🔬 Pre-registered experiment · TIPS-EXP-001

The Italian Draw: a betting hypothesis, registered before the season

Our backtests found the draw systematically underpriced in Italian football. Backtest discoveries are usually mirages — so instead of selling it, we froze the rules in advance, hashed them, anchored them in Bitcoin, and will test them live for a full season. In public. Either way it ends.

Status

Registered · goes live with the first Serie A / Serie B match of the 2026-27 season (August 2026). Selections will appear in the public track record.

The hypothesis

Across six seasons of opening odds, the draw in Italy returned more than it should: Serie B blind-draw EV +7.2% (CI95 +0.5% to +14.3%, positive in 6 of 6 seasons), Serie A +4.9%, with a clean gradient — short draw odds (≤3.5) strongly positive in both leagues, long draw odds negative. Plausible mechanism: nobody loves the draw, so its price receives the least market correction.

The frozen rules (summary)

  1. Every 2026-27 Serie A and Serie B match with Pinnacle prices in our feed.
  2. Flag the draw (X) if and only if Pinnacle's opening draw odds are ≤ 3.50. No other filter, no discretion.
  3. Flat 1-unit stakes; entry at our first Pinnacle snapshot; benchmark = Pinnacle close (same-book CLV).
  4. Primary metric: average CLV with 95% bootstrap CI. Full season, no early stopping.
  5. Success declared only if the CI excludes zero. Otherwise the strategy is declared dead and gets a public autopsy.
  6. Every selection is logged before kick-off in the public track record and cannot be edited or removed.

🔒 Proof the rules were fixed in advance

The complete specification is frozen in a public file. Its SHA-256 hash is committed to our public GitHub ledger and anchored in the Bitcoin blockchain via OpenTimestamps. If we ever changed a comma after seeing results, the hash would change and the tampering would be provable.

SHA-256 2e8ac133f27a897d91a3ee83e31fa6df40835cd391f93e55f760ea599770cfbe

Read the frozen specification → Verify on GitHub (with OpenTimestamps proof) →

Why this might fail (written before, not after)

Italy came out of a scan of 17 leagues, so multiple-testing risk is real. The ≤3.5 band was chosen after seeing the data. Market effects decay. That is exactly why this is a forward test with pre-committed rules and not a betting recommendation: if the edge is a mirage, you will watch it die here, in public, and it will get its own entry in our strategy graveyard. Read the explainer →

Follow the test live in the track record →
Found this useful? Send it to a friend who bets.