League of Legends prediction markets: drafts, patches and splits

What a LoL market contract actually pays

An ordinary binary market pays 1 to the side that resolves true and 0 to the side that resolves false. A rare unknown or 50-50 resolution can pay .50 to each side, but that is a specific rule outcome, not a universal refund or cancellation. So the number you are trading against is the market's own rule text: which source decides, by what deadline, and what happens in edge cases like a postponement or a forfeit. If your reason to hold the position lives only in the title, you are exposed to whatever the rules say in the parts you did not read. Compare that discipline with how other verticals are structured in these esports market guides.

League of Legends adds a second layer: the event rulebook. Riot's official Competitive Operations library is where competition regulations live, and the applicable current event rulebook, not a generic memory of how leagues run, governs format, rosters and match conditions. One concrete example: the 2026 LCP Split 3 handbook makes regular Swiss-stage matches BO3 and qualification and elimination rounds BO5. That is one named competition, not a universal League of Legends rule. A BO3 is a best-of-3 series and a BO5 is best-of-5; the format changes how much a single draft or side choice can decide the series.

2 records, same .7, different evidence

Imagine Team A at 7 wins and 10 maps, and Team B at 14 wins and 20 maps. Both compute to a raw ratio of .7. Read as bare numbers, they look equally strong. They are not equally informative. 10 maps is a small base that a couple of series can swing; 20 maps spread across 2 patches and different opponent pools is not the same object even when the arithmetic matches. The ratio is identical, the uncertainty around it is not. That comparison is an illustrative construction for this guide, not a measured study of any roster.

The decision that follows is not to trust the bigger record. It is to stop calling a historical ratio a current probability. Patch revisions, draft trends, side selection, format changes and roster moves split a season into dated strata, and a ratio only means something inside the strata it came from. If Team A's 7-10 sits entirely before a major patch and Team B's 14-20 crosses 2, you either rebuild the estimate patch by patch or you accept that you are comparing 2 different games. This protocol is author-designed and hypothetical; it is not evidence that any specific window size is enough.

Patch, draft, side: the parts that move under you

A patch is a game update that changes champions, items or systems. A draft is the pick-and-ban phase before a game. Side selection decides who plays blue and who plays red. Each can move a team's results with the roster unchanged, which is exactly why a result from an older patch is weak evidence for a match played now. The handbook tells you the format of a named competition; it does not tell you any current team's win rate, and you should not manufacture one from a format document.

Fearless draft, where champions picked in earlier games of a series cannot be reused later in that series, gets discussed as if it were global. This guide does not assert that. If a market's rules reference a draft format, check the applicable rulebook for that event, because a format that thins the champion pool changes what a deep pool is worth in game 3 or game 4 of a series. Treat every such claim as conditional until the rulebook for that competition confirms it.

Splits, Swiss stages and why BO3 does not feed BO5

A split is a segment of the competitive season, and its internal format determines what kind of evidence you have. In the 2026 LCP Split 3 handbook, the regular Swiss stage runs BO3 while qualification and elimination run BO5. A Swiss-stage format pairs teams by comparable standing rather than randomly, but that is a general property of Swiss systems, not a fact drawn from the LCP handbook. Qualification and elimination rounds are knockout-style with higher stakes and more room for in-series adaptation. A team can genuinely be stronger in one and weaker in the other.

The practical consequence: a BO3 win ratio does not transfer cleanly to a BO5 market. A BO5 gives both teams more chances to adapt after a loss, adjust side priority and change draft priors, and it also changes the variance of the series outcome. If your market resolves on a BO5, look for BO5 evidence or rebuild the estimate with format-specific assumptions and say so out loud. If the rulebook points to a regular Swiss match, your BO3 sample is the relevant one, and pairing it with a knockout-stage result is a category error.

Who declares the result, and when it counts

Resolution rules name the source, the deadline and the edge cases, and the mechanics differ by product type. Ordinary prediction markets on named events are resolved through UMA, while Up/Down markets use a Chainlink time-weighted average price, or TWAP, taken at the start and the end of a window, with the end at or above the start resolving UP. Those are different mechanisms with different failure modes. For a League of Legends match you are usually looking at an ordinary event market resolved by a named source, but confirm the actual rules instead of importing assumptions from a price-based product.

Deadlines carry their own risk. A match can be postponed, replayed or decided by forfeit, and the rule text decides whether your position survives that. If the market mirrors the event's scheduled date and the event moves, the exposure is yours, not the market's. If the rules are silent on postponement, you are holding an edge case you did not price. This is a risk framing, not a promise about how any specific market will behave.

A working checklist before you size anything

Use the table as a short filter. Each row maps something you observe to the question that decides whether the number is usable, and the third column is a conditional choice rather than a verdict. None of the rows are rankings, and none of them replace reading the rulebook for the specific event you care about. The Elo ratings in prediction markets guide shows what changes when a single value compresses a whole history.

What changes when you calibrate forecasts over a season

Treating a ratio as a probability is the core error this guide targets, and it is the same error that shows up across rating systems and forecasting habits. A rating built from a small window will move sharply after one result; a rating built from a long window will lag a real change. Neither is wrong by itself, but the 2 answer different questions. If you maintain per-patch windows and record them before the match, you can later see whether your .7 was well priced or just confident. The calibration in forecasting literature is about exactly that gap between stated confidence and observed frequency.

Ratings carry the same hazard. An Elo-style number compresses a history into one value, which is convenient and also the reason it hides which patch or roster the value was built on. Elo ratings in prediction markets are most useful when the underlying window is stated and the strata are visible. Writing those strata down before the series, in a fixed format, is what makes the estimate checkable afterwards.

Limits and losing cases

Nothing here gives you a current win rate, a roster move, a patch effect or a price. No country-specific authorization, tax or legal position is implied. The 2026 LCP Split 3 handbook describes one competition's format and is not a universal League of Legends rule; the .7 comparison is a hypothetical illustration of evidence size and dated strata, not a measured result. Fees, spreads and opportunity cost are real costs of holding a position, and this guide does not invent dollar values for them.

Losing cases are ordinary, not exceptional. A team can win a Swiss-stage BO3 match and lose a qualification BO5 in the same season, and a 14-20 record can be followed by a losing series because the patch changed what the record was measuring. The next decision is concrete: pull the applicable rulebook, write down the dated window, patch, side and format, decide whether those match the market's resolution rule, and only then decide whether to hold. You can compare your forecasts with resolved outcomes by keeping a prediction journal across the split.