Neural Network Applications in Investment and Trading

Neural networks provide significant benefits in investment and trading applications. They are actively being used for such applications as predicting stock prices, determining asset allocation, and forecasting portfolio changes.

Below we have listed some of the common applications of neural networks in investment and trading. If you are currently using neural networks in your investment and trading application, we would love to hear about it.

NeuroDimension has also used its leading edge neural network technology to develop numerous investment and trading applications with a variety of companies. If you need neural network consulting or trading system development services for your investment and trading application, please contact NeuroDimension.

Sample Applications

Locate common characteristics between available assets (stocks on verge of breakout).
Forecast effects of changes to portfolio or trading approach (both fn approx and TS simulation)
Predict prices or best trading actions based on previous performance.
Group available assets based on similarities (similar trend direction, portfolio diversification)

Locate common characteristics between available assets

Detecting common characteristics in large amounts of financial data is a type of classification problem. Neural networks can be used to solve classification problems, typically through Multi-Layer Perceptron (MLP) and Support Vector Machines (SVM) type networks.

Examples of classification applications in finance include dividing research populations or data into groups for further study. For example, data can be extracted from databases to determine whether a stock is on the verge of a breakout.

Sample Study: Classifying level of return on stock market index

This sample study highlights the usage of neural networks in classifying the “level of return?on a stock index. The forecasting performance of a group of classification models is superior to that of a group of level estimation models. The classification models included in the study are aimed at forecasting the sign (direction) of index return whereas the level estimation models take the conventional approach to estimate the value of the return. The classification models perform better than their level estimation counterparts in terms of hit rate (number of times the predicted direction is correct). More interestingly, the classification models are able to generate higher trading profits than the level estimation models. Forecasting stock indices: a comparison of classification and level estimation models - Mark T. Leung, Hazem Daouk, An-Sing Chen Locate this paper on Google Scholar!

Our NeuroSolutions and TradingSolutions products are an excellent resource for classification and financial modeling applications. For an interactive example of classification in NeuroSolutions for Excel, download the free evaluation version of the software and view the demo called “Testing Classifiers?in the Help menu. For an interactive example of modeling in TradingSolutions, download the free evaluation version of the software and view the “Getting Started?manual in the Help menu.

Forecast effects of changes to portfolio or trading approach

Forecasting the relationship between multiple factors in financial data is a type of function approximation problem. Neural networks can be used to solve function approximation problems, typically through Multi-Layer Perceptron (MLP), Radial Basis Function (RBF) and CANFIS (Co-Active Neuro-Fuzzy Inference System) type networks.

Examples of function approximation in financial include predicting changes to prices and costs. For example, data from studies can potentially predict the next day’s closing price for stocks, Forex or even futures data.

Sample Study: Predicting Next Day’s Closing Price & Sensitivity Analysis

Using NeuroSolutions and TradingSolutions, this study investigates using “Sensitivity about the Mean?(in NeuroSolutions for Excel) to determine the key indicators to be used in the neural network model in TradingSolutions. The study was able to reduce the number of indicators in the model thus making it more efficient and more accurate. Identifying Relative Contribution of Selected Technical Indicators in Stock Market Prediction - Gary R. Weckman & Ranjeet Agarwala Locate this paper on Google Scholar!

Our NeuroSolutions and TradingSolutions products are an excellent resource for function approximation and financial modeling applications. For an interactive example of function approximation in NeuroSolutions, download the free evaluation version of the software and view the demo called “Multi-Layer Perceptron, Basic?in the Help menu. For an interactive example of modeling in TradingSolutions, download the free evaluation version of the software and view the “Getting Started?manual in the Help menu.

Predict prices or best trading actions based on previous performance

Forecasting the relationship between multiple factors in financial data is a type of time-series prediction problem. Neural networks can be used to solve time-series problems, typically through Time-Lagged Recurrent (TLRN) type network.

Examples of time-series predictions in finance include forecasting revenue and expense cost. For example, data from financial studies can forecast the Forex (currency) markets with a higher return.

Sample Study: Forecasting the Forex Market

This sample study highlights that a neural network model is applicable to the prediction of foreign exchange rates. The neural network model in this study outperformed all other models by having the highest Annualized Return of 29.68% in comparison to the other models averaging 15.65%. In addition, the neural network model had the highest percent of winning trades with 57.24% and the other models having an average of 47.75%. (Trading currencies is very risky and you may loose all or some of your investment. More risks of FOREX trading.) Modeling and Trading the EUR/USD Exchange Rate: Do Neural Network Models Perform Better? - Christian L. Dunis and Mark Williams Locate this paper on Google!

Our NeuroSolutions and TradingSolutions products are an excellent resource for time-series and financial modeling applications. For an interactive example of time-series prediction in NeuroSolutions, download the free evaluation version of the software and view the demo called “Time Lagged Recurrent Network?in the Help menu. For an interactive example of modeling in TradingSolutions, download the free evaluation version of the software and view the “Getting Started?manual in the Help menu.

Group available assets based on similarities

Grouping of financial data based on key characteristics is a type of clustering problem. Neural networks can be used to solve clustering problems, typically through Self-Organizing Map (SOM) type network.

Examples of clustering in financial include the detection of key characteristics in demographics and feature extraction. For example, data can be extracted for mutual fund investment managers to determine risk assessment.

Sample Study: Mutual Fund Grouping

This sample study highlights the usage of neural networks in grouping different mutual funds based on total annualized return, net assets, turnover ratio and many more inputs. The funds are separated into 3 separate groups by managers with less than 3 tenures, another cluster managed by managers with slightly more tenure and the third group managed by managers with substantially higher tenure. Financial Applications of Self-Organizing Maps - Guido J. Deboeck Locate this paper on Google Scholar!

Our NeuroSolutions and TradingSolutions products are an excellent resource for clustering and financial modeling applications. For an interactive example of clustering in NeuroSolutions, download the free evaluation of the software and view the demo called “Unsupervised Learning?in the Help menu. For an interactive example of modeling in TradingSolutions, download the free evaluation version of the software and view the “Getting Started?manual in the Help menu.