Good article. I think a discussion of hold periods can be helpful. For example, in my trading it's quite normal to have 3-4 losing trades in a row with short hold periods. This is because the algo model can get faked out picking up the beginning of a new trend which doesn't happen. Then, one good trade held for an extended period of a trend could be 10x+ the losses of the losing trades.

So I think you are on the right track, but it only applies if your trading system as constants for exits (i.e always exit when +10% return or -5% return) so it can be modeled. But other systems where exits are not target based would be much harder to model in this way.

Personally, I think the best way to model "normal" vs "abnormal" losing streaks is just drawdown on the equity curve. That's the real "target" regardless of the type of trading system. So for example if in backtesting (assuming your system is not overfitted) you get drawdowns of up to 20% with an average of 10%, you can use that as a gauge to see if the current "loss streak" you are experiencing is normal (around average), at the max (around -20%), or in excess of that ..... the latter would indicate that your system may be breaking down and the edge/alpha you had found is no longer valid