Home › Methodology

Selection Room

The method behind Selection Room

A plain-English, but complete, description of how every number on this site is calculated — for cricket leagues today, and built to extend to any team sport with published results.

Selection is a comparison problem: who contributed the most, given what they were actually asked to do and the chances they actually got? Selection Room answers that from a league's own published scorecards — every run, every wicket, every catch, priced against what that contribution was really worth in the match and division it happened in, then rolled up per player per game.

It is inspired by the player-value models used in the professional game — most notably the PCA MVP rankings, the Professional Cricketers' Association's own points system for ranking player contribution across professional cricket — adapted here for a Saturday league's own published scorecards, with division-calibrated pricing in place of a professional competition's own reference data. Selection Room is not affiliated with or endorsed by the PCA.

A wicket is not worth a flat number. Dismissing a division's best number 3 is worth more than dismissing a number 11 who rarely faces ten balls — so for cricket leagues, every wicket is priced at the dismissed batter's own POSITION BASELINE: what that batting slot typically scores in that division this season, recomputed from every innings batted at that position, across every division separately. A division with no baseline yet for a position falls back to a flat runs-per-wicket figure for that division instead, so a wicket is never left unpriced.

A wicket's value is then split between whoever actually earned it, by how the batter was dismissed:

• Bowled, lbw, hit wicket, caught and bowled — the bowler earns the full value. No fielder was involved.
• Caught (including caught behind) or stumped — the bowler earns 70%, the credited fielder or wicketkeeper the other 30%. Both contributed to the dismissal; the split reflects that.
• Run out — the bowler earns nothing at all; the credited fielder(s) earn the full value, split evenly for a joint run out. A run out is not a bowling achievement.

A wicket a scorecard cannot name-match to a bowler (mostly games recorded by initials only) is still priced, at a flat division-wide rate, with no fielding share to split.

Every price above is division-specific, not one number for the whole league. For cricket leagues, two figures are re-derived weekly from every innings recorded in that division this season: the division's own runs-per-wicket (its overall scoring rate, the fallback wicket price and the "proven wicket-taker" bowling-average bar), and its par strike rate (the pace batting is typically scored at there). A Premier division and a Division 8 are priced from their own real matches, never from a single league-wide constant.

A run is not worth the same in every match either. Two multipliers adjust batting value for the game it was scored in:

• Match scarcity — the division's own runs-per-wicket divided by that particular match's runs-per-wicket (both innings combined), clamped to a modest range. A low-scoring match makes every run harder-won and worth more; an easy pitch makes runs cheap and worth less.
• Strike-rate-vs-par — rewards or penalises tempo against the division's own par strike rate, damped so a handful of balls faced cannot swing a whole game, and neutral for an innings with no ball-by-ball data recorded.

Every judgement on this site has a minimum sample size before it is allowed to fire at all — a real pattern, not noise from three games. For cricket leagues specifically:

• A batting dismissal-method read needs at least 5 real dismissals before it says anything.
• A bowling dismissal-method read needs at least 8 attributed wickets — a higher bar, for the extra unknowns in attributing a bowling wicket.
• Both reads use a dominance rule, not a priority order: only the strictly largest of the three dismissal clusters (straight, caught, stumped) is ever considered, and it still has to clear its own share of the total — a genuinely mixed profile earns no verdict at all, just the raw counts.
• Every threshold above is a real fraction of a real sample (a quarter of dismissals, half of dismissals, and so on) — never a raw count alone, so a single repeated event in a tiny sample can never masquerade as a pattern.

• Did-not-bat is neutral. A not-out or a did-not-bat innings never counts against a player — there is no dismissal to hold against them.
• Position-fair baselines. A lower-order batter is compared against what THAT position typically scores, never against the division's overall average — which would be unfairly dominated by top-order innings.
• Opportunity floors. A player with too few real opportunities — too few innings batted, too few overs bowled — carries a "limited evidence" badge instead of a ranking judgement. Selection Room never scores a player on chances they were not given.
• Deliberately excluded. No part of the impact score includes a win/result bonus or captaincy. It only ever prices runs actually scored and dismissals actually recorded.