Elo Ratings vs. Betting Odds: How Soccer Win Probabilities Work
One forecast comes from the betting market, the other from past results. Here's what each one knows, how to read them, and why they don't always agree.
- Betting odds can be turned into win percentages by dividing 1 by each price and removing the bookmaker's margin.
- Elo is a strength rating built only from match results. The gap between two ratings, plus home advantage, predicts the match.
- The odds know about injuries, lineups and travel. Elo doesn't, which is the main reason the two disagree.
- MLS is hard to predict: even a heavy favorite loses or draws about 4 times in 10.
- A draw is rarely the single most likely result, even in close matchups.
Two Ways to Forecast a Match
On the Soccer Prediction League match pages, every MLS game shows two sets of win chances. The first comes from DraftKings betting odds. The second comes from Elo ratings. Both answer the same question, how likely is a home win, a draw or an away win, but they get there from completely different starting points.
| Betting odds (DraftKings) | Elo rating | |
|---|---|---|
| Built from | What bettors and the bookmaker expect, right up to kickoff | Past match results only |
| Knows about injuries and lineups | Yes | No |
| Knows about travel, rest and rotation | Yes | No |
| Updates | Constantly, as news breaks | After each match is played |
| Includes a profit margin | Yes, which has to be removed | No |
How Betting Odds Become Win Percentages
US sportsbooks often show American odds like -140 or +250, but the same prices can be written as decimal odds, which are easier to work with. Decimal odds of 2.10 mean a $1 bet returns $2.10 if it wins. Dividing 1 by the decimal odds gives the implied chance of that result:
| Result | Decimal odds | 1 ÷ odds | After removing the margin |
|---|---|---|---|
| Home win | 2.10 | 47.6% | 45% |
| Draw | 3.40 | 29.4% | 28% |
| Away win | 3.60 | 27.8% | 27% |
| Total | 104.8% | 100% |
The raw numbers add up to 104.8%, not 100%. That extra 4.8% is the bookmaker's margin, often called the vig or overround, and it's how the sportsbook makes money whichever way the match goes. Dividing each number by the total strips it out and leaves a fair set of win chances. The percentages on the Prediction League's match pages are calculated this way.
How Elo Ratings Work
Elo was invented for chess by physicist Arpad Elo and has been adapted for soccer by projects such as World Football Elo Ratings and ClubElo. Every team has one number. In the MLS Elo ratings used by the Prediction League, every club started at 1,500 in the league's first season, 1996, and the ratings have moved with every league match since.
After each game, the winner takes rating points from the loser. How many depends on two things. An upset moves more points than an expected result, because the ratings were more wrong. A big win moves more points than a narrow one: in this model the change grows with the square root of the goal difference, so a 3-0 win counts for about 1.7 times as much as a 1-0 win. A draw moves points toward the weaker team.
To forecast a match, Elo compares the two ratings and adds a home advantage worth about 92 rating points. Say a team rated 1,560 hosts a team rated 1,530. With home advantage the host is 122 points better, which the standard Elo formula turns into an expected score of about 67%. That figure counts a draw as half a win, so it isn't the same as a 67% chance of winning.
To split that into home, draw and away percentages, the Prediction League calibrates Elo against about 4,800 MLS league matches since 2016. Draws happen about 27% of the time between evenly rated teams and fall to about 21% when one team is far stronger. The calibration also found that home teams in MLS win more often than the basic formula expects, so the percentages allow for that.
Why the Two Forecasts Disagree
Most weeks, DraftKings and Elo agree on the favorite in most matches. The interesting ones are where they don't, and there are a few usual reasons.
The odds react to team news. If a club's top scorer is injured, or a coach is expected to rest starters before a midweek game, the betting market moves within hours. Elo won't notice until the results start to change.
Elo remembers the whole season. A team that started badly but has just signed two players may still carry a low rating, while the market is already treating it as a different side. The reverse happens too: a team on a lucky run can look stronger in Elo than the market believes.
The market can also overreact. Big-name clubs and recent headlines attract bets, and a team that just lost 4-0 can be priced more harshly than one result deserves. Elo applies the same rule to every match, which makes it a useful check on the market's mood.
What This Means in MLS
MLS is one of the harder leagues to predict. A salary cap, heavy travel and frequent squad rotation keep teams close together. In league matches from 2023 to 2026, home teams won 46% of the time, 26% ended in draws and away teams won 28%. And since 2016, even when Elo rated one team 200 or more points stronger, that team won only about 6 times in 10.
That has a few practical consequences for anyone making picks:
- A forecaster who gets a little under half of MLS results right is doing well. Picking the Elo favorite every week would have been right about 47% of the time over those seasons.
- Draws are common but rarely the most likely single outcome, so picking one only makes sense when both forecasts rate the teams nearly even.
- The matches where DraftKings and Elo disagree are the best places to use your own knowledge, because at least one of the two forecasts is missing something.
Vegas Bot and Elo Bot
To give every player something to measure against, the Prediction League runs two automatic players. Vegas Bot always picks the result DraftKings makes most likely. Elo Bot always picks the higher-rated team after home advantage and never picks a draw. Both appear in every league table. Across the 2026 matchweeks where both had picks, they finished almost level, which says a lot about how close the two approaches are in MLS. Beating both bots over a season is a good sign that your picks are adding something the numbers miss.
Frequently Asked Questions
- What is an Elo rating in soccer?
- An Elo rating is a single number for a team's strength, calculated only from match results. Teams gain points for wins and lose points for losses, with bigger changes for upsets and bigger winning margins. The gap between two teams' ratings, plus an allowance for home advantage, predicts how a match between them should go.
- How do you turn betting odds into a win percentage?
- Divide 1 by the decimal odds for each result. The three numbers add up to a little over 100% because of the bookmaker's margin, so divide each one by their total. For odds of 2.10, 3.40 and 3.60 that gives roughly 45% home, 28% draw and 27% away.
- Why do Elo ratings and betting odds disagree?
- They use different information. Betting odds react to everything the market knows before kickoff, including injuries, lineups, rotation and travel. Elo only knows past scores, so it can't see a missing star player or a team resting starters. Disagreements usually mean something has changed that results haven't shown yet.
- Which is more accurate, Elo or betting odds?
- Betting odds usually have the edge because they include team news, but the gap in a low-scoring, unpredictable league like MLS is small. In Soccer Prediction League's own tracking of the 2026 MLS season, a bot picking the DraftKings favorite and a bot picking the Elo favorite finished almost level over the matchweeks where both picked.
- Why is a draw almost never the most likely result?
- Even in evenly matched MLS games, draws happen only about 26% to 30% of the time, while one of the two teams wins the rest. Because the win chances are split between two teams and the draw is one outcome, the favorite's win chance is usually higher than the draw chance.
