Asian Handicap Fair Odds Calculator

Asian Handicap prices are probabilities of goal-margin settlement states, not simply probabilities of winning a match. A Poisson score matrix or its Skellam goal-difference distribution provides those probabilities. Fair odds then depend on whether the selected line can produce a full win, half win, push, half loss or full loss.

In one minute
  • −0.5 is equivalent to backing the selected team to win.
  • 0.0 is draw no bet: a draw returns the stake.
  • +0.5 wins when the selected team wins or draws.
  • Quarter lines split the stake equally between adjacent half- and whole-goal lines.
  • The difference of independent Poisson goal counts follows a Skellam distribution.
  • Fair odds equal 1/P(win) only when no push or partial settlement is possible.
  • Dixon-Coles changes the probabilities of goal margins −1, 0 and +1.
  • Fair odds are conditional on the supplied λ values and chosen score model.

What is an Asian Handicap?

An Asian Handicap adds a virtual goal adjustment to the selected team before the bet is settled. The adjustment removes or reduces the draw as a betting outcome.

Define the selected team’s regulation-time goal margin:

D = selected team goals − opponent goals

For a handicap h, the adjusted margin is:

Dadjusted = D + h
Adjusted margin > 0 The corresponding line wins.
Adjusted margin = 0 The corresponding line pushes.
Adjusted margin < 0 The corresponding line loses.

If the home team is selected at −0.5, it must win the match. If it is selected at +0.5, a win or draw is enough.

The sign is from the selected team’s perspective. Home −0.75 and Away +0.75 are opposite sides of the same handicap market. Do not apply the same sign to both teams.

Half, whole and quarter-goal handicap lines

Selected-team line Equivalent bet Important settlement state
−0.5 Selected team to win A draw is a full loss
0.0 Draw no bet A draw is a push
+0.5 Selected team or draw A one-goal loss is a full loss
−1.0 Selected team gives one goal A one-goal win is a push
+1.0 Selected team receives one goal A one-goal loss is a push
−0.25 Half stake on 0.0, half on −0.5 A draw is a half loss
+0.25 Half stake on 0.0, half on +0.5 A draw is a half win
−0.75 Half stake on −0.5, half on −1.0 A one-goal win is a half win
+0.75 Half stake on +0.5, half on +1.0 A one-goal loss is a half loss

General quarter-line split

A line ending in .25 or .75 splits into the two adjacent half-goal increments:

h = 0.5 × lower component + 0.5 × upper component

Examples:

  • −1.25 = half on −1.0 and half on −1.5;
  • −0.75 = half on −0.5 and half on −1.0;
  • +0.25 = half on 0.0 and half on +0.5;
  • +1.75 = half on +1.5 and half on +2.0.

Asian Handicap fair-odds calculator

Enter home and away λ, select a team and handicap line, and optionally apply Dixon-Coles. The handicap sign is always from the selected team’s perspective.

Goal rates to handicap fair odds

Model Full win Half win Push Half loss Full loss Fair odds
Goal margin for selected team Independent Poisson Dixon-Coles

The margin table shows the probabilities most relevant to common central handicap lines. The calculator uses a wider internal distribution with at least goals 0–6 per team and expands until each omitted Poisson tail is below approximately 10−13. Small positive probabilities use scientific notation. Use goal rates in (0, 10], decimal points or commas, and ρ strictly inside its computed bounds. Prices exclude bookmaker margin and fees.

From expected goals to the goal-difference distribution

Under independent Poisson:

H ~ Poisson(λhome),    A ~ Poisson(λaway)

The complete score matrix is:

P(H = i, A = j) = Pois(i; λhome) × Pois(j; λaway)

Summing every score with the same difference gives the distribution of:

D = H − A

The Skellam distribution

The difference between two independent Poisson variables follows a Skellam distribution:

P(D = k) = e−(λh+λa) (λh/λa)k/2 I|k|(2√(λhλa))

I|k| is the modified Bessel function of the first kind. In practical software, summing the score matrix is often simpler, especially when Dixon-Coles or another joint model is used.

