BAI8 record, 4 to 8 October: 19 calls settled
Every Desk call that settled between 4 and 8 October: 12 landed, 7 missed, against the 10.5 our own numbers expected. All 19 listed, misses first.
On this page
Nineteen BAI8 Desk calls settled between 4 and 8 October. The side we published won 12 and lost 7. Our own probabilities, added up, expected about 10.5 wins, so the week ran a little hot. Only two of the 19 carried a positive edge at the price we saw, and those two split one each. That is the honest summary: a small sample that proves nothing either way.
What a "call" is here
Every card on the Desk names one side of a match with our chance for it. The verdict then compares that chance with the best price we saw. A positive edge means the price was longer than our fair price. A negative edge means the price was shorter: we rated the side, but the market already had it priced above our number. "Won" and "lost" below mean whether that side won, nothing more. We take no bets, so there is no profit figure to show.
The misses
Seven calls lost. All dates are kick-off dates in UTC.
| Date | Match | Our side | Our chance | Best price | Edge (pts) |
|---|---|---|---|---|---|
| 8 Oct | Yankees v Rays (MLB) | Yankees | 50.0% | 1.67, Bally Bet | -9.8 |
| 7 Oct | Jets v Avalanche (NHL) | Avalanche | 58.6% | 1.59, theScore Bet | -4.3 |
| 7 Oct | White Sox v Guardians (MLB) | White Sox | 52.4% | 1.92, Betfair | +0.4 |
| 7 Oct | Paris Basketball v ASVEL (EuroLeague) | Paris | 55.1% | 1.40, Betsson | -16.4 |
| 6 Oct | Saints v Falcons (NFL) | Saints | 51.1% | 1.87, Betfair | -2.4 |
| 5 Oct | Guardians v White Sox (MLB) | Guardians | 50.5% | 1.67, Betfair Sportsbook | -9.3 |
| 5 Oct | Dodgers v Braves (MLB) | Dodgers | 54.7% | 1.72, 1xBet | -3.5 |
Baseball was the worst of it: one win from five MLB calls. Most of those were close to 50% on our numbers, which is another way of saying we did not have much of a view.
The calls that landed
Twelve won: four NHL (Oilers, Capitals, Hurricanes, Ducks v Panthers), four Nations League favourites (Albania, England, Slovakia, Switzerland), the Dodgers at Braves on 6 October (our one winning positive-edge call, 51.6% at 1.99), the Ravens, Troy in college football and Atherton Collieries in the FA Cup at 39.3%.
What the week does say
Two things are worth keeping an eye on, not concluding from.
First, the big favourites. We had Albania 78.9% against San Marino and England 67.8% against the Czech Republic, while the prices implied far more (1.03 and 1.14). All of those favourites won. One week cannot prove we are too cautious on mismatches, but it is the pattern we will watch.
Second, the accuracy score. The Brier score, the average squared gap between our chance and what happened, was 0.229 across the 19. Saying 50% for every call scores 0.25, and lower is better. So the numbers did a bit better than a coin, on a sample far too small to bank on. For why the price matters more than the hit rate, see what a positive edge actually means.
What this does not tell you
It does not tell you whether our numbers beat the market over time. With two positive-edge calls there is nothing to measure yet. It does not cover calls still waiting to settle. And a winning week is no more evidence of skill than a losing one would be of its absence: see how long a losing run can honestly last.
18+. Bet only what you can afford to lose.
Common questions
How did BAI8's calls do from 4 to 8 October?
Nineteen calls settled. The side we published won 12 times and lost 7. Adding up our own probabilities, we expected about 10.5 wins.
How many of those calls had a positive edge?
Two. Dodgers at Braves on 6 October at 1.99, which won, and White Sox v Guardians on 7 October at 1.92, which lost. The other calls were either priced shorter than our number or had no price.
Does 12 from 19 mean the model is good?
No. Nineteen results are far too few to separate skill from luck. A record has to run for hundreds of calls before the win rate says much on its own.