# Average tournament stack: reproducible authored examples

Created 2026-09-21 for GTO Gecko. These are complete synthetic fields, not actual tournament data or a model of how fields usually develop. No personal data, sampling, random seed, solver or payout model is used.

Run `python generate.py`, then `node verify.mjs` in this directory. Standard Python and Node only. The generator computes with exact rational arithmetic before JSON conversion; the verifier uses independently expressed JavaScript arithmetic and fixed expected outputs. `inputs.json` contains all stacks; `results.json` preserves every result used in the article. `explorer.html` is a self-contained fixed-example explorer; it needs no network connection or account. All main values remain available in its static initial view without JavaScript. It stores and transmits no entries.

Checkpoint: between hands, before forced bets, with every remaining player's complete stack counted and no chips in a pot. Both fields have nine players and 900,000 chips. Hero is the unique 80,000 stack. The mean is 100,000 in both; median 50,000/80,000; Hero rank 3/5 counting from largest. Rank is 1 plus the number of strictly larger stacks; if ties occur this is shared competition rank, not a unique position. For an even player count median averages the middle two.

Big blinds are 2,000 or 4,000; chips stay fixed. A separate elimination transfers A's 10,000 stack to its 360,000 leader and removes the zero stack: 900,000/8 = 112,500 mean, median 55,000, Hero still 80,000 = 40bb at BB2,000. No additions, removals, deductions or chip-rounding adjustments occur. The equal-starting-stack shortcut requires counting every stack actually issued, including re-entries, and adjusting for other changes.

The fields prove that mean and Hero chips do not determine median or rank. They do not establish a typical tournament distribution, cash probability, equity, skill, correct aggression level or a safe folding threshold. The explorer deliberately contains fixed examples only. It is not an app screenshot or a real-time poker assistant. Study actual decisions with the relevant stacks, positions, payouts, blinds/antes and action context.

Definitions: NIST Measures of Location, accessed 2026-09-21: https://www.itl.nist.gov/div898/handbook/eda/section3/eda351.htm

`SHA256SUMS` records package bytes, excluding itself.

Combined change: after that elimination, changing BB to 4,000 gives mean 28.125bb and Hero 20bb. This uses the same 900,000 chips and eight players.