About

How does this work?

This website uses information gathered from Strive's replays on PC to track players and assign them Glicko-2 ratings based on their performance. This system is widely used in competitive games and is especially suitable for 1v1 play. Matchups are as follows:

Rating difference Win probability
050%
10063%
20075%
30085%
40091%
50095%
If you take on someone 300 points above you in a FT10, you can expect to take about 2 games off of them.

What's a good rating?

Average rating is about 1200, if you want to compare yourself to others. I've got the "strong player" cutoff at 1800 since those are good enough to beat the average player 90% of the time or so, but above them there are plenty of 2100s that do the same to them.

What does the '±' part mean?

This represents how confident the system is in your rating. If you've got a rating of 1450 ±150, that means the system is 95% confident that your rating is between 1300 and 1600. New players/characters started out at 1500 ±700, which means the system thinks they could be anywhere between an 800-rated newbie or a 2200 rated pro. As you play more games, especailly against opponents close to you in skill, this number will naturally go down.

How often does this page update?

New data is scraped from Strive every minute, and ratings are recalculated every hour. This lets the system use multiple games each time it wants to update the ratings, resulting in a more stable system.

If you're asking about the website itself getting new features, updates are sporadic. I still have a few core things I want to add, though. Feel free to ask me on Github or Twitter!. DMs etc. are open.

What's does 'expected wins' mean?

This is how often the system expects you be able to beat your opponent in a game, based on your relative ratings. It doesn't take your personal matchup history into account.

This is also the number that you need to exceed on average if you want your rating to go up. If you can take down opponents the system thought was stronger than you, then clearly it was wrong!

What's the difference between 'Win rate (raw)' and 'Win rate (rating adjusted)'?

Raw win rate is what it sounds like - all of your wins vs that character divided by all your games against this character.

Adjusted win rate takes into account your chance of winning based on diference in rating. Instead of simply adding together wins and losses, wins and losses are weighted by your expected win rate. If you beat someone that you had a 90% chance to win against, you only get .1 added to your "wins" total. If you lose to someone who you had a 70% chance to win against, you get the full .7 added to your loss bucket. Then at the end the adjusted win rate is your win bucket divided by your win bucket plus your loss bucket. Final estimation is (win_bucket) / (win_bucket + loss+bucket)

Can I gain more rating by farming lower skilled players?

No, not unless you can beat them even more consistently than the game thought you should. If anything, going up against lower sklled players carries an element of risk, since losses to them causes you to lose a lot more rating than losing to a similarly rated opponent would.

Is this PC only?

No! You can find console games at console.ratingupdate.info. Once crossplay is implement the websites will be unified.

Does rating change over time?

Your actual rating value won't change over time but if you don't play for a while your deviation will gradually increase. This might remove you from the rankings if you go over 150, but you can quickly get the deviation down again by just playing some games.

It does this because if you haven't played in a while the system can't be as sure about how good you are.

How do I read the matchup chart?

If you're the character on the left (written in full), and your opponent has the character on top (with a shortened name), the grid cell gives you the win probability.

Some of my games are missing.

The site only looks at replays from the tower, so if you've been playing in the park those games won't show up.

I only started grabbing replays in December, and my scraping had a lot of downtime up until about January fifth. Games after that should all be present though.

My question wasn't answered here.

Send me a message on Twitter or open an issue on Github!