Byrne Hobart has a useful analysis [$] of stop-losses today. It's as fun as it is useful.
Poker players have long used (and debated, and been proud about not using) stop-losses. How do Byrne's points look through the poker lens?
First, a big loss means that your leverage has changed. The (rough) poker translation of this is that your edge in a game relative to your bankroll has changed. This is strikingly absent from a lot of poker discussion. But a large loss is often at least 10% of a reasonable bankroll, and often a much bigger portion of the bankroll someone is actually using. Even if there's still a justification for playing the game, the decision relative to some smaller game often looks different.
Second, a big loss means that there's information you're missing, so persisting includes a tax on your finite attention. Sometimes this just isn't true in poker (if, e.g., your loss is due to simple bad beats). But this is more true than poker players think: game conditions when you're losing are often different than you took them to be. And even if they weren't, the fact of the loss itself changes them, because people play differently against opponents with losing images than with winning ones.
The way I've often thought about this is different still, but related to each of these. When you observe an extreme outlier, you should usually vastly reduce your credence in the model with respect to which the event counts as an outlier. In crude terms: suppose that you're 95% sure that some model of the situation is right. If you observe an event that is a four-sigma outlier according to the model but much less likely if the model is wrong, then (by Bayes' theorem) your model is almost certainly wrong. So if you were making commitments on the basis of that model, you should stop doing so.
In poker terms: if you're stuck that bad, it's much likelier than you thought that (i) you're on tilt, (ii) you're not as good as you think you are, (iii) you're not as good in that game as you think you are, (iv) you're getting cheated, or (v) something else is going on that you haven't even thought of.
About a decade ago, it happened pretty often that baseball teams were making comebacks (to win divisions or make playoffs) that the preferred models suggested were 100:1 (or much worse) underdogs. It was disappointing to me how many people accepted the conclusion that they were living in an age of miracles. They would have done better to reject the models.