1 min read

When do we really learn something?

In trading, it is often very easy to identify an error or place for improvement in a post session review/reflection period. However, we cannot really say that error has been learned or addressed if that is all that happens. If you commit the same error two weeks from now, did you really learn?

Observation of a phenomena is not sufficient for learning to take place.  One must identify the phenomenon then implement mechanism to address it and then forward test that mechanism for its durability.

Say you "learn" that trade X does not work in condition Y. For starters, that has to be an observation from more than a handful of datapoints (especially if all in a tight time period/market regime). Then, you have to abstract from the positive claim that trade X does not work in condition Y, the normative claim: I should not take trade X in condition Y.

And here is the important step. Now you have a norm and need to use more than will power to enforce it. Mechanism must be built out here that works towards accomplishing that norm. I.e. this alert will go off in condition Y that invalidates the trade. Or the alternative is you have not really learned it this time, and it is likely you will have to "learn" it a few more times in painful ways before that learning actually sticks.