hidden layers. Neurons in consecutive layers are densely connected but neurons within the same layer do not interact with each other. Figure 1 shows a neural network that is capable of self-learning through forward propagation and back propagation.

Figure 1.

Structure of neural network.

First, we collected K-line indicators (K), D-line indicators (D), J-line indicators (J), RSI, MACD, BIAS, and WR Williams indicators of Funshine Culture and Sanxiang Impression during 2020–2022 to be independent variables (X). Second, we collected the two companies’ closing prices as the dependent variable (Y). The data was organized to associate seven indicators’ current values with the closing price of the next period. The first 500 pieces of data for each company were selected and divided into 5 groups, each group containing 100 pieces. After the division, 4 groups were used for training, and 1 group for testing. This cross-validation was repeated for multiple times to ensure more scientific and accurate testing results. To improve the accuracy of stock closing price predictions, this study adopted both the conventional neural network model and the genetic evolution-based neural network model to train and test the data.

4.1 Building normal artificial neural networks (ANN)

The number of neurons in the hidden layer of an artificial neural network is often determined by taking the sum of the neurons in the input and output layers and dividing it by two. Since the exact number of hidden layers cannot be determined, this study tried various configurations, including 1, 2, and 3. As there were seven types of independent variable X, we build 7-4-1, 7-4-4-1, and 7-4-4-4-1 as the general neural network models.

4.2 Building a genetic evolution-based neural network (GA-BPN)

The Genetic Evolution-based Neural Network (GA-BPN) uses genetic algorithm to optimize the initial weights and thresholds of the BPN neural network. After the optimization, the BPN neural network will generate more accurate predictions. The genetic algorithm parameters are set as follows: the iteration count was set to 1,000, the population size was set to 50, the crossover rate was set to 0.2, and the mutation rate was set to 0.04. By setting different parameters, the optimal network structure was found. The optimal network architecture for the Funshine Culture GA model is shown in Table 4, The optimal network architecture for the Sanxiang Impression GA model is shown in Table 5. The optimal structure was represented in green, which would be selected in this study to build the genetic evolution-based neural network, allowing for more accurate predictions.

Table 4
Optimal network architecture of Funshine Culture GA model

Group I

Group II

Group III

Group IV

Group V

7, 8, 6, 7, 1

7, 8, 1

7, 10, 1

7, 8, 8, 7, 9, 1

7, 7, 6, 8, 9, 1

7, 8, 5, 5, 7, 1

7, 8, 8, 1

7, 8, 1

7, 8, 8, 1

7, 6, 5, 8, 9, 1

7, 7, 6, 5, 8, 1

7, 8, 7, 1

7, 10, 1

7, 8, 8, 1

7, 7, 4, 8, 9, 1

7, 6, 6, 5, 8, 1

7, 7, 1

7, 8, 10, 7, 1

7, 9, 10, 1

7, 7, 4, 7, 8, 1

7, 6, 5, 6, 1

7, 7, 8, 1

7, 10, 1

7, 1

7, 7, 4, 7, 8, 1

7, 6, 4, 9, 4, 1

7, 8, 8, 1

7, 9, 1

7, 6, 8, 1

7, 7, 4, 7, 8, 1

Table 5

Optimal network architecture of Sanxiang Impression GA model

Group I

Group II

Group III

Group IV

Group V

7, 8, 10, 8, 9, 1

7, 1

7, 6, 19, 1

7, 8, 8, 1

7, 10, 1

7, 7, 8, 9, 8, 1

7, 7, 8, 8, 1

7, 5, 19, 1

7, 8, 9, 1

7, 10, 1

7, 7, 11, 8, 1

7, 7, 1

7, 5, 1

7, 7, 8, 8, 1

7, 1

7, 7, 8, 8, 1

7, 8, 1

7, 6, 5, 1

7, 6, 8, 8, 1

7, 1

7, 8, 10, 8, 1

7, 8, 8, 1

7, 1

7, 8.8.1

7, 1

7, 7, 1

7, 6, 5, 1

7, 10, 9, 1

7, 6, 1

This study adopted four evaluating metrics to compare the predictive performance of the four models: 7-4-1, 7-4-4-1, 7-4-4-4-1, and Genetic Evolution-Based Neural Network models. MSE is the mean of the squared difference between the predicted and true values. RMSE is the square root of the ratio of the sum of the squares of the deviations of the predicted values from the true values. MAPE is the mean of the absolute percentage erro. MAD is the average of the absolute values of the deviations of all individual observations from the arithmetic mean.

