The graphic on the left shows the "optimal" variant of the strategy on the simulated historical prices (in sample data). This variant has the best performance as measured in Sharpe Ratio (SR) among all combinations of entry day, side, holding period and stop loss percentage.
The green line represents the stock prices. The blue line represents the profit or loss of the trading strategy. Notice how the blue line gets more and more profitable over time as the computer continues to optimize the system to fit historical data. In a matter of seconds or minutes, the computer creates what appears to be a very profitable equity curve based on the input historical stock price dataset.
The graphic on the right shows what happens when the computer-generated trading strategy developed on the left is deployed against new (i.e., "out of sample") stock price data. In most runs, the right strategy performance typically shows no resemblance to the "backtested and optimized" results shown on the left. In particular, the Sharpe Ratio ("SR") of the right-hand graph is typically much lower than the Sharpe Ratio of the left-hand graph. It is often negative (indicating an investment loss).
For a rigorous mathematical analysis of this phenomenon, see our technical paper in the Notices of the American Mathematical Society.
Click on the image above to play a video showing the progress of optimization process.
| Parameters for this run | |
| Maximum Holding Period | 7 |
| Maximum Stop Loss | 10 |
| Sample Length | 1000 |
| Standard Deviation | 1 |
| Seed | 1732347837 |
| Real-World Stock Market Data Used | No |
| Sharpe Ratio (SR) of OOS Data | |
| Deflated Sharpe Ratio (DSR) of IS Data | |