Stock market prediction - Wikipedia

McGill University researchers have developed a software program to make day to day predictions of stock price variations. The prediction algorithm is based on a new hypothesis about the predictability of stock prices: McGill University is looking for users to further test and validate the prediction model, and for partners to commercialize the software.

The other trading strategies were: Simulation results show that, when taking into account real-life constraints, such as transaction costs, slippage and taxes, the TCE turns out to be the best performing strategy, leading to significant gains as compared to the three other approaches. McGill researchers have designed an algorithm to test the validity of the Live Market Hypothesis, called the Temporal Correlation Estimator TCE.

This algorithm estimates the temporal covariance between stock price variations, and use this data to make short term predictions. A simulated trading environment has been designed in order to assess the potential of the TCE. This environment is designed to realistically represent a real trading environment and its constraints, so as to establish whether a trading strategy based on a given predictive algorithm has potential for yielding profit in the real world.

Stock Market Forecast Tools Download

For each active trading day within the input range, it then queries the input algorithm to know which stocks, among the list of market entities given as input, are expected to rise in value from the current day to the next.

Help Center Join AUTM. Home Organizations Technologies Needs Capabilities Startups Success stories People.

Background Historically, work in the area of stock market prediction has been divided into four categories: However, despite the tremendous amount of research efforts in the area of stock market prediction only limited success has been obtained to this point, leading many to doubt the feasibility of stock market prediction. Market behaviorists propose that it is not possible to consistently outperform the market through decisions based on past historical data, since stock prices already convey all publicly known information and reflect the true underlying value of the stock the company represents.

According to this view, short term stock prices variations follow a random walk, and such variations are essentially noise that cannot possibly be predicted.

Short term stock market forecasting. Track Code Manager Derrick Wong Background Historically, work in the area of stock market prediction has been divided into four categories: McGill researchers rather see the market as a complex, but organized and coherent system.

In this global market, millions of various entities make decisions that affect future outcomes.

More specifically, McGill researchers postulate that the changes in stock prices, both in the long and in the short term, are indirectly caused by the actions taken by market entities.

That is, the entities that are part of the market govern the future of the market. Furthermore, because of the inherent interactions between market entities companies competing, cooperating, buying each other, etc.

Stock and Fund Prediction. On-demand Stock Forecasting

McGill researchers call this the Live Market Hypothesis. According to this hypothesis, the market, in the short term, is not fundamentally unpredictable, and stock prices do not follow a random walk. Instead, McGill researchers propose that short term predictions can be made, by analyzing the existing relationships between market entities.