How Elo Rating Works
Every signed-in REPPED player has an Elo rating — a single number that estimates your skill relative to everyone else, and updates after every ranked match. It's the same rating system chess has used since the 1960s, adapted to 30-second fitness battles.
Starting point
Everyone starts at 1000. From there, every match moves your rating up or down based on the result and the strength of your opponent.
The core idea: expected outcome
Before a match, the system calculates the probability you "should" win based on the rating gap between you and your opponent. If you're rated much higher, you're expected to win — so winning nets you very little, and losing costs you a lot. If you're the underdog, the reverse is true: beat a much stronger opponent and you gain significantly more than beating someone at your own level.
The K-factor
REPPED uses a K-factor of 32 — this controls how fast your rating moves per match. A higher K-factor means faster-moving ratings (good for reflecting real, recent improvement) but more volatility; a lower one is more stable but slower to catch up to your true skill. 32 is a middle ground: a hot streak or a bad night moves your rating meaningfully without a single match swinging your whole rank.
The floor
Your rating can never go below zero, no matter how brutal a losing streak gets. Nobody gets stuck in negative-Elo purgatory.
Seasons and soft resets
Every calendar month is a new season. At the start of each one, ratings pull 25% of the way back toward 1000 — never a full wipe, but enough to keep the leaderboard fresh and give returning players a real shot at climbing again. Full mechanics are on our Seasonal Ranking page.
The rank ladder
Your Elo places you into one of seven visible ranks:
Why this feels fairer than a plain win counter
A leaderboard sorted by raw wins rewards volume — play constantly against easy opponents and you can out-rank someone far stronger who plays less often. Elo corrects for this automatically: your rating converges toward your real skill level over time, regardless of how many matches you've played.
Related reading
- Check the current leaderboard
- How the AI judges every rep — the score your Elo is built on
- The full history of Elo, on our blog
A worked example, with the actual numbers
Say you are rated 1000 and you meet somebody rated 1200. The formula puts your chance of winning at roughly 24 per cent. Win and you gain about 24 points; lose and you drop about 8.
Now stand on the other side of it. Rated 1200 against the 1000, a win is worth about 8 points and a loss costs about 24. Nobody is being rewarded or punished for who they are — the system simply pays out more for results it did not predict, and it does so symmetrically.
The consequence people find counter-intuitive is that the board cannot be farmed. Ten wins against somebody 400 points below you are worth less than a single win against somebody 200 above. Avoiding strong opponents does not protect a rating so much as freeze it.
Your first few battles do not mean much
A brand-new rating of 1000 is not a measurement, it is a placeholder. It says nothing about you because nothing has been observed yet, and it takes a genuine run of battles before the number carries information.
That matters because the first week is exactly when people read the most into it. A rating that swings 200 points either way across your first five battles is not telling you that you have improved or collapsed — it is the system working out roughly where you belong. Judge it after a few dozen battles, not a few.
What a rating does not measure
It measures how you did in thirty seconds, at whatever the roulette picked, against the people you happened to be matched with. That is a real thing, and it is a narrow one.
It is not a fitness score and it is not a health metric. Somebody rated below you may well out-train you over an hour, be considerably stronger in absolute terms, or be better at every form of exercise that is not this one. It is the rating for this game — exactly as much as any rating anywhere should claim.