A map handicap is a different event from the series moneyline, not a safer substitute at every price. The definitions below apply to ordinary completed series without a starting map advantage. Check the event format and settlement clauses before applying them. A higher cover probability can be offset by a shorter quote; no universal +1.5 advantage follows from the extra winning scoreline. The useful starting point is a complete terminal-score tree with one probability assigned to each mutually exclusive outcome.
1. The Mechanics of Map Spreads: What You Are Actually Betting
A map handicap adjusts the final series score by a fixed number of maps (in Dota 2 and LoL, "games"). The half-point removes the push. Before you price anything, make sure you know exactly which scorelines win and lose.
- Bo3, Favourite -1.5: wins only on a 2-0 sweep. A 2-1 series win is a loss on this ticket.
- Bo3, Underdog +1.5: wins on any scoreline except a 0-2 loss. That means underdog 2-0, underdog 2-1, or underdog 1-2. Formally, P(+1.5) = 1 − P(Favourite Sweep).
- Bo5, Favourite -1.5: wins on 3-0 or 3-1. Loses on 3-2 and on any series loss.
- Bo5, Underdog +1.5: wins if the underdog takes the series or loses 2-3. In practice you are betting that the series reaches a fifth map, or better.
- Bo5, Favourite -2.5: wins only on a clean 3-0.
- Bo5, Underdog +2.5: wins if the underdog takes at least one map. P(+2.5) = 1 − P(Favourite 3-0).
Constant probabilities as a teaching baseline
Assume independent maps with the same favourite probability p. In Bo3, the favourite sweep probability is p². In Bo5, a favourite −1.5 ticket wins on 3-0 or 3-1, with probability p³ + 3 p³(1−p). The three placements of the lost map explain the coefficient 3. The series-win probability also includes six 3-2 paths. Those assumptions simplify the tree; they do not establish realistic map strength for an actual final. In particular, winning a series and covering a spread remain separate quantities even when both favour the same team.
Hypothetical independent maps, favourite perspective
Bo3 p=.65; q=.35
P(Fav2-0)=p^2=.4225
P(Fav2-1)=2*p^2*q=.29575
P(Favseries)=.71825
P(Dog+1.5)=1-.4225=.5775
EV(Dog+1.5 at1.68)=.5775*1.68-1=-.0298
Bo5 p=.62; q=.38
P(Fav3-0)=.238328
P(Fav3-1)=3*p^3*q=.27169392
P(Fav3-2)=6*p^3*q^2=.2064873792
P(Favseries)=.7165092992
P(Fav-1.5)=.51002192
P(Dog+1.5)=.48997808
P(Dog+2.5)=.761672Map-specific strength and the complement
Now assume the favourite has p1 = 0.70, p2 = 0.55 and p3 = 0.62. Under independence its sweep chance is 0.385, so the opponent +1.5 covers with probability 0.615. The decider does not affect this particular handicap, since the ticket has already won if the series reaches 1-1. It does affect the opponent’s chance of winning the series, 0.3174. At hypothetical prices 1.68 for +1.5 and 3.10 for the moneyline, EV is +3.32% and −1.606% respectively. Different quotes can reverse that comparison.
If maps are dependent, the sweep uses P(W1)P(W2|W1), not two unconditional averages. A shared form shock can change both maps, and draft rules may alter later choices. Build those branches explicitly rather than subtracting a universal correlation haircut. Nor should you force Bo5 probabilities toward 50% on every map: fatigue and adaptation require event-specific evidence and may act in different directions. A personal estimate of a comfort pick is a model input, not a guaranteed minimum win rate.
Synthetic coverage requires disjoint states
In an ordinary completed Bo3, the underdog +1.5 consists of underdog 2-0, underdog 2-1 and favourite 2-1. The events are disjoint, so equal-return dutching can reproduce the handicap. With hypothetical prices 7.50, 5.80 and 3.60, the reciprocal sum is 0.583524904215 and the synthetic decimal price is 1.713722915299. This is better than a direct 1.68 only if the intended stakes are accepted and settlement is compatible. A second construction, underdog moneyline plus favourite 2-1, can have a worse combined price. Equivalent outcomes do not imply equivalent costs.
Hypothetical dutching, total stake100
S=1/7.50+1/5.80+1/3.60=.583524904215
syntheticPrice=1/S=1.713722915299
stake_i=100*(1/odds_i)/S
stakes=22.849638870650;29.546946815496;47.603414313854
return in each covered state=171.372291529875
loss if favourite2-0=-100
EV at pCover=.615: .615*1.713722915299-1=.053939592909
Two-leg alternative:1/(1/3.10+1/3.60)=1.665671641791The losing state still loses the full total stake. This is a synthetic handicap, not an arbitrage across every outcome. A rejected or void leg also changes the intended payoff tree. Inspect the case where only one accepted ticket remains, rather than assuming the worst leg can always be filled first at the intended price. Commission on an exchange and fixed transfer costs need their own cash-flow treatment. Do not count potential returns as available bankroll before settlement or use borrowed funds to complete an uneven set of positions.
Price, risk and an auditable decision
A single binary ticket with cover probability c and decimal price d has net-return variance d²c(1−c) per unit staked. A higher cover chance alone does not determine variance; the price and size matter. Compare the same total stake, then examine joint loss across related series positions. Fractional Kelly reduces the stake from the single-bet optimum under a chosen probability, but it cannot validate the probability or guarantee a drawdown bound. Choose any personal cap before the result, not as a way to force an appealing ticket into the plan.
For a reproducible review, retain event rules, veto or draft scenarios, conditional map inputs, full market quotes, source times, costs and accepted stakes. Test the sum of terminal probabilities and the complement identities before calculating EV. A historical finals table needs an explicit list of events and outcomes; none is presented here as measured evidence. A sweep in a famous final is not a general base rate. Keep CLV, calibration and realised profit separate. The decision may be direct handicap, a properly priced synthetic, another market or no position.
When auditing the tree, distinguish the chance that a map is played from the chance that a team wins it if played. A Map 3 estimate in Bo3 is conditional on a 1-1 state; it should not be added as an independent terminal outcome beside 2-0. In a Bo5, enumerate paths until either team reaches three wins, then stop. Potential fourth and fifth maps may be generated mathematically if unused outcomes are marginalised and each terminal score is counted once. The error is treating those maps as played or counting a terminal outcome again. Apply the handicap to each final score and sum the corresponding weights. The same procedure handles a hypothetical starting advantage, but that is a different contract and needs a new tree. Check also integer spreads, which can introduce a push and therefore no longer follow the simple binary EV formula. A walkover, withdrawal or replay can invoke special settlement outside the normal score tree. Preserve those clauses in the record instead of silently treating every administrative result as a played sweep. If the rule is unclear, mark the contract unevaluable before staking.
