Variance

Raise pocket aces from the cutoff, the button reraises with kings, and it's all in before the flop. You're about an 82% favorite - the best you'll do getting it in preflop - and a king comes on the board. You lose with the best hand, played right on every street. That's variance: the gap between what should happen on average and what shows up in one sample. Run that spot a thousand times and the aces win close to 820, but on a given night you can lose it twice running and still have done nothing wrong.

Variance is why a winning strategy can bleed money for weeks, sometimes months, before the math catches up. A player with a real edge can watch a bankroll shrink over a few thousand hands on bad run-outs and coolers alone, no adjustment required. Bankroll management exists for this reason: the swings between now and the long run can bust a player who is favored in every pot they play. Knowing the size of your own swings, and keeping enough behind you to survive them, is as much a skill as reading a board.

The beginner mistake is treating a small sample as proof of anything. Win a big pot with a bluff and you're a genius; lose three flips running and your game is broken - both readings forget that a few hundred hands tell you almost nothing about skill. One session, good or bad, is mostly noise. Judge each decision on whether it was correct with the information you had at the time, and judge results only over a sample big enough for the swings to cancel out.