The formulas for these evaluating metrics are as follows:

Mean Square Error (MSE)

$\displaystyle\textit{MSE}=\frac{\sum\limits_{i=1}^{n}{(y_{i}-\hat{y}_{i})^{2}}% }{n}$ (1)

Root Mean Square Error (RMSE)

$\displaystyle\textit{RMSE}=\sqrt{\frac{\sum\limits_{i=1}^{n}{(y_{i}-\hat{y}_{i% })^{2}}}{n}}$ (2)

Mean Absolute Percentage Error (MAPE)

$\displaystyle\textit{MAPE}=\frac{100\%}{{n}}\sum\limits_{i=1}^{n}{\left|{\frac% {y_{i}-\hat{y}_{i}}{y_{i}}}\right|}$ (3)

Mean Absolute Deviation (MAD)

$\displaystyle\textit{MAD}=\frac{\sum\limits_{i=1}^{n}{|{{y}_{i}-\hat{y}_{i}}|}% }{n}$ (4)

Where $n$ denotes the number of samples, $y_{i}$ denotes the true value and $\hat{y}_{i}$ denotes the predicted value.

The smaller the value of the four evaluating metrics, the smaller the error will be, and the higher predictive accuracy the model will achieve.

Table 6

Evaluation results of 5 indicators of Funshine Culture

Company	Model	MSE	RMSE	MAPE	MAD
Funshine Culture	GA	2472.580273	49.72504674	0.301650855	32.57613719
	7-4-1	2938.2638	54.2057543	0.519589651	40.94366726
	7-4-4-1	2820.172617	53.10529745	0.703013616	45.69056758
	7-4-4-4-1	2778.35142	52.71006944	0.570564413	42.41701546

According to Table 6, Calculations based on Eqs (1)–(4), among the four models based on the data of Funshine Culture, the Genetic Evolution-Based Neural Network (GA) model yielded the lowest values. The stock of Funshine Culture had an MSE of 2472.580273, RMSE of 49.72504674, and MAPE of 0.301650855. Thus, the GA model outperformed the other three models.

Table 7

Evaluation results of 5 indicators of Sanxiang Impression

Company	Model	MSE	RMSE	MAPE	MAD
Sanxiang Impression	GA	0.209972413	0.458227469	0.085145388	0.37399272
	7-4-1	0.316485168	0.562570145	0.098565427	0.440116968
	7-4-4-1	0.267077824	0.516795728	0.092590307	0.410059624
	7-4-4-4-1	0.277449875	0.526735109	0.09635457	0.426097958

According to Table 7, Calculations based on Eqs (1)–(4), among the four models based on the data of Sanxiang Impression, the Genetic Evolution-Based Neural Network (GA) model yielded the lowest values. The stock of Sanxiang Impression had an MSE of 0.262170526, RMSE of 0.562570145, and MAPE of 0.093501921. The GA model outperformed the other models again. The analysis shows that GA model is most suitable for predicting the stock closing prices of the companies.

4.3 Genetic evolution-based neural network (GA-BPN) model’s predictions

The GA model was used to predict the closing prices of the two companies. According to the results in Table 8. The predicted closing price for Funshine Culture on January 3, 2023 was 42.605765. The predicted closing price for Sanxiang Impression on February 16, 2022 was 3.040785. The line charts show the actual closing stock prices and the predicted closing stock prices of the companies. It can be observed that the predicted curve aligns closely with the actual curve.

