Résultats de recherche pour « Stock Price Forecasting »

With the world’s rapid economic growth and the expansion of stock market, it produced a large amount of valuable data information. That data become an important investors in stock investment analysis subject Thorough analysis the short-term stock price forecast problem and comparing a variety of stock price forecasting method, on the basis of BP neural network (BPNN) [1] and principal component analysis (PCA）[2] and genetic algorithm and the feasibility of short-term prediction of stock price .BP neural network can use the study of historical stock market data, find out the inherent law of development and change of the stock market, so as to realize the future stock price data changes over a period of time.

With the world’s rapid economic growth and the expansion of stock market, it produced a large amount of valuable data information. That data become an important investors in stock investment analysis subject Thorough analysis the short-term stock price forecast problem and comparing a variety of stock price forecasting method, on the basis of BP neural network (BPNN) [1] and principal component analysis (PCA）[2] and genetic algorithm and the feasibility of short-term prediction of stock price .BP neural network can use the study of historical stock market data, find out the inherent law of development and change of the stock market, so as to realize the future stock price data changes over a period of time.

ABSTRACT A long‐standing puzzle to financial economists is the difficulty of outperforming the benchmark random walk model in out‐of‐sample contests. Using data from the USA over the period of 1872–2007, this paper re‐examines the out‐of‐sample predictability of real stock prices based on price–dividend (PD) ratios. The current research focuses on the significance of the time‐varying mean and nonlinear dynamics of PD ratios in the empirical analysis. Empirical results support the proposed nonlinear model of the PD ratio and the stationarity of the trend‐adjusted PD ratio. Furthermore, this paper rejects the non‐predictability hypothesis of stock prices statistically based on in‐ and out‐of‐sample tests and economically based on the criteria of expected real return per unit of risk. Copyright © 2011 John Wiley & Sons, Ltd.

Prediction of future movement of stock prices has been the subject matter of many research work. On one hand, we have proponents of the Efficient Market Hypothesis who claim that stock prices cannot be predicted accurately. On the other hand, there are propositions that have shown that, if appropriately modelled, stock prices can be predicted fairly accurately. The latter have focused on choice of variables, appropriate functional forms and techniques of forecasting. This work proposes a granular approach to stock price prediction by combining statistical and machine learning methods with some concepts that have been advanced in the literature on technical analysis. The objective of our work is to take 5 minute daily data on stock prices from the National Stock Exchange (NSE) in India and develop a forecasting framework for stock prices. Our contention is that such a granular approach can model the inherent dynamics and can be fine-tuned for immediate forecasting. Six different techniques including three regression-based approaches and three classification-based approaches are applied to model and predict stock price movement of two stocks listed in NSE - Tata Steel and Hero Moto. Extensive results have been provided on the performance of these forecasting techniques for both the stocks.

Prediction of future movement of stock prices has been the subject matter of many research work. On one hand, we have proponents of the Efficient Market Hypothesis who claim that stock prices cannot be predicted accurately. On the other hand, there are propositions that have shown that, if appropriately modelled, stock prices can be predicted fairly accurately. The latter have focused on choice of variables, appropriate functional forms and techniques of forecasting. This work proposes a granular approach to stock price prediction by combining statistical and machine learning methods with some concepts that have been advanced in the literature on technical analysis. The objective of our work is to take 5 minute daily data on stock prices from the National Stock Exchange (NSE) in India and develop a forecasting framework for stock prices. Our contention is that such a granular approach can model the inherent dynamics and can be fine-tuned for immediate forecasting. Six different techniques including three regression-based approaches and three classification-based approaches are applied to model and predict stock price movement of two stocks listed in NSE - Tata Steel and Hero Moto. Extensive results have been provided on the performance of these forecasting techniques for both the stocks.

Abstract Recent studies have improved stock price forecasting with the emerging deep learning models. Despite advancements in deep learning, stock price prediction faces significant challenges. Existing studies predominantly focus on forecasting future prices, with limited attention to nowcasting, which predicts current or near-future market states. Additionally, most methods use univariate data, neglecting the valuable interactions between multiple financial variables. This study addresses these challenges by evaluating both forecasting and nowcasting approaches using deep learning. We incorporate multivariate inputs, including opening price, high, low, close, volume, inday-change and trend, to enhance the predictive power of our models. We implement and compare several deep learning methods: Long Short-Term Memory (LSTM) networks, Gated Recurrent Units (GRU), Convolutional Neural Networks combined with LSTM (CNN-LSTM), and the transformer-based Patch Time Series Transformer (PatchTST). Our experimental results reveal that the standard LSTM model achieves superior performance compared to the more recent PatchTST and CNN-LSTM models. Specifically, models perform better in nowcasting scenarios, likely due to smaller price fluctuations over shorter periods. Furthermore, our analysis shows that including variables such as opening prices, highest prices, and lowest prices enhances predictive accuracy, whereas trading volume tends to reduce performance. These findings suggest that deep learning models are more effective for real-time or near-term stock price prediction and highlight the importance of multivariate inputs in developing robust prediction models. This study provides valuable insights for enhancing the accuracy and reliability of stock price forecasts, with significant implications for financial analysts and investors.