Skellam is a consequence of independent Poisson. Once the two team scores are dependent or have non-Poisson marginals, their difference is not generally Skellam.

The underlying models are compared in Poisson vs Dixon-Coles vs bivariate Poisson.

The universal Asian Handicap fair-odds formula

Let the selected line have probabilities:

pFW Full win probability.
pHW Half win probability.
pP Push probability.
pHL Half loss probability.
pFL Full loss probability.
Decimal odds O Total return per unit stake on a full win.

At decimal odds O, expected net profit on one unit is:

E[profit] = (O − 1)(pFW + 0.5pHW) − (pFL + 0.5pHL)

Set expected profit to zero and solve for O:

Ofair = 1 + [pFL + 0.5pHL] / [pFW + 0.5pHW]

Why 1/P(win) is often wrong

The shortcut 1/P(win) works only when every non-win is a full loss. It fails when a draw or exact margin returns some or all of the stake.

Line type Fair-odds formula
Half-goal line, no push 1 / P(full win)
Whole-goal line 1 + P(full loss) / P(full win)
Quarter-goal line Use full/half settlement formula above

Worked example: λ home 1.70, λ away 1.10

Use the same goal rates across all lines:

λhome = 1.70,    λaway = 1.10

Under independent Poisson:

Home handicap Fair odds: Poisson Fair odds: Dixon-Coles ρ = −0.03 Key settlement margin
−1.0 2.744 2.756 Win by exactly one pushes
−0.75 2.226 2.240 Win by exactly one is a half win
−0.5 1.945 1.958 Any home win is a full win
−0.25 1.712 1.717 A draw is a half loss
0.0 1.478 1.475 A draw pushes
+0.25 1.388 1.382 A draw is a half win
+0.5 1.326 1.320 Home win or draw is a full win
+0.75 1.225 1.222 Lose by exactly one is a half loss
+1.0 1.1241 1.1235 Lose by exactly one pushes

Home +1.0 is shown to four decimals because the Dixon-Coles shift is smaller than the usual rounding: 1.1241 vs 1.1235. Full cancellation of the correction starts on wider lines — at these rates, Home +1.5 is identical under both models because the −1, 0 and +1 goal margins all settle the same way.

Detailed example: Home −0.75

Home −0.75 splits into Home −0.5 and Home −1.0.

Actual home margin −0.5 half −1.0 half Combined settlement
Home wins by 2+ Win Win Full win
Home wins by exactly 1 Win Push Half win
Draw or away win Loss Loss Full loss

The Poisson settlement probabilities are:

Full win Half win Full loss
27.870% 23.532% 48.597%
Ofair = 1 + 0.485974 / (0.278704 + 0.5 × 0.235322) = 2.226

Detailed example: Home +0.25

Home +0.25 splits into Home 0.0 and Home +0.5:

  • home win: full win;
  • draw: half win;
  • away win: full loss.
Ofair = 1 + P(Away win) / [P(Home win) + 0.5P(Draw)] = 1.388

Special Asian Handicap lines and 1X2 probabilities

Let H, D and A be fair home-win, draw and away-win probabilities.

Home −0.5

No push is possible:

O(Home −0.5) = 1 / H

Home +0.5

O(Home +0.5) = 1 / (H + D)

Home 0.0: draw no bet

O(Home 0) = 1 + A/H = (H + A)/H

Home −0.25

A draw is a half loss:

O(Home −0.25) = 1 + [A + 0.5D]/H

Home +0.25

A draw is a half win:

O(Home +0.25) = 1 + A/[H + 0.5D]

These central lines can be priced from fair 1X2 probabilities alone. Lines such as −0.75, −1.0 or +1.0 require exact goal-margin probabilities, not only Home/Draw/Away totals.

How to price quarter-goal lines correctly

Quarter lines combine two bets at the same quoted decimal odds. The probabilities of half-win and half-loss states must therefore be weighted by 0.5.

