A soccer betting strategy you can test is a pricing method: estimate each side's expected goals, turn them into a chance for every scoreline, and bet only when a bookmaker's price is longer than your fair price. In this page's illustrative match, expected goals of 1.52 for the home side and 1.08 for the away side give the home win a 47.41% chance, a fair $2.11. So $2.15 sits just above that fair price, and $2.05 below it.
It uses football's own tools: expected goals (xG) for the inputs, the Poisson distribution for the scorelines, Asian lines for the bets and the 90-minute rule for settlement. It prices a result rather than predicting one, and a priced edge still loses often.
A soccer betting model in five steps
- Rate each side's attack and defence from expected goals over a long sample.
- Turn the ratings into expected goals for this match, adjusted for team news.
- Price every market from one scoreline grid, so your prices agree with each other.
- Take the margin out of the bookmaker's market and back a selection only when its price is longer than your fair price.
- Log each price you took beside the final price before kick-off, and judge the method on closing line value.
From xG to fair prices for one match
One simple rating, with illustrative figures: home expected goals = the home side's xG per home match x the away side's xG conceded per away match / the league's average home xG per match.
- Home side: 1.71 x 1.33 / 1.50 = 1.516, rounded to 1.52.
- Away side, from its xG per away match, the home side's xG conceded per home match and the league's away average: 1.12 x 1.06 / 1.10 = 1.079, rounded to 1.08.
The Poisson distribution then gives each side's chance of scoring 0, 1, 2 or more goals, and multiplying the two sides' chances gives every scoreline. Working from the rounded figures:
| Market | Chance | Fair price |
|---|---|---|
| Home win | 47.41% | $2.11 |
| Draw | 25.64% | $3.90 |
| Away win | 26.96% | $3.71 |
| Over 2.5 goals | 48.16% | $2.08 |
| Both teams to score | 51.60% | $1.94 |
| 1-1 | 12.19% | $8.20 |
Two of those need no grid. Over 2.5 = 1 - e^-2.60 x (1 + 2.60 + 2.60 x 2.60 / 2) = 1 - 0.07427 x 6.98 = 48.16%, where e is the constant 2.718 and 2.60 is the expected total. Both teams to score = (1 - e^-1.52) x (1 - e^-1.08) = 0.7813 x 0.6604 = 51.60%. The ratings, not the Poisson step, are where models differ and where they go wrong.
The draw: why a three-way market needs its own number
In a three-way market, neither side's price means anything until the draw has a number, because it takes its share from both. The model adds up every level score, 0-0 at 7.43%, 1-1 at 12.19%, 2-2 at 5.00% and so on, to give the draw 25.64%.
Check it against the market. Say one bookmaker offers the home side at $2.15, the draw at $3.50 and the away side at $3.45: 1 / 2.15 + 1 / 3.50 + 1 / 3.45 = 104.07%. Scaled down to 100% in proportion (the fair odds calculator calls it multiplicative), that is home 44.69%, draw 27.45% and away 27.85%.
Your draw is about 1.8 points lower. A model built on independent Poisson counts can give too few 0-0 and 1-1 results, so a low draw is the first number to doubt.
Moving that gap onto the draw, in proportion from each side, leaves the home side 46.25% and the away side 26.30%:
| Bet on the home side | Expected value, model's draw | Expected value, market's draw |
|---|---|---|
| 1X2 at $2.15 | 0.4741 x 2.15 - 1 = +1.9% | 0.4625 x 2.15 - 1 = -0.6% |
| Draw no bet (Asian 0) at $1.62 | 0.4741 x 0.62 - 0.2696 = +2.4% | 0.4625 x 0.62 - 0.2630 = +2.4% |
| Asian -0.25 at $1.90 | 0.4741 x 0.90 - 0.5 x 0.2564 - 0.2696 = +2.9% | 0.4625 x 0.90 - 0.5 x 0.2745 - 0.2630 = +1.6% |
The 1X2 edge disappears, while draw no bet barely moves: its fair price, 1 + loss chance / win chance, depends only on the ratio of the two sides' chances, which a proportional shift leaves alone.
Risk: Betting involves risk. Your chances come from a model with real error in it, and a +EV bet can still lose, and often does. There is no guarantee of profit. See responsible gambling for limits and support.
Asian quarter lines: less margin, a split stake
At the same bookmaker, the home side's -0.25 line is $1.90 against $1.98 on the away side at +0.25: 1 / 1.90 + 1 / 1.98 = 103.14%, against 104.07% for the 1X2. Add up each market before you compare, because a two-way line does not always carry less margin than the 1X2.