Table 8
Prediction results of GA-BPN model

Company	Date	Actual value	Predicted value	Error
Funshine Culture	January 3, 2023	52.16	42.605765	9.554235
Sanxiang Impression	February 16, 2022	3.58	3.040785	0.539215

4.4 Reasons for the fluctuations of the closing prices

Figure 2.

Closing price chart of Funshine Culture.

Figure 3.

Closing price chart of Sanxiang Impression.

After observing Figs 2 and 3, we discover that both stocks are highly volatile within a certain period, but overall they maintain a relatively stable closing price with a decline and a rebound. The stock prices are based on the supply and demand on the market. In essence, they represent the companies’ earnings potential and operational results. In the cultural industry, where Funshine Culture and Sanxiang Impression belong, a company’s performance is also influenced national policies, the overall economy, and the market environment. It is also affected by the artistic value of the cultural products.

Since the 21st century, globalization has blurred the borders of different economies, and competitions among countries have intensified. To race ahead, many countries have shifted their economic development models to build the capacity of domestic culture. China called on its artists and writers to boost their awareness of and confidence in Chinese culture, create new prospects for Chinese literature and art, and add new luster to Chinese culture. To pave the way for stronger and more inclusive creative industries, the government provided special funds and preferential tax policies for cultural companies. The government also builds a “National Project Library for Cultural Industry Development”, in which a number of leading, fundamental, and strategic projects are stored as typical cases, to lead the cultural industry and point out future directions for companies. It has also optimized the industrial structure for the healthy, sound, and systematic development of the cultural industry. Thanks to the efforts, total the output value of China’s media industry reached RMB 2.52 trillion in 2020, with a 6.51% year-on-year (YoY) increase. The total operating revenue of China’s cultural industry reached RMB 11.9 trillion in 2021, which was an 16.0% YoY increase, and its average growth rate from 2020 to 2021 was over 8.9%. Under the strong support of the government and macro policies, the cultural industry has expanded rapidly and advance in a more systematic way. This lays a solid foundation for the excellent performance of cultural companies on the stock market. Specifically, there are a number of factors that can affect the stock price of a performing arts company, such as: Cultural policies, government cultural policies may encourage or restrict the development of the industry, which can affect the company’s stock price; Consumer behavior, changes in consumer behavior can directly affect the industry; International relations, changes in international relations can also have an impact on stock prices. Changes in trade policies, political instability or increased international tensions may lead to increased uncertainty in the market and stock prices may be negatively affected.

If the cultural industry, especially the sector of performing arts, wants to further develop, it should follow the principle of “the content is king”. Motivated by excellent cultural policies, the performing arts practitioners have created many high-quality artistic works in recent years. Cultural companies innovated their products, such as red operas, dance dramas, stage plays, TV dramas, and films, on the basis of Chinese history. The themes of the products focused on the 40th anniversary of reform and opening-up, the 70th anniversary of the founding of the People’s Republic of China, the completion of the building of a moderately prosperous society in all respects, and the 100th anniversary of the founding of the Communist Party of China, as well as important historical events and figures. Moreover, they have made “creative transformation and innovative development” of cultural heritages. Inspired by Chinese traditional culture and history, the cultural professionals have produced a large number of popular productions with a huge investment of resources, such as The Eternal Wave, The Journey of a Legendary Landscape Painting, and Li Bai. Such high-quality productions are growing and there is a fantastic array of new works entering the market each year. This indicates that China’s cultural industry is in its golden age. As the quality of artistic works improves, they can appeal to a wider audience. This will further bolster the diversity and vitality of the performing arts industry and provides a nurturing and creative environment where people can access quality arts. With solid economic foundation, market foundation, and cultural foundation, stocks related to the cultural industry have not fluctuated much in recent years. Their closing prices appear to be relatively stable with a mild upward trend.