Selected team −0.25

Result 0.0 half −0.5 half Settlement
Selected team wins Win Win Full win
Draw Push Loss Half loss
Selected team loses Loss Loss Full loss

Selected team +0.75

Result +0.5 half +1.0 half Settlement
Win or draw Win Win Full win
Lose by exactly 1 Loss Push Half loss
Lose by 2+ Loss Loss Full loss

Use the combined expected settlement to price quarter lines. The arithmetic average of component fair odds is generally wrong. It agrees only in the special case where both components have equal winning-stake probability weights. The full/half-win and full/half-loss formula above works without relying on that exception.

How Dixon-Coles changes Asian Handicap fair odds

The correction parameter is not a free number. For the calculator, all four τ multipliers must remain strictly positive, which bounds ρ:

max(−1/λhome, −1/λaway) < ρ < min(1/(λhomeλaway), 1)

For the worked rates λhome = 1.70 and λaway = 1.10 the strict interval is approximately −0.588 < ρ < 0.535; the calculator enforces the corresponding dynamic bounds for the entered rates.

Dixon-Coles reallocates probability among:

  • 0-0 and 1-1, both with goal difference 0;
  • 1-0, with home goal difference +1;
  • 0-1, with home goal difference −1.

Therefore, at fixed λ, it changes the goal-difference probabilities at −1, 0 and +1 while preserving total probability and each team’s marginal goal distribution.

Negative ρ in the worked example

With ρ = −0.03:

  • draw probability increases;
  • one-goal home and away win probabilities decrease;
  • larger exact goal margins remain unchanged directly.

This explains why Home −0.5 becomes slightly less likely and its fair odds rise from 1.945 to 1.958, while Home +0.5 becomes slightly more likely and its fair odds fall from 1.326 to 1.320.

Dixon-Coles does not change every handicap line. If all three affected goal margins settle identically on a sufficiently wide line, their net probability transfer can cancel. Central lines are generally the most sensitive.

The correction is derived in the Dixon-Coles football model guide.

How to compare fair Asian Handicap odds with market prices

Asian Handicap books usually quote two opposing sides, such as Home −0.75 and Away +0.75. Convert both odds to raw implied probabilities and remove the margin before comparing the market’s probability with the model.

Example two-way market

Selection Offered odds Reciprocal price
Home −0.75 2.12 47.170%
Away +0.75 1.80 55.556%
Booksum 102.725%

Using proportional normalization — the current ImpliedScore de-vig baseline — produces a de-vigged reciprocal price share for Home −0.75 of about 45.918%. That share is an effective break-even settlement weight, not a full-win probability, because the line contains partial settlement states.

Compare odds or expected settlement—not raw “win probability”

The cleanest comparison is:

  • model fair decimal odds from settlement probabilities;
  • market de-vigged fair decimal odds under the same quarter-line settlement;
  • expected return at the actually offered price.

For model settlement probabilities and offered decimal odds O:

EV = (O−1)(pFW + 0.5pHW) − (pFL + 0.5pHL)

This formula respects half settlements. Treating a half win as a full win overstates expected return.

Different methods for removing the market overround are compared in No-Vig Calculator and margin-removal guide.

Can Asian Handicap odds identify expected goals?

One handicap market gives one independent fair-price relationship. It is insufficient to identify both λhome and λaway.

What an AH line mainly constrains

Asian Handicap is primarily sensitive to the goal-difference distribution:

D = Home goals − Away goals

Match totals mainly constrain:

Λ = λhome + λaway

Combining one total market with one handicap market can identify or numerically fit the two rates under a specified score model.

Why a quoted main line contains extra information

The handicap value itself—such as Home −0.75 rather than −0.25—summarizes the bookmaker’s view of team imbalance. The two prices at that line then describe the balance around the selected settlement threshold.

Line and price are both data. Using only odds 1.90/1.90 without recording whether the handicap was −0.25, −0.75 or −1.5 discards the main directional information.

