Elo, MMR, and Matchmaking in League of Legends: How the System Really Works

The term Elo originally refers to a rating system developed by chess player Arpad Elo in the 1960s. In the context of League of Legends, the word Elo is commonly used colloquially to mean rank or skill level, even though Riot's actual rating system is more complex than the original Elo formula. Riot uses a system called TrueSkill-influenced MMR — Matchmaking Rating — which is a hidden numerical score that more accurately reflects your current skill level than your visible tier and LP.

Your visible rank — Iron, Bronze, Silver, Gold, Platinum, Emerald, Diamond, Master, Grandmaster, Challenger — is a display layer built on top of your MMR. It uses LP gains and losses to create a ladder experience with milestone promotions, but the underlying matchmaking uses your raw MMR score to find opponents. Your visible rank and your hidden MMR can diverge significantly, particularly after periods of high win or loss streaks, which is why some players describe themselves as being above or below their rank in terms of their matchmaking queue.

MMR updates after every game based on whether you won or lost relative to the expected outcome. If you are matched against players whose MMR is similar to yours and you win, your MMR increases by a small amount. If you win against players with higher MMR, it increases more. If you lose to lower-MMR players, it decreases more significantly than losing to higher-MMR players. This mechanism is designed to make your MMR converge toward your true skill level over time.

Riot's matchmaking algorithm attempts to create games where each team has roughly equal aggregate MMR, with individual player MMR as close as possible within each role. The system first identifies the role each player queued for, then finds a combination of five players per team where the MMR distribution is balanced. The goal is a predicted 50 percent win probability for each team before the game begins, though draft and individual performance will deviate from this prediction.

Queue times create trade-offs in matchmaking quality. When the algorithm cannot find a perfect MMR-balanced lobby within a few seconds, it widens the acceptable MMR range to reduce wait times. During off-peak hours when fewer players are in queue, the system accepts wider MMR ranges to avoid excessive wait times, which is why some games during late-night hours feel more imbalanced than games during peak hours when the matching pool is larger.

Position autofill is a feature where the system assigns you a role different from your preference when it cannot find a player to fill a specific role need within a reasonable queue time. Autofill games predictably produce lower win rates for the autofilled player because they are playing outside their comfort zone. The matchmaking system is aware of autofill assignments and adjusts its predictions slightly, but autofilled players still tend to underperform their MMR in unfamiliar roles.

LP gains and losses are calculated based on the relationship between your visible rank LP and your hidden MMR. When your MMR is significantly higher than your visible rank — a situation that often occurs early in a season or after a win streak — the system gives you more LP for wins and takes less for losses to accelerate your promotion to the rank that matches your MMR. When your MMR is lower than your visible rank, the opposite occurs: you gain less LP for wins and lose more for losses.

This explains the common complaint that players get 18 LP for a win but lose 22 for a loss. It does not mean the system is rigged — it means your MMR is slightly below your visible rank, and the system is slowly pulling your visible rank down to match reality. Players who climbed quickly during a win streak and then stabilized often experience this correction period where LP math feels unfavorable. The fix is to improve your win rate back above 50 percent to raise MMR.

LP gains and losses are independent of your individual performance in the game. You receive the same LP for a 15-kill stomp as for a close 30-minute grind, because the matchmaking system only cares about the team outcome, not individual contribution. This design is intentional — rewarding individual stats rather than wins would incentivize stat-padding behavior that is detrimental to team play. The only path to consistent LP gain is consistent team winning.

Riot does not publicly expose MMR values through its API. Third-party tools that claim to show your exact MMR are making estimates based on observable data — your LP gains, the average rank of your opponents, and historical patterns for how the system behaves at different MMR levels. Sites like WhatIsMyMMR.com and the MMR displays in OP.GG and U.GG are well-calibrated estimates but not exact values. Treat them as approximate indicators within a range of plus or minus 100-200 rating points.

The most reliable indicator of your MMR is the average visible rank of the players in your recent games. If you are Gold 2 but consistently being matched with and against Platinum 3 and Platinum 4 players, your MMR is calibrated at Platinum-level and you should expect faster LP gains until your visible rank catches up. If you are Gold 2 and being matched with Gold 4 and Silver 1 players, your MMR is at or slightly below your visible rank.