We categorize stock price patterns by their five consecutive extrema following rules outlined in literature. We smooth stock prices in a fixed window and locate extrema from the maxima and minima in the smooth prices. We then plot a digital image using 200 pixel values of each pattern and turn the image into grey scale and then black and white to locate the pixel values. We determine each unique pixel value representing the extrema. We repeat the process to determine that the pixel no values categorize the patterns the same as the price time series. Then we train a machine learning classifier on each labeled pattern example represented by their consecutive extrema. We find a high degree of accuracy on both in sample and out of sample classification data. The method chosen is gradient boosted trees. We then remove the extrema one at a time and re classify using approximately 50% of the data for training on all five extrema and 50% for testing on removed extrema. We see that misclassification occurs between top and bottom patterns. We then re train our classifier using 10% of the data and categorize all of our top patterns as 'top' and bottom as 'bot. Key Takeaways: 1. Digital image processing and machine learning can be used for early detection of technical stock price patterns prior to their completion. 2. Misclassification occurs between top class and bottom class patterns (i.e. a Rectangle Top is Classified as a Head and Shoulders) 3. Patterns categorized by 'top' or 'bottom' are better than 98% accurate for forecasting on pixel values representing their extrema compared to around 50% for time series observations so image processing is a better method for detecting and forecasting patterns via Machine Learning as opposed to using raw time series.

ABSTRACT We study the short‐term predictability of price limit hits. This limit on the trading price is a policy measure imposed with the intention of stabilizing the markets and has been in place for several decades in the Chinese stock markets. We employ feature engineering on past return data and train machine learning models for each individual stock. The results show that a mildly complex model based on ensembling and downsampling the historical information of the majority class (“non‐hit” samples) can substantially improve the forecast performance of a naive guess of 50 % to about 66 % in terms of balanced classification accuracy between true positives and true negatives. We also find that price limit hits of older stocks and of stocks belonging to the tertiary sector are more predictable. We interpret this result with the argument that certain stocks with a longer history are more susceptible to speculative behavior, thus increasing the probability and predictability of such price limit hits.

Accurate stock price forecasting remains a challenging problem because financial markets are highly dynamic, nonlinear, and influenced by changing economic conditions. Deep learning models, particularly Long Short-Term Memory (LSTM) networks, have shown strong capability in capturing temporal dependencies in financial time series. However, their performance is often sensitive to hyperparameter selection, which can affect convergence, stability, and generalization. This research presents a comparative forecasting framework using three bio-inspired hyperparameter optimization algorithms: Sand Cat Swarm Optimization (SCSO), Particle Swarm Optimization (PSO), and Grey Wolf Optimization (GWO). Each of the optimization algorithms works on a common hyperparameter search space comprising the learning rate, dropout rate, batch size and number of units in the LSTM layers, while the input sequence length is fixed at 60 trading days. The models are compared with a fixed-hyperparameter baseline LSTM to validate whether hyperparameter optimization improves price forecasting. The models are evaluated using daily data from five DAX-index equities over two sample periods: 2018–2023 and 1996–2024, respectively, with 2023 and 2019–2024 being their respective held-out test sets. The LSTM predicts the next-trading-day log return, which is transformed into a forecast of the next adjusted closing price. Historical prices and technical indicators representing trend, momentum, and volatility are incorporated into the modeling process. Chronological data splitting, training-only scaler fitting, validation-based hyperparameter selection, and a held-out test set used exclusively for final evaluation are employed to limit look-ahead bias, with every model evaluated across three random seeds. Model performance is assessed using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), coefficient of determination (R2) and directional accuracy, along with support from Diebold–Mariano tests and Ljung–Box and ARCH residual diagnostics. LSTM–SCSO achieves the strongest aggregate error performance during 2018–2023, reducing mean RMSE by approximately 2.5% relative to the baseline. During 1996–2024, PSO–LSTM ranks first on RMSE, MAE, and MAPE, reducing mean RMSE by approximately 3.9%, while GWO–LSTM produces closely comparable results. Optimized models record the lowest equity-level mean RMSE for four of five equities in the short-term experiment and three of five equities in the long-term experiment. However, Diebold–Mariano significance varies across model–equity comparisons, and aggregate directional accuracy remains below 50% for all models in both periods. Long-term residual diagnostics also identify substantial autocorrelation and ARCH effects. The findings demonstrate that bio-inspired metaheuristic optimization can improve one-step-ahead LSTM forecasts, but its benefits depend on the optimizer, equity, and historical evaluation period.