Mistake #3: Paying for a Sweep Without Pricing It
A favorite to win a best-of-three series is not equally likely to win it 2-0. A match-winner bet accepts either 2-0 or 2-1; a favorite -1.5 map handicap normally needs the sweep. Moving to the second market because its decimal odds look more attractive changes the event you are buying. It does not improve the original forecast. The useful question is whether the probability of the required score justifies the offered price.
This guide uses hypothetical probabilities and results to explain that difference. It does not report a tracked sample of professional matches or claim a universal favorite failure rate. A -1.5 ticket can have positive expected value, and an underdog +1.5 ticket can have negative expected value. Neither sign follows from the team's reputation or from the handicap alone.
Check the event first: In a completed Bo3 with no awarded maps, the favorite's -1.5 wins only at 2-0. The opponent's +1.5 wins whenever that opponent takes at least one map. Check the operator's treatment of forfeits, retirements and map advantages before treating the two bets as exact complements.
1. Separate CS2 vetoes from Dota 2 drafts
In CS2, an event's veto procedure determines the maps, their order and the rights associated with selection. Do not assume the favorite always chooses the first map or the decider is equally favorable to both teams. Read the actual event rules and confirmed veto. The ESL July 2026 rules update also illustrates why an old map-pool list needs a date and an event boundary.
Dota 2 hero drafting is a different process. Teams do not alternate between a CS2-style pool of different maps. In Captains Mode the captain selects heroes, as described in FACEIT's lobby guidance. Drafts, sides, adaptation and roster availability can change game probabilities within a series. A CS2 argument about an opponent's chosen map cannot be copied into a Dota 2 prediction.
A team's recent map win rate is also not automatically its chance against today's opponent. Opponent strength, sample size, lineup, patch and starting side can differ. A 60% historical rate is an observation about a sample, not a rule that the team must take a map. Record how the forecast was formed and keep uncertainty visible instead of inventing a universal cutoff.
2. Calculate the sweep as a joint event
Let A be the favorite. The general identity is P(A wins 2-0) = P(A wins map 1) × P(A wins map 2 | A won map 1). Multiplying two unconditional map estimates additionally assumes independence. The second-map estimate may change after the first game's result, a substitution or new draft information. Use the conditional form when those dependencies are part of the forecast.
Hypothetical inputs, not observed team rates:
p1 = 0.74
P(A wins map 2 | A wins map 1) = 0.58
P(A 2-0) = 0.74 × 0.58 = 0.4292
Fair decimal odds = 1 / 0.4292 = 2.329916...
Offered odds = 1.85
Break-even probability = 1 / 1.85 = 54.054054...%
EV per unit = 0.4292 × 1.85 - 1 = -0.20598
EV = -20.598%The negative EV belongs to these assumed inputs and this price. It is not evidence that all favorites are bad handicap bets. Expected loss is an average over repeated comparable decisions under the model; a single ticket either wins or loses. Our series calculator helps explore map inputs, but its simplified assumptions do not validate a real forecast.
A match-win probability cannot replace the sweep probability. For example, under an independent-map model with probabilities 0.74, 0.58 and 0.65, A wins the series with probability 0.72924 while its sweep probability remains 0.4292. The difference comes from paths that go to the decider. A quoted match price alone does not identify those paths or the individual map probabilities.
3. The opposite side is not automatically value
If the completed-series outcomes are complementary, P(B +1.5) = 1 - 0.4292 = 0.5708. At hypothetical odds 1.60, that bet has EV = 0.5708 × 1.60 - 1 = -8.672%. Both the favorite -1.5 at 1.85 and the underdog +1.5 at 1.60 are negative under the same probability model. Bookmaker margin allows that outcome; value does not have to move from one side to the other.
On a narrow screen, scroll the table sideways to see every column.
| Hypothetical selection | Probability | Odds | EV |
|---|---|---|---|
| A -1.5 | 42.92% | 1.85 | -20.598% |
| B +1.5 | 57.08% | 1.60 | -8.672% |
| B +1.5 at another price | 57.08% | 1.96 | +11.8768% |
The final row changes the quote, not the outcome probability. It illustrates price sensitivity rather than recommending an available operator or market. Compare both prices at the same timestamp and with the same settlement terms. The overround is 1/O1 + 1/O2 - 1 for a complete two-way market; it is not a separately deducted commission and cannot be inferred from one price. There is no documented fixed margin range in this example. See handicap mechanics for the score conditions.
4. A constructed result set is not an audit
Consider an explicitly invented set of 200 completed series: A sweeps 89, wins 2-1 in 78, and loses the remaining 33. With $100 stakes, A -1.5 at 1.88 produces 89 × $88 - 111 × 3,268, or -16.34% of $20,000 turnover. B +1.5 at 1.96 produces 111 × $96 - 89 × 1,756, or +8.78%. These are accounting checks on chosen outcomes, not a backtest or observed strategy return.
To turn such a table into research, publish the event list, selection rule, actual accepted prices and timestamps, voids and a separate validation period. Choosing outcomes after seeing results can make almost any rule look convincing. Even a genuine historical profit would need uncertainty and execution costs before being used to justify future stakes. Keep the map and match winner distinction in the dataset.
Do not confuse average odds with average return. The constructed table uses a fixed price for every ticket, so multiplication is sufficient. With changing odds, calculate the profit of each winning bet and add losses separately. Substituting the average quote of all tickets can give a different result when high and low prices belong to different outcomes. A void is not a full-stake loss either.
A team can be the likely winner and still be a poor purchase at the offered price. Another can lose more often and have positive expected value at sufficiently high odds. These are model statements, not earnings guarantees. If a modest plausible change in the inputs reverses EV, keep both calculations. Do not turn a disputed point estimate into a confident recommendation because one version looks attractive.
5. A decision checklist
- Confirm Bo3 format, map advantages and settlement terms. A Bo5 -1.5 has different winning scores.
- Use the confirmed veto for CS2 or the relevant game and draft information for Dota 2. Do not assume the lower-ranked team chooses map 2.
- Estimate the required joint probability and test less favorable inputs. An arbitrary 68% map threshold or 5% buffer cannot replace this calculation.
- Compare the probability with 1/odds and calculate EV. Use the odds converter to check the arithmetic.
- Keep the original forecast and accepted quote in the ledger. If the probability cannot be defended, passing is a complete decision; there is no obligation to switch to +1.5.
