How it works · Daily 300

The first number is how many hands sit in the base the pool is measured against. Not my hands – all of them: six players at a table, and every one of their actions is in the file. The second is how many times the instruction behind all this has been rewritten. It changes after every error found, and after every correction I make to a calculation or to how it is read. That makes it the most honest measure of how often the method turned out to be wrong.

Every session is checked from several sides

One method is not enough – each has its own blind spot. So there are five, and they argue with each other.

The pool

How the opponents actually play, not how a textbook says they should. How often they open from each seat, how often they surrender the blind, how often they pay on the last card. All counted from the base rather than assumed. That is the set of hands every decision is then measured against.

The probabilities

Some numbers come from enumerating every way the cards can fall – millions of layouts, no model involved. The result, the fees, “the pot splits 85.7% of the time” come from there. No opinion can spoil a number like that.

My own commentary

While playing I say out loud the decisions that were not automatic: what I saw about the opponent and what I expected from him. Each remark is attached to its hand and checked against the arithmetic. Sometimes the arithmetic disagrees with me – and that is the most interesting thing an analysis can contain.

The solver

Contested spots are solved for equilibrium: how this hand is played if both sides play perfectly. Solved at the real pot and stack sizes – not similar ones, the actual ones. That matters more than it sounds: a neighbouring spot yields tidy numbers about a different hand.

The key hand, separately

One hand per session goes to GTO Wizard. Not the one that cost the most, but the one where the decision was most expensive: a multiway pot, or the point where the next bet is already about the whole stack. Big losses usually mean the cards arrived, not that anything was decided.

Business insights

While playing, or during the analysis, I try to spot the situation or the decision that could be the seed of a deeper insight, one that applies beyond poker. Claude helps me with that, and so does my own experience in life and in business.

And the check at the end

The finished text is checked independently against the source files: every sum, every percentage, every card. It regularly finds errors, which is a normal part of the process rather than an exception. What it finds is fixed before publication.

And one rule above all five: a decision is judged blind to its outcome. First the arithmetic says whether the action was right; only then does the ending come into view. A mistake is what the arithmetic condemns, not what happened to lose.

And one limit. Some numbers are exact enumeration and cannot be argued with. Others rest on the opponent's range, and a range is constructed rather than measured – his cards are rarely visible. Those I call estimates or give as a band.