Valorant prediction markets: maps, agents and series formats

A map contract and a series contract are different bets

Before you price anything in Valorant, decide which question the contract actually settles. A single map contract pays on the winner of one map. A series contract pays on the winner of the match, which may be a best of 3 or a best of 5. These are different bets with different probabilities even for the same 2 teams on the same day. If you carry a series probability into a map contract, or the reverse, your number is wrong from the start.

Market rules, not titles, define settlement. Polymarket's resolution documentation states that ordinary binary markets pay 1 to winning shares and 0 to losing shares, with rare unknown or 50-50 resolutions paying 0.50 each rather than a universal cancellation or refund. The rules name the source, the deadline and the edge cases. For a Valorant market, look for which operator or dataset is the source and whether settlement happens after the final map, after the series, or on a calendar date. If a match is postponed, forfeited or replayed, the rule text is what decides your payout, so read it before sizing.

Riot's Competitive Operations library is the canonical place to confirm specific competition rules for a given event. It collects competition regulations and updates, and the applicable current rulebook is the reference, not a generic description of how leagues usually work. The captured index itself does not establish a named map pool, a patch, a roster or a win rate. If a market rule says "per official rules" without linking the version, treat that as extra uncertainty rather than as a resolved detail.

Series length changes the probability, not the skill

Take a hypothetical team that wins each map with probability 0.6, with every map treated as an identical independent draw. This is an author-designed illustration, not a measured result. In a best of 3 the team can win by taking the first 2 maps, or by losing one map and winning the last 2. The counts work out as: win in 2 maps, 0.6 times 0.6 equals 0.36; win after a loss, 2 orderings each 0.6 times 0.4 times 0.6, so 0.144 twice. The series probability is 0.36 plus 0.144 plus 0.144, which is 0.648.

Now extend the same assumption to a best of 5. Summing the sequences in which the team reaches 3 wins gives 0.68256. The jump from 0.648 to 0.68256 is 3.456 percentage points, and it comes purely from the format, with no change in the teams. Longer series compress the underdog position: a team that loses more than half its maps wins a best of 5 less often than a best of 3, and wins a single map most often of all. That direction matters when you compare a map market with a series market on the same fixture.

Notice what the arithmetic does not do. It does not tell you a fair price in a real match, because the assumption of identical independent maps is false once veto, agent picks, side and patch enter. It only tells you how much the format alone moves the number, which is often more than people expect and rarely zero.

A worked decision: which contract do you actually have

You believe a team wins 0.6 of its maps against this opponent. 2 markets are offered: one on a best of 3 series, one on a best of 5 series. Under the independence assumption your model probability is 0.648 for the first and 0.68256 for the second. If the best of 3 is priced at 0.60, the modelled gap is 4.8 points. If the best of 5 is priced at 0.60, the modelled gap is 8.256 points. On those numbers the best of 5 looks like the larger edge, which is exactly why the format is worth checking before you compare 2 prices.

The losing case is easy to construct and worth holding in mind. Suppose the veto lets the opponent remove the map where your team is strongest and steer play to a map where your team's attack side is weak on the current patch. Your realised per-map rate can fall below 0.6, and in a best of 5 the opponent gets more chances to keep steering. The longer format that looked like the bigger edge becomes the bigger exposure. A positive model gap is a reason to check the rules, the veto procedure and the patch, not a reason to place the bet.

So the order of operations is: identify whether the contract is a map or a series, read the settlement rules, confirm the format and veto procedure against the applicable event rulebook, and only then compare your assumption with the price. If any of those steps is unclear, the honest position is that your number has wider error bars than the model gap suggests.

Why a constant per-map rate is a placeholder, not a model

The 0.6 example assumes every map is the same test of the same strength. Real matches violate that in several ways at once. Map veto changes which maps are played, so a team with one strong map and one weak map gets very different series odds depending on pick and ban order. Agent composition and the current patch change which playstyles are viable, so a composition that dominated last patch can be ordinary the next. Side choice on attack or defense creates map-specific advantages, and a roster move or substitute can reset a team's baseline entirely.

Because those factors are event-specific, a price should be compared with an event-specific estimate rather than a season average. That is where the Riot competitive library earns its place: it is the canonical source for which regulations, and by extension which patch and format, apply to the event you are looking at. If the market does not make the patch identifiable, that gap belongs in your uncertainty, not in your confidence.

Reading the fine print on a Valorant market

For a binary market, winning shares normally pay 1 and losing shares normally pay 0, and the rare unknown or 50-50 outcome pays 0.50 to each side instead of refunding everyone. That is a different payoff shape from a void bet, and it is set by the documented rules. On a Valorant fixture, the details worth extracting are the named source, the settlement timing, and the treatment of forfeits, replays or technical pauses. Where a rule points to "official rules", keep following the pointer until you reach the version that governs this event.

The competitive operations library is the start of that chain, not the end: it is an index of regulations and updates, and the capture used here is the index rather than a statement about a specific patch, map pool, win rate or roster. If no rulebook version is identifiable, that is a reason to treat the market as less well defined than a market whose rules are explicit.

Where Valorant sits among esports markets

Valorant is one title in a wider set of esports markets, and the same discipline travels between games: check the format, check the rules, and do not assume a constant per-map rate. If you are new to this area, a good starting point is an overview of esports prediction markets, which explains how these contracts are structured and settled across titles. From there, the natural next steps are the mechanics of Elo ratings in esports for building a strength estimate, how odds are set and interpreted for reading a price, and calibration checks for probability forecasts for testing whether your stated probabilities match results over time.

Each of those topics stands on its own and each has its own pitfalls. The specific point of the Valorant walkthrough above is narrower: your per-map assumption and the contract's map-or-series distinction are 2 separate decisions, and neither substitutes for the other. Get the second wrong and no amount of modelling the first will help.