How Walk-Forward Analysis is different from Backtesting
Traditional backtesting services find a fixed set of indicators for some time period in the past, and then presents the computed optimal performance as if it represents actual attainable results. Using these static indicators in the future is very unlikely to provide similar performance to the optimal performance, since that was computed with foreknowledge of how the stock price would behave.
Stock-O-Matic instead does the following:
1. Rewinds the clock to 5 years ago, and calculates the optimal set of technical indicators for the 5 years prior to that point (i.e., 10 years before today).
2. These indicators are then used to generate buy and sell signals for the next quarter. These are generated without knowing how the stock price will behave during this time period. Thus, they represent signals that would actually have been generated.
3. At the end of this quarter a new set of optimal technical indicators are computed for the previous 5 years.
The above process is repeated for each quarter of the past 5 years. The results shown for the walk-forward analysis represent actual buy and sell signals that would have been generated, and not just optimal, but purely theoretical results.
Read more about walk-forward analysis and backtesting at Wikipedia.