Member-only story
I take a pretty simple view of making money. You just have to work harder and think better than all other market participants, and put yourself in the position to get lucky. “Thinking better” may actually mean acknowledging and accepting there are hard limitations to your thinking, such as cognition’s poor ability or inability to control your emotions, or that there are fundamentally unknowable or uncertain facts of about the world. It’s in this context that I’m considering the process of building trading models.
When building trading models, the usual process is to take an idea, build a model with clearly-defined trading rules and implement it on historical data while being mindful of and accounting for various biases, use various statistical analyses to see if there’s a favorable upside, and then implement it going forward.
The problem is that even properly “defining” backtest environments is difficult, before even going into formulating good strategies and actually backtesting them. Defining backtest environments may entail realistically modeling commissions, slippage, availability of stocks to borrow and short, Level 2 quotes that may change in response to your actions, one-off market events, etc. This all falls under the general heading of “explain” what happened in the past in order to realistically backtest.
