Abstract:We consider the application of machine learning models for short-term intra-day trading in equities. We envisage a scenario wherein machine learning models are submitted by independent data scientists to predict discretised ten-candle returns every five minutes, in response to five-minute candlestick data provided to them in near real-time. An ensemble model combines these multiple models via a weighted-majority algorithm. The weights of each model are dynamically updated based on the performance of each model, and can also be used to reward model owners. Each model's performance is evaluated according to two different metrics over a recent time window: In addition to accuracy, we also consider a `utility' metric that is a proxy for a model's potential profitability under a particular trading strategy. We present experimental results on real intra-day data that show that our weighted-majority ensemble techniques show improved accuracy as well as utility over any of the individual models, especially using the utility metric to dynamically re-weight models over shorter time-windows.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to solve the challenges of stock prediction in short - term intraday trading, especially how to improve the accuracy and potential profitability of prediction through the ensemble learning method. Specifically: 1. **Challenges in intraday trading**: In a high - frequency trading environment, the latest K - line data (i.e., opening price, highest price, lowest price, closing price, and trading volume) is provided every 5 minutes, and it is required to predict the return rate of the next 10 K - lines in real - time. Due to the intense market fluctuations and large amount of information, it is difficult for a single model to maintain high accuracy continuously. 2. **Application of ensemble learning**: To overcome the limitations of a single model, the author proposes a framework named Numin, which combines prediction models submitted by multiple independent data scientists using the Weighted - Majority Algorithm (WMA). These models adjust their weights dynamically based on historical performance to ensure that the most reliable predictions are given higher weights. 3. **Dynamic weight adjustment**: The weight of each model is updated dynamically according to its performance in the past period. The weight is not only based on accuracy but also introduces a "utility" metric, which is a proxy for the potential profitability of the model under a specific trading strategy. In this way, the system can quickly adapt to market changes and identify the best - performing models. 4. **Experimental verification**: The author conducted experiments using real - world intraday trading data, and the results show that, compared with a single model, the weighted - majority ensemble method has a significant improvement in both accuracy and utility, especially when the utility metric is used to adjust the weights dynamically in a shorter time window. ### Formula representation - **Weight update formula**: \[ w_j^{(r)}=\alpha\cdot\tilde{s}_j^{(r)}+(1 - \alpha)\cdot w_j^{(r - 1)} \] where \(\alpha=\frac{2}{|win|+ 1}\), \(\tilde{s}_j^{(r)}\) is the normalized score, and \(w_j^{(r)}\) is the weight of model \(j\) after the \(r\)-th round. - **Utility metric**: \[ U=\sum_{t = 1}^{m}(\text{actual return}-\text{expected return}) \] The utility metric evaluates the profitability of the model in actual trading. Through these methods, the Numin framework can take advantage of multiple models in a constantly changing market environment and provide more accurate and more profitable trading suggestions.

Numin: Weighted-Majority Ensembles for Intraday Trading

Fusion of Multiple Indicators with Ensemble Incremental Learning Techniques for Stock Price Forecasting.

An efficient real-time stock prediction exploiting incremental learning and deep learning

AI-Driven Intraday Trading: Applying Machine Learning and Market Activity for Enhanced Decision Support in Financial Markets

Stock market prediction employing ensemble methods: the Nifty50 index

Quantitative trading models based on Sufficient Dimension Reduction and Ensemble Learning

DoubleEnsemble: A New Ensemble Method Based on Sample Reweighting and Feature Selection for Financial Data Analysis

Stock Price Forecasting with Empirical Mode Decomposition Based Ensemble \(\nu \)-Support Vector Regression Model

Reinforcement Learning based dynamic weighing of Ensemble Models for Time Series Forecasting

An Intelligent Fusion Model with Portfolio Selection and Machine Learning for Stock Market Prediction

A Time Series Analysis-Based Stock Price Prediction Using Machine Learning and Deep Learning Models

An efficient hybrid approach for forecasting real-time stock market indices

Temporal mixture ensemble models for intraday volume forecasting in cryptocurrency exchange markets

A stock market trading framework based on deep learning architectures

An Ensemble Method of Deep Reinforcement Learning for Automated Cryptocurrency Trading

Novel Modelling Strategies for High-frequency Stock Trading Data

Volatility forecasting with machine learning and intraday commonality

Discrete Wavelet Transform-Based Prediction of Stock Index: A Study on National Stock Exchange Fifty Index

Prediction of Stock Price Using Statistical and Ensemble learning Models: A Comparative Study

Intraday Return Forecasts and High-Frequency Trading of Stock Index Futures: A Hybrid Wavelet-Deep Learning Approach

Predicting stock price of construction companies using weighted ensemble learning