JUNE 24, 2014
Variance Without The Mathematics
Variance is the reason a good decision and a bad result can be the same event. It is also the single hardest thing to convey to anyone who has not sat through a long stretch of it, because from inside a session there is no visible boundary between being unlucky and being wrong.
The mathematics of it is not the interesting part, and it is not the part that does damage. The damage is done by the experience of a signal that arrives late, in the wrong order, and mixed with noise that looks exactly like signal.
Feedback that lies for a while
Most skills give feedback quickly enough to learn from. Play a wrong note and it sounds wrong immediately. Poker does not work that way. A player can make a run of genuinely poor decisions and be rewarded for all of them, then correct those decisions and be punished for weeks. The feedback is real, it is just arriving from a distribution rather than from the decision itself.
This produces a specific kind of confusion. The player who is losing does not know whether to change anything. The player who is winning does not know whether to trust anything. Both are looking at the same evidence and neither can read it, and the honest answer to what the last month meant is usually that it meant very little.
Why the swing is not the story
It is tempting to treat variance as drama, because swings look like drama on a graph. But the lived version is closer to weather than to plot. Nothing happens for a long time, then a great deal happens at once, and the person it happens to has no more influence over the sequencing than a person has over rain.
The thing worth attention is not the swing. It is what the swing does to the interval between decisions: how a player who has been losing starts making choices faster, in larger sizes, with less of a gap in between. The distribution stays the same. The person using it does not.