When a tournament changes your view of a hero
A hero wins nine of eleven maps and suddenly appears in every draft discussion. That record deserves inspection, but it does not tell you how many percentage points to add to the next team that picks the hero. Strong teams may be selecting it, favorable matchups may explain the wins, or the same specialist may account for much of the sample. Start by identifying those cases before calling the hero overpowered.
Tournament adaptation can change what teams prioritize. It need not shrink every event to a fixed number of heroes. Different formats, participants and patches produce different paths. Count what happened in the event under review rather than importing a claim that group stages always contain more than ninety heroes or playoffs always revolve around twenty-five. A cumulative count also cannot be compared directly with a count from a much shorter playoff window.
Measure priority with a consistent denominator
For a hero, record picks, bans, first-phase actions and maps in which it was available. Define the denominator explicitly. A pick-plus-ban rate measures how often a hero is contested, not how often it wins when selected. With ten maps and six drafts containing either a pick or ban, the contest rate is 60%. Calling that 60% win equity would confuse two different quantities. Compare equal windows and keep patch changes separate.
You may define a working priority pool as heroes contested in at least 20% of the last thirty maps. That is an analyst's chosen threshold, not an established tournament statistic. Moving the threshold to 15% or changing the window can alter the count substantially. Show that sensitivity. Also record which teams remain in the tournament: elimination of a specialist can reduce a hero's presence without revealing anything new about the hero's inherent strength.
Shrink the record without inventing a team edge
Use a hypothetical Beta prior with mean 0.50 and total weight twenty. Nine wins from eleven maps then produce posterior mean (9 + 10)/(11 + 20) = 19/31, approximately 61.29%. Twenty-two wins from thirty-four produce (22 + 10)/(34 + 20) = 32/54, approximately 59.26%. These are outputs of the chosen prior and simplified sampling model. The prior weight is not claimed to be empirically optimal.
Neither posterior gives the causal benefit of drafting that hero. Selection is not random. Team strength, player experience, role, opponents, other heroes and side all affect the result. Multiplying the excess over 50% by an invented “draft dilution” constant would create a precise-looking team adjustment without evidence. Compare models with and without hero terms on later events, and keep the estimated effect separate from raw and shrunk win percentages.
Patch history matters to both availability and interpretation. Valve's 7.41 changes removed Facets, so a draft record split by a former Facet cannot be applied as a current selection rule. Retain the original version alongside historical data. A hero's old role or skill interaction may also differ. Reading patch notes tells you which hypotheses need retesting; it does not quantify the new competitive effect by itself.
Format changes are conditional branches
A Bo2 can finish 2–0, 1–1 or 0–2. Under independent maps with constant assumed p = 0.60, the probabilities are 0.36, 0.48 and 0.16. Now make a different hypothetical assumption: after winning the first map, the favorite's second-map probability is 0.52; after losing the first, it remains 0.60. The outcomes become 0.312, 0.528 and 0.16. Both sets sum to one, but they describe different models.
The second example does not demonstrate that teams experiment after a first-map win. That needs observations and tournament context. Standings incentives, advancement rules, rest and the opposing draft may point in different directions. Record the reason for a conditional adjustment before the next map. Do not apply a universal discount to all group-stage winners or increase stakes automatically when the bracket reaches elimination play.
Treat workload as evidence to investigate
An event schedule can put consecutive series on one day. Record actual maps played, elapsed play time, breaks, travel and substitutions. Those facts establish workload, not a fixed performance penalty. A later error can have many causes. Labeling it fatigue after seeing the loss is weaker than making a testable forecast beforehand. Do not infer a universal late-game error increase from a team's lower-bracket position.
For sensitivity only, the independent constant-map Bo5 function gives 0.593126875 at p = 0.55 and 0.5374600192 at p = 0.52. The series difference is about 5.57 percentage points. This is a comparison of two fixed-input scenarios, not the correct calculation for a penalty that begins only on later maps. A map-dependent fatigue hypothesis requires probabilities for the actual sequence and conditional paths.
Check adoption and the available price
Log when a strategy appears, who copies it, which counters are tried and whether the originating team can play alternatives. Rising bans may indicate perceived danger, but they also reflect opponent preparation and opportunity cost. There is no documented universal 24-to-48-hour delay before sportsbooks respond. Use time-stamped prices from the same market to test a specific delay hypothesis. A line may have incorporated the idea before you saw the draft.
Your tournament sheet should therefore distinguish observations, model adjustments and unresolved questions. Update the sheet when new maps arrive, including evidence against the initial story. A favorite losing twice is a prompt to inspect opponents and drafts, not a mandatory three-point penalty. An underdog introducing a new idea is worth studying, not automatically backing. The next draft is the place to test whether the supposed discovery remains available and relevant.
Check the information set before the price
For a live assessment, put the game clock beside the time the information reached you. A broadcast frame, a public scoreboard and an available betting price can describe different moments. Record the source of each observation and keep an unknown field unknown. If the price has already incorporated a fight that your video has not shown, a neat calculation using that video is evaluating an earlier game state. Slow-moving inputs also need this check: a player can spend gold, change position or lose an objective between updates.
Keep a short decision record: event and map, ruleset, roster, relevant state, probability range, quoted odds and observation time. Write the reason before the result arrives. Include the cases you rejected, because studying only placed bets hides selection mistakes. A useful review asks which observation would have changed the decision and whether that observation was actually available at the time. An explanation invented after a win is not a forecast.
Test the estimate, including the decision to pass
Expected profit per unit at decimal odds O is p × O − 1, provided the bet has only win and loss outcomes. Settlement rules can add pushes, voids or partial payouts, so read them before applying that equation. The break-even probability is 1/O. Removing an overround from the offered prices produces a market reference under a chosen allocation method; it does not reveal the true probability. A disagreement with that reference is a research question until the independent estimate has supporting evidence.
Use later matches to test parameters fitted on earlier matches. Split by event or time rather than randomly scattering snapshots from the same map across training and testing. Otherwise the model can recognize a match it has effectively seen already. Check calibration, probability scoring and realized returns with uncertainty, alongside changes in patches and rosters. Closing-line movement is another diagnostic, but a favorable move alone cannot establish accurate probabilities or sustained profitability. When a plausible alternative assumption reverses the expected value, passing is a complete decision. A smaller stake changes exposure; it does not repair a faulty probability.
Related models and reading
These pages explain related methods. Their models do not independently validate the numerical assumptions in this article.
- Dota 2 betting strategy guide
- permaban and hero priority analysis on EsportsMath.org
- Series Win Probability Calculator
- first blood and 10 kills props
- tournament meta and form regression on EsportsMath.org
- draft phase win probability
Version context and organizer references
These official references provide patch or event context. They do not establish the hypothetical model coefficients, a current buyback cost formula or a complete ruleset for every event. Check the active client and the specific event rules for operational details.