ImpliedScore currently fits λ from 1X2 and Over/Under. Asian Handicap can be added as an alternative or additional target, but more targets do not guarantee exact compatibility with a two-parameter Poisson matrix.

How to validate Asian Handicap probabilities

  1. Freeze the full market before kickoff. Store line, both prices, source, timestamp, λ values and model.
  2. Record regulation-time goal margin. Do not substitute advancement after extra time or penalties.
  3. Apply the exact settlement rule. Store full win, half win, push, half loss or full loss.
  4. Score expected return, not only hit rate. A half win and full win are economically different outcomes.
  5. Compare identical lines. Home −0.5 and Home −0.75 should not be pooled as the same binary event.
  6. Use paired model comparisons. Test Poisson, Dixon-Coles and market-implied prices on the same fixtures.
  7. Separate probability fitting from betting results. Short-run profit can be dominated by variance and offered-price selection.
  8. Check central and wide lines separately. Dependence corrections can affect them differently.

Useful evaluation quantities

  • average realized net return at model fair odds;
  • calibration of full-win and full-loss components;
  • frequency of exact push margins;
  • paired loss or likelihood of the full score distribution;
  • closing-line comparison after margin removal.

A model specifically developed for the Asian Handicap market should be evaluated against the settlement structure, not only against Home/Draw/Away accuracy.

Limitations of Poisson Asian Handicap pricing

Expected goal rates can be wrong

Every handicap probability inherits errors in λ from historical ratings, projected xG or market inversion.

Skellam requires independent Poisson goals

Goal dependence, non-Poisson dispersion and changing game state can alter the margin distribution.

Dixon-Coles is local

It changes only score cells corresponding to margins −1, 0 and +1. It does not directly correct broader tail behaviour.

Market settlement rules vary

In-play Asian lines can reset the effective score at bet placement on some platforms. Full-time pre-match models must not be applied to such a market without matching its rules.

Quarter-line fair odds are model-sensitive

A small change in exact one-goal-margin probability can alter half-win or half-loss frequency, even when overall match-win probability changes little.

Two-way de-vigging is not unique

Proportional, additive, power and other methods can imply different market fair prices.

Pre-match probabilities are not live probabilities

Elapsed time, goals and red cards change the remaining goal-margin distribution.

Structural model risks are discussed in where Poisson football models fail.

Practical conclusion

  • Build or derive the complete goal-difference distribution.
  • Split quarter lines into their two component handicaps.
  • Use full and half settlement weights in the fair-odds formula.
  • Use 1/P(win) only on lines where every non-win is a full loss.
  • Expect Dixon-Coles to affect central one-goal and draw-sensitive lines most.
  • Record both the handicap line and its prices when fitting from market data.
  • Validate with exact settlement outcomes on frozen out-of-sample matches.

References

  1. Skellam, J. G. (1946). The Frequency Distribution of the Difference between Two Poisson Variates Belonging to Different Populations . Journal of the Royal Statistical Society, 109(3), 296.
  2. Maher, M. J. (1982). Modelling Association Football Scores . Statistica Neerlandica, 36(3), 109–118.
  3. Dixon, M. J. & Coles, S. G. (1997). Modelling Association Football Scores and Inefficiencies in the Football Betting Market . Journal of the Royal Statistical Society: Series C, 46(2), 265–280.
  4. Constantinou, A. C. (2021). Investigating the Efficiency of the Asian Handicap Football Betting Market with Ratings and Bayesian Networks . Journal of Sports Analytics, 7(4), 307–321.
  5. Betfair Help. Exchange: What Is Asian Handicap Betting? . Settlement examples for whole, half and quarter lines.
  6. ImpliedScore. Market-Implied Goals from Betting Odds .
  7. ImpliedScore. Poisson vs Dixon-Coles vs Bivariate Poisson (linked above).

Educational material only. Asian Handicap probabilities and fair odds do not guarantee results or profit. Betting involves financial risk and is intended only for adults in jurisdictions where it is permitted.

Scroll to Top