Last week we published our GW42/43 Heinz — the six Championship and League One fixtures where our model showed the biggest gap versus Bet365. We were clear that one weekend proves nothing either way. Then all six lost.
This is the follow-up we said we’d write. Not to make excuses — there aren’t any — but because transparency is the whole point of doing this in the open.
The Six Picks — What Happened
Six picks. Six losses. A flat-token return of −6 BTP tokens on 1 BTP token per pick. That’s the bad news, and we’re not dressing it up.
BTP tokens are an imaginary unit used purely to track model performance over time. They represent no real money — see the note at the bottom of this post.
The Picks That Didn’t Make the Heinz — Four from Four
Here’s what makes this weekend worth analysing beyond a simple “the model was wrong.” The four correct calls from that same pool of fixtures were all picks that didn’t make the Heinz cut — they had positive EV but ranked below the top six:
Four from four, including Portsmouth at 6.25 — the upset of the weekend. These weren’t published as Heinz picks because their EV ranked 7th–10th that week. That’s the nature of a top-6 cut-off.
Is that a coincidence? Almost certainly, given the sample size. But it does illustrate something real: higher EV typically means higher odds, and higher odds means higher variance. The Heinz picks are the ones where the model diverges most sharply from the market — and the market isn’t always wrong.
The Full Season Picture
Across all tracked picks at EV ≥ 1.20 since we started publishing this data — tracked in BTP tokens, our imaginary unit for measuring model performance:
A 28.6% correct call rate sounds low until you look at the average odds: 4.76. The model’s token tracking shows that four correct calls from fourteen, at those average odds, produces a positive token return — which is what the +35.9% token ROI reflects. This is a measure of model calibration against market odds, not a financial result.
By league:
Championship picks are averaging odds of 6.27 — long-shot territory — and the two correct calls produce a +108.9% token ROI on that small sample. League One sits at −9.4% over eight picks with shorter average odds of 3.62. Both figures are far too small a sample to draw conclusions from.
THE HONEST CAVEAT
Fourteen picks is not a sample size. It is an anecdote with a spreadsheet attached. A good run or a bad run across 14 data points tells you almost nothing about whether the model has genuine edge over the market.
We need hundreds of tracked picks before the token ROI figure becomes meaningful. We are publishing these results every gameweek because building that record — in public, without editing — is the only honest way to find out. We’ll keep going.
GW44 Heinz Coming Friday
The GW44 Heinz will publish on Friday as usual, based on GW44 and GW45 fixtures across both leagues. Same methodology, same filters, same accountability post if it goes wrong again.
If you want to understand the model before the next set of picks lands, the explainer posts are here: Championship model | League One model.
Not betting advice. BTP tokens are an imaginary unit used solely to track model performance over time. No real money is implied or recommended. Football probability models are for analytical and entertainment purposes only. If you do bet, please do so responsibly. 18+ BeGambleAware.org.
