The question
"It's 0-0 and the draw keeps getting shorter even though nothing is happening. Then someone scores and it leaps out. Why does in-play football price movement look like that?"
Because two different things move an in-play price. One is the clock, which works a little every minute. The other is goals, which work all at once. Understanding both is the foundation of every in-play trade.
The idea in one sentence
As time passes without a goal, the expected goals left in the match shrink and prices drift smoothly; when a goal goes in, the score changes and prices jump to a new level in an instant.
The picture
Think of the match as a store of expected goals being used up. A side expected to score 1.5 goals in 90 minutes has 0.75 left at half-time, if nothing has happened, and 0.25 left with 15 minutes to go.
That drives the smooth part:
- The draw shortens at 0-0. Fewer goals left means the level score is more likely to last.
- Over 2.5 drifts. Three goals in the time remaining gets less likely every minute.
- Both win prices drift at 0-0. Each side has less time to score the goal it needs.
Then a goal arrives. Betfair suspends the market, and when it reopens the price has jumped. There's no trading in between: the price doesn't pass through the levels on the way. A red card is a smaller jump of the same kind: it doesn't change the score, but it changes how many goals each side is likely to get from here.
This combination of steady drift and sudden jumps is what mathematicians call a jump process. The model underneath is the Poisson process from Lesson 4.1.
| Market | Chance | Fair price |
|---|---|---|
| Home | 46.4% | 2.15 |
| Draw | 25.8% | 3.88 |
| Away | 27.8% | 3.60 |
| Over 2.5 | 48.2% | 2.08 |
| Under 2.5 | 51.8% | 1.93 |
Worked Betfair example
A match where the model expects 1.5 goals for the home side and 1.1 for the away side over 90 minutes. Goals are assumed to arrive at a steady rate, and stoppage time is ignored to keep the arithmetic clean. These are model fair prices. (Illustrative figures.)
Part 1: the drift at 0-0
| Minute (still 0-0) | Home | Draw | Away | Over 2.5 |
|---|---|---|---|---|
| 0 | 2.15 | 3.88 | 3.60 | 2.08 |
| 15 | 2.27 | 3.47 | 3.68 | 2.71 |
| 30 | 2.45 | 3.01 | 3.84 | 3.97 |
| 45 | 2.75 | 2.52 | 4.17 | 7.00 |
| 60 | 3.38 | 2.00 | 4.93 | 17.40 |
| 75 | 5.31 | 1.47 | 7.50 | 101.69 |
- Expected goals left. At 30 minutes, two-thirds of the match remains: home 1.5 × 60 ÷ 90 = 1.0, away 1.1 × 60 ÷ 90 = 0.73.
- The draw. Adding up the chance of every level score from here (0-0, 1-1, 2-2 and so on) gives 33.2%, a fair price of 3.01. At kick-off it was 25.8% (3.88).
- Over 2.5. Three or more goals from 1.73 expected gives 25.2%, a price of 3.97, nearly double its kick-off price of 2.08. Not a thing has happened on the pitch.
Part 2: the jump
- Home goal at 60 minutes. Just before, the draw is 2.00 (50.1%). After the market reopens at 1-0, the away side needs a goal just to draw: 16.9%, a price of 5.93. The home price jumps from 3.38 to 1.25.
- Over 2.5 after the goal. Two more goals are now needed in 30 minutes: 21.5%, a price of 4.64, in from 17.40.
Part 3: what a lay-the-draw trade is really betting on
At 30 minutes, 0-0, you lay the draw at 3.0 for £50 (liability £100).
- If the home side scores at 40 minutes, the model's draw price jumps to about 5.4. Green up (Lesson 8.1): back £50 × 3.0 ÷ 5.4 = £27.78. Locked result about +£22.22, or +£21.78 after 2% commission.
- If it's still 0-0 at 60 minutes, the draw has shortened to 2.00. Green up: back £50 × 3.0 ÷ 2.0 = £75. Locked result −£25.
- How often is that? With 1.73 goals expected in the last hour, the chance of no goal between 30 and 60 minutes is e^(−2.6 × 30 ÷ 90) = 42%.
Verdict: laying the draw is a bet that a goal comes soon. The clock is against you every minute, and a goal is the only thing that pays. Whether it's a good trade depends on whether your goal expectation beats the one in the price, just like any other bet.
The formula
Expected goals remaining
- λ is a side's expected goals over the full 90 minutes.
- t is the minute now.
In plain English: at a steady rate, the goals still to come shrink in a straight line as the clock runs down.
The chance of any final score
- λ_h and λ_a are the home and away expected goals remaining.
- i and j are the extra goals each side scores from now, so the final score is (current home goals + i) to (current away goals + j).
In plain English: treat each side's remaining goals as a Poisson count, multiply the two, and add up every final score that settles your market. The price is 1 ÷ that probability.
The chance of no goal in the next few minutes
- λ_h + λ_a is the total expected goals for the full match.
- m is the number of minutes ahead.
In plain English: the longer the window and the higher the scoring rate, the less likely a quiet spell. This is the drift you're betting against if you lay the draw.
Try it
Set home 1.5 and away 1.1 and drag the clock from 0 to 75 minutes at 0-0: watch the draw shorten and Over 2.5 drift. Then press "Home goal" at 60 minutes and see the draw jump from 2.00 to 5.93.
Common mistakes
- Reading the drift as a trend. A draw shortening at 0-0 isn't momentum. It's the clock, and it's predictable (Lesson 8.4).
- Assuming a steady scoring rate. Real goal rates rise later in matches and shift with the score. The simple model is a starting point, and real prices already allow for this.
- Forgetting stoppage time. Minute 90 isn't the end. A model that stops at 90 prices late draws too short.
- Thinking you can exit before the jump. The market is suspended on a goal. A trade that needs you to get out "just before" the goal can't do it.
- Treating a green as proof of an edge. A trade that greened up after an early goal was a winning bet on a goal coming soon. Judge in-play trades over hundreds of matches, like any other strategy (Lesson 7.3).
In-play edges still need a real reason to exist: why good models still lose money.