Esports prediction markets: games, formats and analysis

What this hub covers

Esports prediction markets let you take a position on an outcome like a series winner, a single map winner or a handicap. They are not all the same object. The game title, the tournament format and the individual market rules decide what actually resolves; confusing them can make the analysis answer a different question.

This hub maps those moving parts and routes you to game-specific guides: CS2 maps, veto and lineup, VALORANT map plus side and agents, Dota 2 draft, patch and lineup, and League of Legends draft, split and patch. There is also an ELO rating guide for the method behind rating systems. This page gives methods, not match tips or live odds.

Events, markets and outcomes are separate objects

An event on a prediction market platform groups one or more markets, and the event, the condition and the outcome token identifiers are all distinct. Tags can overlap instead of forming a strict hierarchy. That matters when a single tournament page holds several markets, for example a series winner and a correct-score market that look similar but resolve differently.

Check the active, closed and acceptingOrders flags on each individual market rather than trusting the parent event. A live event can contain one market that is already closed and another that still accepts orders. If you filter only by the event page, you may analyze something you cannot actually trade.

Resolution rules state the source, the deadline and the edge cases, so the title alone is not enough. Ordinary binary markets pay 1 to the winning side and 0 to the losing side. In a rare unknown or 50-50 resolution both sides can receive 0.50, but that is not a universal cancellation or refund, and you should not assume it.

Series format is part of the probability

A best-of-3 (BO3) series ends when one team reaches 2 map wins; a best-of-5 (BO5) ends at 3. A single map win rate is not a series win rate. If a team wins 60% of its maps, that does not mean it wins 60% of its series, because the number of maps and the order of wins both matter.

Take an illustrative team with a 60% chance to win any given map, treating maps as independent and identical. In a BO3 it wins the series by taking the first 2 maps, or by splitting the first 2 and winning the decider: 0.6 x 0.6 + 2 x (0.6 x 0.4) x 0.6 = 0.36 + 0.288 = 0.648, about 64.8%. A 60% map edge becomes roughly 64.8% at series level.

The same 60% map rate in a BO5 gives a different number because the series can end in 3, 4 or 5 maps: 0.6^3 + 3 x 0.6^3 x 0.4 + 6 x 0.6^3 x 0.4^2 = 0.216 + 0.2592 + 0.20736 = 0.68256, about 68.3%. The map rate did not change. The format did.

A concrete decision when formats differ

Suppose a tournament page lists 2 markets on the same matchup, one on a BO3 series winner and one on a BO5 series winner. Under the same illustrative 60% map rate, the BO3 number is about 64.8% and the BO5 number is about 68.3%. These are outputs of a stated model, not observed results or forecasts.

The decision is to confirm the scheduled format from the event rules before you compare anything. If the rules say BO3, the BO5 figure describes the wrong object; if the rules say BO5, do not reuse the BO3 figure. This is a rule-matching step, and it will not tell you whether either price is attractive.

That protocol is author-designed and hypothetical. It assumes independent maps at a constant 60% map win rate, which is a simplification. Real map pools, vetoes, side selection and roster changes break that independence, so the BO3 and BO5 numbers are a way to see the format effect, not a forecast of any match.

Where each game guide applies

Each title has its own variables. The CS2 maps, veto and lineup guide covers map pools, the veto sequence and roster questions. The VALORANT map and side agent guide covers map plus side and agent composition. The Dota 2 draft, patch and lineup guide covers draft order, patch version and lineup. The League of Legends draft, split and patch guide covers draft, split context and patch changes.

If you are working on rating methods rather than a single title, start with the ELO rating guide. A rating system is a way to structure an analysis; none of them is automatically correct for a given esports market, and the guide explains the method rather than endorsing a number.

Use each guide for the dimension that actually drives your question, and treat the worked examples in them as illustrations of method. None of them is a match tip, and none of them claims live odds.

Rules, postponements and edge cases vary

Postponement, stand-in and forfeit are handled by each contract and tournament rule, so there is no uniform refund. A market's rules should name the resolution source and deadline. For a given competition, the applicable current event rulebook is what matters, not a generic description of how leagues usually work.

Riot's official Competitive Operations library is where competition regulations and updates are published. That capture is a library index, not proof of any particular VALORANT patch, map pool, win rate or roster. It tells you where to find the applicable rulebook, not what a specific match will produce.

Before reading any price, verify the actual rules for that market and avoid extending one duration or source to another. When the rules are silent on an edge case such as a mid-series stand-in, treat that as uncertainty the title alone cannot resolve.

What to check first

Identify the outcome type: match winner, map winner or handicap. Identify the format: BO3, BO5 or something else. Then read the individual market flags and the resolution rules, including the source and the deadline.

These checks do not guarantee a profit and do not remove the possibility of losing; a mispriced market can still go against you even when your reading of the rules is correct. They mainly keep you from spending effort on the wrong object, which is where the format and outcome mix-ups cost the most.