You've probably heard of Elo ratings. They're everywhere - chess, video games, sports betting, even dating apps. But what actually is an Elo rating?
Here's the simple version: It's a number that goes up when you win and down when you lose. The amount it changes depends on how good your opponent is.
That's it. That's the core idea.
But the details matter, because Elo solves a specific problem that simpler systems can't handle. Let's dig into what that problem is, and how Elo addresses it.
Imagine you're running a weekly board game night. After 10 weeks, here's what your records look like:
Straightforward, right? Riley is clearly better. But then you notice something:
Riley has been playing mostly against beginners. Sam has been facing your three best players every single week. Riley's 70% against weak competition might actually be less impressive than Sam's 60% against tough competition.
How do you compare them fairly?
This is the problem Elo ratings solve: accounting for opponent strength. Simple win-loss records treat all wins the same. Elo doesn't. Beating a strong opponent matters more than beating a weak one.
When you first join a competition using Elo, you get an initial rating - usually something like 1200 or 1500. Everyone starts at the same place. This number is your rating. Higher = better.
When two players face each other, Elo calculates an expected score based on their current ratings.
If you're rated 1600 and your opponent is rated 1400, the system "expects" you to win. You're the favorite.
And the reverse for your opponent:
This is the magic of Elo. Beating weak opponents barely moves your rating. Beating strong opponents jumps it significantly.
Let's watch some ratings change in real-time.
Starting positions:
Ben is rated 200 points higher. The system expects Ben to win about 75% of the time.
Maya wins.
This is an upset. Maya beat someone rated much higher.
Ben is rated 376 points higher than Zoe. The system expects Ben to win about 95% of the time.
Ben wins.
Expected outcome. Ratings barely move.
Maya wins.
Expected. Small change.
New standings:
Maya moved up significantly by beating Ben. Ben barely gained anything from beating Zoe. Over time, these adjustments converge on everyone's true skill level.
You don't need to know this to use Elo, but here's the basic formula:
Expected Score = 1 / (1 + 10^((Opponent Rating - Your Rating) / 400))
New Rating = Old Rating + K × (Actual Score - Expected Score)The K-factor controls how much ratings change per game (usually 16 or 32 for casual competitions).
Example calculation:
Expected Score = 0.24
New Rating = 1400 + 32 × (1.0 - 0.24) = 1424
That's the whole formula.
Elo works really well when people show up irregularly, play different numbers of games, or face opponents of varying skill. If someone plays 5 games against strong players and someone else plays 20 games against beginners, Elo keeps the comparison fair. Ratings stay valid whether you play every week or once a month, and they remain meaningful over long periods.
Chess uses Elo because competitions mix grandmasters (2700+) with club players (1600) and beginners (1000). When they play each other, the system accounts for these differences automatically. Competitive video games like League of Legends use similar systems (called MMR) to rank millions of players who compete at different times against different opponents.
On the other hand, if you're running a short 8-week league where everyone plays everyone once, simple win-loss records work great. Same if everyone's roughly equal skill, or if your group just prefers "count the wins" simplicity. Round-robin formats especially don't need ratings - everyone plays everyone the same number of times anyway. No need to over-engineer it.
Elo tells you probability, not certainty. A 1600-rated player might have a 75% chance to beat a 1400-rated player. But that means 25% of the time, the underdog wins. Upsets happen.
Ratings are relative. A 1800 rating in your board game group doesn't mean you'd be 1800 in a professional league. You're being compared to the pool of players you actually compete against.
Ratings adjust over time. If you improve, your rating rises. If you get rusty, it falls. It tracks your current skill, not a permanent label.
You have a few options:
Manual calculation: Use the formula above after each game. Tedious but doable for small weekly groups.
Spreadsheet: Set up a Google Sheet with the formulas. Enter results, ratings update automatically.
Use a tool: Software handles all the math for you - just enter who won each game. Tools like shmelo are built for casual competitions (board games, ping pong, whatever you're tracking). Elo updates happen behind the scenes.
Elo ratings are a way to rank people that accounts for opponent strength. That's the whole point.
It's worth using when your competition has irregular participation, people play different opponents, and skill levels vary. It's overkill for short-term leagues where everyone plays everyone equally, or when you just want to keep things simple.
There's no one-size-fits-all. Elo is a tool. Use it when it solves your specific problem. Don't use it when simpler approaches work fine.
The good news is you don't need to understand the math to use it. You just need to know what problem it solves. And now you do.
Want to try Elo rankings for your group? Shmelo handles all the calculations automatically - just track who wins and the leaderboard stays fair.
Create a leaderboard →