Your LP gain per win is the clearest data point. Consistent gains of 25 or more LP per win indicate above-rank MMR. Gains of 20 to 24 per win indicate your MMR is close to your visible rank. Gains of 15 to 19 per win indicate below-rank MMR. These thresholds are approximate and vary based on your exact tier, but the principle holds across all rank brackets above Iron.

At the start of each ranked season or split, League of Legends runs placement games that help the system calibrate your starting MMR. The system does not start completely fresh — it retains a significant portion of your previous split MMR as a prior estimate, then updates based on your placement game results. This is why a player who was Platinum last split will typically be placed in Gold or high Silver after placements, not Iron, even if they lose multiple placement games.

The number of placement games and the weight given to their results has changed across League's history. In recent years, Riot has reduced the influence of placements on final starting rank because high-stakes placement games produced volatile and often inaccurate starting positions due to the psychological pressure and tendency to play unusual compositions or champions. A longer calibration period spread across the first 50 or so ranked games of a split provides a more accurate initial rank than 10 heavily-weighted placements.

Fresh accounts — new players or smurf accounts with no previous ranked history — receive a more aggressive MMR calibration during their placements because the system has no prior information. The system gives extra weight to early game outcomes to converge toward an accurate estimate faster. This is why smurfs who dominate placement games get placed significantly higher than players who win and lose evenly — the calibration is designed to identify outlier skill quickly.

Playing in duo queue pairs you with a friend and keeps you on the same team, but it introduces a matchmaking complexity. A duo of two players has a coordination advantage over solo players on both teams, so the system compensates by attempting to match the duo against another duo or by slightly weighting the opposing team's MMR higher to account for the coordination bonus. How aggressively this compensation is applied varies by the MMR gap between the two duo partners.

When duo partners have significantly different MMR levels — for example, a Diamond player queuing with a Gold friend — the system must compromise between matching the Diamond player against appropriate opponents and keeping queue times reasonable. The result is typically games where the higher-MMR player is the strongest player in the lobby and the lower-MMR player is being carried against opponents above their individual skill level. This benefits the lower-ranked player's MMR but produces imbalanced games.

Five-person premade parties bypass normal MMR balancing because the system cannot match five premades against each other on demand. Organized five-stacks in normal queue can have very uneven games as a result, which is one reason Riot restricts full premades to flex queue rather than solo/duo in competitive play. The matchmaking concessions required to accommodate five-player coordination are larger than the system can reliably balance in solo/duo queue.

Understanding MMR mechanics has direct implications for climb strategies. The most important insight is that win rate drives everything — a consistent 53 to 55 percent win rate over many games will eventually climb any rank because your MMR will outpace your visible rank, triggering favorable LP math. Players who focus on winning rather than individual statistics are aligned with what the system actually rewards.

Early season play significantly impacts your MMR trajectory for the split. If you can maintain above 50 percent win rate in the first 30 games of a season, your calibration MMR will be set above the average for your rank tier, giving you easier LP gains throughout the split. Conversely, a poor start — 30 to 40 percent win rate in the first 20 games — can create a difficult hole where your MMR is calibrated below your visible rank, making every subsequent LP gain difficult even when you are playing well.

Champion pool discipline helps maintain consistent win rates across varied game conditions, which is the underlying driver of MMR improvement. Players who play three or fewer champions develop deeper game-state pattern recognition on those champions, making consistent decision-making more achievable than players who rotate a large pool of 15 to 20 champions and never develop champion-specific depth. The data consistently shows that players who narrow their pool during climb phases achieve faster LP progression than those who play for variety.

League of Legends applies rank decay to players in Master tier and above who do not play ranked games within a specified time window, reducing their LP over time until they return to activity. This decay mechanism does not apply to Diamond and below — a Gold player can take six months off and return to the same rank they left. However, their MMR may be treated as stale by the system, which can affect matchmaking quality during the readjustment period after returning.

When returning from a long break, expect your initial games to feel harder than they did before your break. This is partly because the meta has shifted and your mental models are outdated, and partly because the matchmaking system may be testing your actual skill level against the new population distribution. Win rates typically dip in the first 10 to 20 games back before stabilizing as both you and the matchmaking system readjust.

Pre-season is a particularly volatile period for MMR because Riot partially resets ranks and MMR values at the start of each ranked split. The size of the reset varies by year and is communicated in Riot's pre-season patch notes. During the reset window, matchmaking quality is lower than mid-season because everyone's calibration is running simultaneously across the player population. Competitive players who want the most stable matchmaking environment typically avoid the first two to three weeks of a new ranked split.