Every pick, won or lost
Every pick here went out before its puck drop and was graded from the final score at the horn. None has been edited or taken down since, so the misses stand next to the hits, and each figure tells you how many picks it rests on.
The record covers 32 days and 9,253 picks so far — early days. A short run cannot separate a sound model from a hot streak, so treat these figures as a first look and check back as the sample grows.
Daily win rate
Since 11 Oct · every graded dayMonth by month
All graded picks, then the best bets on their own| Month | Win rate | Best bets | Picks |
|---|---|---|---|
| Nov 2026 | 60.4% | 84.4% 38 of 45 | 2,214 |
| Oct 2026 | 61.8% | 74% 71 of 96 | 7,039 |
By market
Every settled pick, split three waysBy confidence
Does the model’s own score predict anything?| Confidence | Picks | Won | Win rate |
|---|---|---|---|
| 90 and overany market, not only best tips | 65 | 51 | 78.5% |
| 80 to 89 | 214 | 179 | 83.6% |
| 70 to 79 | 2759 | 1939 | 70.3% |
| 60 to 69 | 2966 | 1878 | 63.3% |
| Under 60 | 3249 | 1643 | 50.6% |
Trust is the model’s own measure of how sure it is, so the higher bands ought to win more often than the lower ones. The sample is still too small to prove it either way: each band shows how many picks sit behind its rate, and a rate means only as much as that count.
Day by day
The last two weeks, newest first| Day | Picks | Won | Win rate | Best bets |
|---|---|---|---|---|
| Wed, 11 Nov | 56 | 33 | 58.9% | — |
| Tue, 10 Nov | 207 | 125 | 60.4% | 75% 3 of 4 |
| Mon, 9 Nov | 64 | 37 | 57.8% | 75% 3 of 4 |
| Sun, 8 Nov | 411 | 250 | 60.8% | 85.7% 6 of 7 |
| Sat, 7 Nov | 468 | 291 | 62.2% | 80% 8 of 10 |
| Fri, 6 Nov | 156 | 97 | 62.2% | 75% 3 of 4 |
| Thu, 5 Nov | 136 | 86 | 63.2% | 100% 1 of 1 |
| Wed, 4 Nov | 140 | 79 | 56.4% | 100% 3 of 3 |
| Tue, 3 Nov | 72 | 43 | 59.7% | 100% 2 of 2 |
| Mon, 2 Nov | 72 | 37 | 51.4% | 75% 3 of 4 |
| Sun, 1 Nov | 432 | 260 | 60.2% | 100% 6 of 6 |
| Sat, 31 Oct | 567 | 383 | 67.5% | 85.7% 6 of 7 |
| Fri, 30 Oct | 616 | 374 | 60.7% | 66.7% 4 of 6 |
| Thu, 29 Oct | 272 | 180 | 66.2% | 66.7% 2 of 3 |
Each recent day links to its board, every pick on it graded.
How the record is kept
Each pick is a single call on a single market — the winner, the handicap or the total. It goes out before the puck drops with the odds it was set against and a trust figure, and the final score grades it. A push or a game that is not played settles nothing either way, so it sits in neither column.
Once a game starts its picks are frozen, and nothing is ever taken out. That is what this page stands on, and it can be checked: today’s hockey predictions go out before the games, this page grades the very same picks afterwards, and both read from one source.
You will not find a profit figure here. A return on stakes counts money no one actually wagered, two or three long shots can swing it more than a month of picks, and it frames predictions as an investment, which they are not. The fair test of a model is how often it is right and over how many calls — the two numbers this page shows.
Frequently asked questions
Yesterday’s hit rate is no promise for tonight: the picks are a model’s judgement, never a sure thing. Stake only what you could lose without it mattering. How we make our predictions