However, the arts industry, as well as the artistic experience, has been devastated by COVID-19 pandemic (2020 $\sim$ 2021) as events, venues, markets and access to art were shut down in order to protect people from the virus. According to the sample survey from the China Association of Performing Arts, nearly 20,000 events in China were canceled or postponed from January to March 2020, resulting in a direct box office loss of over RMB 2 billion. The mass cancellation of performances indirectly led to stock price declines of companies from the cultural industry. The easing of Covid control restrictions at the second half of 2022 reinvigorated the performing arts sector. Governments at all levels introduced a set of policies to help cultural enterprises resume work and production. Through fiscal subsidies, tax incentives, and financial support, the government offered targeted assistance to the cultural industry. Cultural enterprises revolutionized their operations and structures through cultural and creative crossovers to other industries and turned challenges into opportunities. In this way, online and offline activities could have a synchronized development. Under the joint efforts from the government, the industry, and the practitioners, the performing arts activities returned towards pre-COVID levels, and the cultural industry market began to recover and so did the stock prices of related companies. Therefore, in recent years, the stock of cultural companies experienced a slight decline before rebounding. The overall stock price was relatively stable, with minor fluctuations due to the macroeconomic consequences of the COVID-19 pandemic or others. This article selects the GA-BPN model to predict the closing price of performance company stocks. The company can use this model to predict the future closing price of stocks and observe the trend. This can control risks in advance, adjust business strategies in a timely manner, and achieve better operations.

5. Conclusions

By comparing the four models GA, 7-4-1, 7-4-4-1 and 7-4-4-4-1, it was determined that the GA-BPN model is the most effective model for predicting the closing price of stocks. In the future, using the GA-BPN model to predict the closing prices of performing arts companies and observe their movements can help performing arts companies manage risks in advance and adjust their operation strategies in time. The GA-BPN model combines the advantages of genetic algorithms and back-propagation neural networks, which can more accurately predict future closing price movements.

With this model, performing arts companies can collect historical data and train it to build a prediction model. The model analyzes multiple factors such as historical market data, company financials, and industry dynamics to predict future closing prices. Performing arts companies can understand market trends based on these predictions in order to make timely and effective strategic adjustments.

By utilizing the GA-BPN model for forecasting and trend observation, performing arts companies can better avoid risks and seize market opportunities. It is a powerful decision support tool for performing arts companies to help them adjust their operation strategies in advance to avoid the decline of revenue or the expansion of loss. In conclusion, closing price prediction using the GA-BPN model can provide performing arts companies with important information to help them better manage operational risks.

In this paper, only two performing arts company stocks are studied, the generalization of the research results needs to be strengthened, and the closing price of the stock will be affected by a variety of factors, and its trend is difficult to predict, so it can be strengthened by the study of external economic factors, focusing on the macro-economic environment, industry development trends, policy impacts, and other factors, in order to improve the accuracy of the prediction, in order to obtain more accurate results.

The fluctuation of stock closing prices is affected by a variety of factors, such as macroeconomic factors, industry trends, company policies, social stability, and so on. This makes it difficult to predict the accuracy of stock prices. Therefore, we need to continuously improve the accuracy of the model to obtain more realistic and precise prediction results. In the application of stock closing price prediction, GA-BPN model does have great advantages and potentials, so the application of GA-BPN model in stock closing price prediction has an important role in promoting the research and development of the securities market, which helps to improve the growth and development of China’s social economy. Therefore, further related research work is needed in the future to promote the application and development of GA-BPN model in stock market forecasting and risk management.

References

Huang

Mercado

. Artificial intelligence recruitment text automatic generation based on light detection and improved neural network algorithm. Optical and Quantum Electronics. 2024; 56(2): 5770.

Wen

Zhao

Fang

Zhang

. MWDINet: A multilevel wavelet decomposition interaction network for stock price prediction. Expert Systems with Applications. 2024; 238(E): 2019.

