James Duckworth v Cameron Norrie, Bybit Stockholm Open
James Duckworth
Cameron Norrie
WonFavourite
Final score 7-5 6-0
Why the model leans this way
Tournament startHover or focus a bar for its exact value and the inputs behind it.
The bars add up to log-odds of -0.445 for Duckworth, which is Duckworth 39.1% and Norrie 60.9%. No effect here: missing serve data, unranked.
Log-odds are the model's own units; near 50%, 0.1 is about 2.5 percentage points. Inputs overlap (overall and surface Elo especially), so one bar is not that input's independent effect, and none of this is a cause.
Player comparison
As the model saw themValues stored with this forecast, as the model saw them. Bars compare the two players on each row; rows use different units, and rows that can be negative show values only.
Forecast history
Every stored version- Tournament startNorrie 60.9%Scored
Filled points are the scored (first-issued) forecasts; hollow points are later revisions. Points are separate stored forecasts, not a continuous line.
VerificationStored before the scheduled startCode matches the frozen protocol#2 in the ledgerShow recordsHide records
- Tournament start · issued
- Prediction id
- 2514ed7db98f4cc2f0abf688157026706af0e03a27d80dfd70ab42a3eada348e
- Inputs (SHA-256)
- de87d1c16fbe6c8103856a3f4453e072a47d1919a8c873717246b5afea88222a
- Model artifact
- a34233a3876d89fe3cd70c782b0d0a31de9192ad0895e05332c7bb5a93fd246c
- Code
- 21911efc3219c875e35155290de8ff29c257e182e5c9b99785cc99daa58981e8
- Ledger
- position 2, chain hash 9b5e099b3bf586080630e8904e7eb9809b1a6851223d5ce80f8beb998eb59514
Decision log
- Tournament start: Forecast issued.
- Tournament start: Checked again: inputs unchanged, so no new forecast.
- Tournament start: No forecast: it was too close to the start (forecasts are issued at least 30 minutes before play).
- (2 runs)Round-aware: Neither player has played earlier in this tournament, so there is nothing for the round-aware update to add.
- Round-aware: No forecast: it was too close to the start (forecasts are issued at least 30 minutes before play).
Result reported by ESPN; each source is stored separately and a disagreement is never scored.
These checks show that a forecast was recorded before the match and has not changed since. They say nothing about whether it was right; that is what the Model Lab measures. Every forecast is also listed in the Forecast Archive.