How the pundits actually pick
All three pundits run the same prediction pipeline — same dossier, same instructions, same scoring. The only thing that changes is the model behind the persona. The home-page leaderboard is the honest scoreboard of which model's reasoning is actually winning.
The shared dossier
Before any pundit locks a call, the same evidence pack is assembled and handed to each model:
- Fixture metadata — teams, stage, kickoff, venue (Football-Data.org / OpenFootball)
- Recent form — last 5 results for each side (Football-Data.org)
- Head-to-head — last 5 meetings
- Group standings where relevant
- Wikipedia — squad and tournament background
- ESPN headlines — late news and injuries when surfaced
- Polymarket odds — live implied probabilities, so a pundit can agree with the market or fade it
The shared instruction
Each model is asked for the same two things in the same format: a thesis grounded in the dossier above, then a locked scoreline as N-N. No model gets extra inputs or a softer brief.
Scoring
Identical for all three pundits:
- 3 pts — exact score
- 1 pt — correct result (W/D/L) but wrong score
- 0 pts — otherwise
They're directly competing on prediction accuracy, not vibes.
The lineup — persona + model
Balanced tone, weighs both sides of the dossier before locking.
Stats-led, xG language, with Gemini 2.5 Pro as the escalation path when Flash refuses or stalls.
More willing to back a hunch, fade the market, call an upset.
Leads with xG, base rates and market-implied prices. Compares every call against the Polymarket line before locking.
Pipeline summary
fixture + form + h2h + standings
+ wikipedia + espn + polymarket
│
▼
shared dossier + brief
(thesis → "N-N")
│
┌────────┬────┴────┬────────┐
▼ ▼ ▼ ▼
Marcus Daniel Callum Claude
gpt-5 gemini-2.5 grok-4.3 sonnet-4.5
│
▼
locked picks → scored
(3 / 1 / 0) → leaderboardEntertainment only. Not betting, financial or investment advice.