A quarter line splits your stake across the two lines beside it, so -0.25 puts half the stake on the level line and half on -0.5: a win pays both halves, a draw refunds one half and loses the other. Asian handicap explained settles every line. The split gives each line a fair price from your win, draw and loss chances:
| Home line | Fair price formula | Fair price | Offered | Expected value per $1 |
|---|---|---|---|---|
| 0 | 1 + loss / win | $1.57 | $1.62 | +2.4% |
| -0.25 | 1 + (half the draw + loss) / win | $1.84 | $1.90 | +2.9% |
| -0.5 | 1 / win | $2.11 | $2.15 | +1.9% |
| -0.75 | 1 + (draw + loss) / (win by 2 or more + half of win by 1) | $2.47 | $2.45 | -0.9% |
The model puts a win by exactly one goal at 23.47% and by two or more at 23.94%, so -0.75 works out at 1 + (0.2564 + 0.2696) / (0.2394 + 0.5 x 0.2347) = $2.47. That is the core of an Asian handicap strategy: one view of the match, priced on every line, then the line and the bookmaker with the best expected value. Here it is -0.25, but on the market's draw it would be the level line.
Risk: Betting involves risk. Paying less margin leaves your chance of winning where it was, and a quarter-line bet still loses in full on a defeat. See responsible gambling for limits and support.
The 90-minute rule in your model
The NSW Bookmaker Declared Betting Events Betting Rules cover bookmakers authorised there, and their clause 5.2.7.3 settles soccer bets on the score when normal time ends, stoppage time included, unless another basis was agreed when you bet. Other bookmakers set their own terms, so check yours. For your model, that means:
- Price normal time, stoppage time included. Goals in added time count, so they belong in your expected goals.
- Never add extra time to a match price. A cup tie level after 90 minutes is a draw for every market settled on normal time, as extra time and penalties shows.
- Allow for the tie, not just the match. A side protecting a first-leg lead may settle for a draw, which a rating built on league matches cannot see.
Team news and kick-off times in Australia
From 25 October 2026, when UK and European clocks go back, to 27 March 2027, Sydney, Melbourne, Canberra and Hobart run 11 hours ahead of London and 10 ahead of Central European time. Brisbane is an hour less, Perth three hours less and Adelaide half an hour less than Sydney.
So European kick-offs fall overnight in the east, and the Premier League said in August 2024 that each starting XI would be out some 75 minutes before kick-off: for a match starting at 7 am in Sydney, about 5:45 am. EPL betting from Australia converts each kick-off slot.
Price the evening before and you carry the line-up risk; wait, and you take whatever price is left. The risk is measurable. Say the home side's leading striker is ruled out and you take 0.14 off its expected goals, from 1.52 to 1.38:
| Before the news | After the news | |
|---|---|---|
| Home win chance | 47.41% | 43.73% |
| Fair 1X2 price | $2.11 | 1 / 0.4373 = $2.29 |
| Fair draw no bet price | $1.57 | 1 + 0.2939 / 0.4373 = $1.67 |
| Expected value at $2.15 | +1.9% | 0.4373 x 2.15 - 1 = -6.0% |
If $2.15 is still on offer after the news, leave it.
Testing the method against closing prices
A few weeks of results prove little, because one goal swings a match. The quicker test is closing line value: your price / the closing price - 1. Take $2.15 the evening before on a side whose price at kick-off is $2.02, and you beat the close by 2.15 / 2.02 - 1 = 6.4% before any margin is removed.
Closing line value covers removing the margin from the close, and positive EV betting covers expected value and staking. A model that keeps beating a margin-free close over a few hundred bets is adding something.
Risk: Betting involves risk. Beating the close is evidence about your prices, not a promise of profit, and past results do not predict future results. See responsible gambling for limits and support.
Mistakes that cost money
| Mistake | What it costs in the worked match |
|---|---|
| Ratings from a short run of goals | Rate the home side at 1.72 instead of 1.52 and $2.15 shows a false +12.6% edge (52.35% x 2.15 - 1) |
| Trusting the plain model's draw | The home side's +1.9% at $2.15 becomes -0.6% on the market's draw |
| Comparing prices on different lines | On the market's draw, $1.90 at -0.25 is +1.6% and $2.15 at -0.5 is -0.6%: only the first refunds half on a draw |
Soccer betting maps the markets this method can price.
Checking your prices against B337's screeners
B337's +EV screener runs the same comparison from the market's side. It works out a fair price for each selection from the market itself, removing the margin from prices across it, exchanges and global market references included, then lists the bookmaker prices above that fair price, ranked by edge. You can change how that fair price is built, and it is an estimate, so set your own model's price beside it.
Terminal Pro and Full Automation include the rows, and a Screeners plan offers them alone, set up with the team and with its price confirmed before you pay. On a free account you see the live count of prices above fair, not the rows. You place each bet yourself, in bookmaker accounts you hold, and any bet placed through B337 uses credits. The board collects prices in cycles, so confirm each one with your bookmaker.
For free and confidential support call 1800 858 858 or visit gamblinghelponline.org.au.
Risk: Betting involves risk. The screener's fair price and your model's are both estimates, a listed price can move before you bet, and a bookmaker can cap your stakes or void a bet. There is no guarantee of profit. See responsible gambling for limits and support.