Chauhan

Shivaprakash

Sabireen

Quadir

Venkataraman

. Stock price forecasting using PSO hypertuned neural nets and ensembling. Applied Soft Computing. 2023; 147(C): 147.

Nedbal

Stieninger

. Success factor analysis for cloud services: A comparative study on software as a service. International Journal of Grid and Utility Computing. 2020; 11(3): 315-329.

Tajmazinani

Hassani

Raei

Rouhani

. Modeling stock price movements prediction based on news sentiment analysis and deep learning. Annals of Financial Economics. 2022; 17(1): 38.

Hoxha

Hasani

. Decision-making biases in property investments in Prishtina, Kosovo. Journal of Property Investment & Finance. 2023; 41(2): 15581.

Gite

Khatavkar

Kotecha

Srivastava

Maheshwari

Pandey

. Explainable stock prices prediction from financial news articles using sentiment analysis. PeerJ Computer Science. 2021; 1(28): 340.

Khennour

Bouchachia

Kherfi

Bouanane

. Randomising the simple recurrent network: A lightweight, energy-efficient RNN model with application to forecasting problems. Neural Computing & Applications. 2023; 7(14): 58.

Kamalov

. Forecasting significant stock price changes using neural networks. Neural Computing & Applications. 2020; 32(23): 1765567.

10.

Zhao

Zeng

Liang

Kang

Liu

. Prediction model for stock price trend based on recurrent neural network. Journal of Ambient Intelligence and Humanized Computing. 2021; 12(1): 74553.

11.

Liu

Soto Breceda

Kuhlmann

Karoly

Grayden

Zhao

Cook

Schmidt

. Brain model state space reconstruction using an LSTM neural network. Journal of Neural Engineering. 2023; 20(3): 781.

12.

Cao

Parry

. Neural network earnings per share forecasting models: A comparison of backward propagation and the genetic algorithm. Decision Support Systems. 2009; 47(1): 3241.

13.

Asgarian

Ghasemi

Momtazi

. Generative adversarial network for sentiment-based stock prediction. Concurrency and Computation-Practice & Experience. 2023; 35(2): e7467.

14.

Liu

Guo

Wang

Zhang

. Prediction of stock market index based on ISSA-BP neural network. Expert Systems with Applications. 2022; 204(15): 117604.

15.

Kirscher

. Steel location Kazakhstan – current situation and outlook. Stahl Und Eisen. 2000; 120(9): 10712.

16.

Chen

Zhou

. Stock prediction based on genetic algorithm feature selection and long short-term memory neural network. IEEE Access. 2021; 9(9): 906672.

17.

Sezer

Ozbayoglu

Dogdu

. A deep neural-network based stock trading system based on evolutionary optimized technical analysis parameters. Procedia Computer Science. 2017; 114(11): 47380.

18.

Cao

. Financial time series forecasting model based on CEEMDAN and LSTM. Physica A-Statistical Mechanics and Its Applications. 2019; 12739.

19.

Qiu

Song

. Predicting the direction of stock market index movement using an optimized artificial neural network model. PLOS One. 2016; 11(5): e0155133.

20.

Huang

Chen

Pan

. Using multi-stage data mining technique to build forecast model for Taiwan stocks. Neural Comput & Applic. 2012; 21(8): 205763.

Predicting stock prices for Chinese performing arts companies using genetic algorithm-based backpropagation neural networks

Abstract

Keywords

1. Introduction

2. Literature review

3. Indicator analysis

3.1 Selection of indicators and their definitions

3.2 Selection of sample data for performing arts companies

Table 1 Introduction of performing arts companies

Table 2 Descriptive statistics of Funshine Culture

4.2 Building a genetic evolution-based neural network (GA-BPN)

Table 4 Optimal network architecture of Funshine Culture GA model

Table 8 Prediction results of GA-BPN model

References

Table 1
Introduction of performing arts companies

Table 2
Descriptive statistics of Funshine Culture

Table 4
Optimal network architecture of Funshine Culture GA model

Table 8
Prediction results of GA-BPN model