At 3.25, how many actually drew?
Take every match in the database where a bookmaker quoted a price for the draw, sort them into bands by that price, and count what happened. Two readings come out of it: how often a band drew, and what share of all draws that band holds. Then, band by band, the exact scores those matches finished on. Nothing here is a prediction — it is a count of matches that have already been played.
| Draw price | Matches | Drew | Draw rate | Share of all draws | Price implies | Backing every draw | Scores |
|---|---|---|---|---|---|---|---|
| under 2.80 | 828 | 36.23% | 37.01% | ||||
| 2.80 – 2.90 | 1,694 | 34.77% | 35.07% | ||||
| 2.90 – 3.00 | 3,866 | 34.69% | 33.90% | ||||
| 3.00 – 3.10 | 6,642 | 33.12% | 32.80% | ||||
| 3.10 – 3.20 | 9,645 | 31.15% | 31.77% | ||||
| 3.20 – 3.30 | 13,186 | 30.06% | 30.80% | ||||
| 3.30 – 3.40 | 16,289 | 29.28% | 29.89% | ||||
| 3.40 – 3.50 | 17,167 | 28.76% | 29.03% | ||||
| 3.50 – 3.60 | 15,694 | 27.11% | 28.22% | ||||
| 3.60 – 3.70 | 13,579 | 26.71% | 27.45% | ||||
| 3.70 – 3.80 | 10,567 | 25.40% | 26.72% | ||||
| 3.80 – 3.90 | 8,074 | 25.19% | 26.02% | ||||
| 3.90 – 4.00 | 6,450 | 24.00% | 25.36% | ||||
| 4.00 – 4.50 | 17,911 | 22.53% | 23.81% | ||||
| 4.50 – 5.00 | 7,755 | 20.16% | 21.22% | ||||
| 5.00 – 6.00 | 6,637 | 16.68% | 18.53% | ||||
| 6.00 – 8.00 | 4,171 | 12.88% | 14.85% | ||||
| 8.00 and up | 2,004 | 8.98% | 10.07% |
Bands are half-open — "3.20 – 3.30" is every price from 3.20 up to but not including 3.30. Each match contributes its closing price where one was recorded, otherwise the only price there is. "Price implies" still contains the bookmaker's margin, which is why it sits above the draw rate in almost every row: that difference is the margin.
Take the price the question usually starts from, 3.25. It falls in the 3.20 – 3.30 band, which holds 13,186 matches. Of those, 3,963 finished level — a draw rate of 30.06%. Those draws are 9.29% of every draw in the priced set, so roughly one drawn match in 11 was priced there.
The price itself implied 30.80%, which is higher than the 30.06% that actually happened — the gap is the bookmaker's margin, and it is the whole story of the last column. Backing the draw in all 13,186 of those matches returns −2.40% per unit staked.
Now the column that matters. Backing the draw in every priced match returns −3.59% per unit, and 16 of the 18 bands lose — the worst, 6.00 – 8.00, at −13.20%. 2 bands finish above break-even: 3.00 – 3.10 at +1.01% over 6,642 matches, and 2.90 – 3.00 at +2.34% over 3,866 matches. That is 6.48% of the priced set clearing zero by a point or two, on samples where the draw rate's own margin of error is about a point — in other words, inside the noise, and in the thin short-price bands where a single season's results move the figure. It is not an edge, and it is left visible rather than smoothed away.
A draw rate that looks generous in one band is the market pricing those matches correctly and keeping its margin, not an opening. "Price implies" sits above the draw rate almost everywhere, and the size of that gap is roughly the size of the loss underneath it.
Two different questions get confused easily, so both columns are here. "Draw rate" is how often a match in this band drew. "Share of all draws" is how many of the database's draws happen to sit in this band — high in the middle simply because most matches are priced in the middle, and low in the tails even where the draw rate is high. Neither says anything about the next match.
Every final score the matches in this band produced, most common first. Level scores are marked. Read "Of the band" for the plain version of "1-1 at this price was 10%"; read "Of the band's draws" for how a draw here tends to look once it happens. They are not the same number, and the difference is large.
| Score | Matches | Of the band | Of the band's draws |
|---|---|---|---|
| other | 464 | 11.12% | — |
| 2-0 | 420 | 10.07% | — |
| 2-1 | 402 | 9.64% | — |
| 3-0 | 383 | 9.18% | — |
| 1-0 | 358 | 8.58% | — |
| 3-1 | 273 | 6.54% | — |
| 1-1 | 251 | 6.02% | 46.74% |
| 4-0 | 238 | 5.71% | — |
| 1-2 | 160 | 3.84% | — |
| 4-1 | 146 | 3.50% | — |
| 2-2 | 145 | 3.48% | 27.00% |
| 0-1 | 141 | 3.38% | — |
| 0-2 | 128 | 3.07% | — |
| 0-0 | 118 | 2.83% | 21.97% |
| 3-2 | 111 | 2.66% | — |
| 4-2 | 74 | 1.77% | — |
| 1-3 | 72 | 1.73% | — |
| 0-3 | 69 | 1.65% | — |
| 0-4 | 53 | 1.27% | — |
| 2-3 | 49 | 1.18% | — |
| 1-4 | 43 | 1.03% | — |
| 4-3 | 28 | 0.67% | — |
| 3-3 | 21 | 0.50% | 3.91% |
| 2-4 | 15 | 0.36% | — |
| 3-4 | 7 | 0.17% | — |
| other draw | 2 | 0.05% | 0.37% |
Scores are capped at 4-4: anything above it is collected under "other draw" (a level score of 5-5 or higher) or "other". Those two are kept apart so the draw column still adds up to 100%.
This page is a count of finished matches. It does not forecast, it does not rate a price as good or bad value, and it carries no staking plan. The one thing it settles is a question people ask a lot — "how often does a draw at this price actually come in?" — and the answer is: close enough to what the price said that the margin swallows the difference.
Two bookmakers are offered because they cover different matches. Pinnacle is the sharper book and closes almost everything it prices, so its bands are the ones to trust on rates; Bet365 reaches more matches but mostly at opening prices, which drift. If the two disagree on a band, that disagreement is the finding.
Every match in FDB4 that carried a bookmaker's draw price, sorted into bands by that price and counted: how often each band ended level, what share of all the database's draws it holds, what the price itself implied once the margin is left in, and what backing the draw in every match of the band would have returned. Each band then breaks down into the exact scorelines its matches produced, so the share of 1-1 results at a given price can be read two ways — as a share of the band's matches and as a share of the band's draws. Descriptive statistics over finished matches, not a forecast and not betting advice.