Too much nonsense! Good for clicks, but what a waste of time!
I'm fed up with all these pseudo informed posts about backtesting. A poster on another thread provided a good argument against his work's results: human brain and computer have to work well together. If your brain lets you down, no computer will yield an useful result. You might just not be aware of your weaknesses, so be a little more reserved not to look foolish.
Backtesting has a very limited usefulness in trading, and even that only when it is done properly.
It's not enough to be a programmer, even a good one. You have to fully understand what you test. In engineering, a good test engineer understands the product as well as the designer.
Otherwise: GIGO.
Backtesting is limited because it relies on a small, non-repeatable set of data, with very specific characteristics. This applies even when working with hundred years of tick data.
Backtesting of a non linear system can't be done only by testing its elements, because the system result is not the sum of its elements' results. Separately testing the strategy's entry and its exit won't give you useful information about the synergetic action of both.
I'm not against backtesting as a tool, but I understand its limitations.
Quote from nononsense:
What's the watchemecall Jack's method?
How does it backtest?
...
