How Momus uses the bookmaker's odds

There's a lazy way to talk about football models: "the bookmakers don't know what they're doing." They do. Sharp books move millions and bake in team news, injuries, and the smartest money in the market. A model that pretends the price doesn't exist isn't clever — it's flying blind.

So Modal doesn't ignore the odds. It uses them, deliberately, in three places. Here's exactly how.

1. First, we de-vig them

A bookmaker's odds don't add up to 100%. The extra is the margin (the vig, or overround) — the house's cut baked into every price. Take it at face value and every outcome looks a little more likely than it really is.

So step one is to strip the margin out and recover the market's true implied probabilities. A price of 2.00 isn't "50%" once you account for the book charging you on both sides — de-vigging turns the raw prices back into probabilities that actually sum to 100%. (We break the maths down in odds vs value.)

Only a de-vigged number is worth comparing anything to. Everything below runs on those.

2. We take the sharpest source available

Not all odds are equal. A recreational book shades its lines toward the crowd; a sharp book (like Pinnacle) prices to be right, not to be popular. So when a sharp price is quoted, we de-vig that. When it isn't, we fall back to the market average across books — noisier, but still a real consensus.

The cleaner the market signal, the better every number downstream.

3. The market as an anchor — when our own data is thin

This is the part most "AI predicts football" pitches won't admit: sometimes the model shouldn't lead.

Our Dixon-Coles model needs enough history on both teams to have a real opinion. For a well-covered league, it does — and it leads. But for a thinly-covered fixture (a minnow, a young competition, a cup mismatch), forcing a confident-looking model number would be dishonest. So there, Modal anchors on the de-vigged sharp price instead of inventing certainty.

You always get a grounded read. You just also get the honesty about where it comes from — the model when the model has earned it, the sharpest market when it hasn't.

4. The market as the thing to beat

Here's where it gets interesting. Once the market is de-vigged, it becomes both a benchmark and a target:

• Benchmark: every match page shows the model's read next to the market's — model vs market, side by side, so you can see exactly where they agree and disagree.
• Target: the gap between them is the edge. When the model rates a side higher than the de-vigged market does, that's a disagreement worth inspecting — and it's what the value board ranks the whole slate by. Not a pick. A gap you can argue with.

And we keep ourselves honest about it: on the track record, the model is graded head-to-head against the de-vigged market on the same fixtures. Same games, same scoring, in the open. "Beats the bookmaker" isn't a slogan — it's a number you can check.

What we don't do

We never turn the odds into a stake, a bet slip, or a "back this now." The odds go in as probabilities and come out as a read and a disagreement — nothing more. That's the whole line Modal holds: analysis, not tips.

The market is the sharpest opponent in the room. We don't pretend it away, and we don't blindly copy it. We de-vig it, lean on it when we should, and try to beat it where we can — and show our work either way.