Cash Games
Position is the only edge you are handed for free
Acting last is worth more than any starting hand, because it turns two guesses into one and lets the same cards be played for a wider profit.
Bankroll & Discipline
Variance is not bad luck, and a downswing is not a verdict
A run of losing sessions tells a player almost nothing on its own. What it should change is the stake, the session length and the review, in that order.
Poker pays out in a scatter rather than a stream. Two players making identical decisions across the same number of hands can finish a stretch far apart, and neither result on its own says which of them decided better. That scatter is variance, and it is a permanent property of a game where the best hand does not always win and the winning hand is not always shown.
Why results resolve slowly
A single hand has an enormous range of outcomes and only one of them happens. A session is a few hundred hands, which is enough for the scatter to dominate and not nearly enough for skill to separate from it. This is why a losing week is uninformative and why a winning week is equally uninformative in the other direction. Neither is long enough for the underlying quality of the decisions to be visible through the noise.
The practical consequence is that results are the wrong thing to review. The right thing to review is the decisions: the spots that were unclear at the time, the hands where a rule of thumb was applied without checking whether it fitted, and the calls that were made because the pot was large rather than because the price was right. Those can be examined immediately. The result cannot.
What a stake is actually for
Choosing a stake is a way of setting how much of the scatter a player is exposed to. A stake small enough that an ordinary losing run is uncomfortable rather than damaging leaves the player able to keep playing, keep reviewing and keep improving. A stake large enough that an ordinary losing run is damaging changes the decisions themselves, because a player protecting something cannot call correctly at a marginal price.
The general rule that follows is unglamorous: move down before it is obviously necessary and move up only when the level below has stopped being interesting. Neither move is a judgement on the player. Both are adjustments to the size of the swing being carried.
Session length and the quality of decisions
Decision quality falls with fatigue long before a player notices it falling, and the first thing to go is usually the willingness to fold. The habit worth keeping is a fixed stopping condition set before sitting down, either a number of hands, a length of time, or a state of concentration, and a commitment to stop at it whether the session is winning or losing.
Stopping because the session went badly reinforces the idea that results should drive behaviour. Stopping at a pre-set point does the opposite, and it removes the longest and least profitable part of most sessions: the stretch played after the point where the player stopped thinking clearly.
None of this is a system for guaranteeing a result, and no stake selection removes the scatter. What it does is keep the scatter survivable, so that the decisions made a thousand hands from now are made by a player who is still at the table and still paying attention.