Name | Modified | Size | Downloads / Week |
---|---|---|---|
readme.txt | 2015-04-06 | 1.1 kB | |
copyright.odt | 2015-04-06 | 12.0 kB | |
license.odt | 2015-04-06 | 85.1 kB | |
Totals: 3 Items | 98.2 kB | 0 |
GENENG This is a research project to develop software which can predict changes in a stock price. The basis is genetic engineering. We start with a population of 'programs' (called proglets) and historical data about price changes of some stocks. We rank the proglets by their accuracy in price changes. We cull the worst. We breed the rest. And we repeat that for many generations. The final stable population of proglets is assessed for predictive power again using historical data. The input to each proglet is a set of price changes in other stocks. The output is whether the price of a target stock rises or falls. It is expected that this is done on a daily time step - inputs are start of trading prices, and output is close rise or fall. This project is intended to discover the effectiveness of this approach. Questions include: Choice of target and other stocks Choice of market Choice of trading conditions Time scale for future prediction. This work is copyright (c) 2015 Walter Milner Contact the copyright holder at wm@waltermilner.com