I've created an AI system tuned for predicting Rocket League esports matches. The ranking puts all the estimated skills of the teams in order. Click on the bars to see the two week rating trend. Learn more about the AI in this video. The match data comes from ShiftRLE's database. Keep in mind that not all esports matches will be added to the database.
When do the ratings get updated?
The ratings should auto-update 6 AM UTC every day unless there is some issue. The date displayed on this page will be the date of the latest game in the database.
Why is my favorite team missing?
In order to appear on this ranking, your team has to have played in an A-Tier or higher tournament in the past two weeks, or they have to be on ShiftRLE's list of active teams.
Why is my favorite team so low? Why are you hating on minor regions?
There are no manual adjustments of ratings. No matches get removed from the dataset. Placement in tournaments do not matter. The only thing that matters is whether you're able to beat the teams in front of you. The AI doesn't necessarily hate on minor regions. I believe most people just underestimate how much depth EU and NA really have. Just because a team misses out on the major by a few points, doesn't mean they're all of the sudden worse than the minor region team that made it. Most minor region teams have done really well one tourney and fallen short multiple times in a row. That is no different than many of the EU and NA teams in regionals.
Keep in mind that sometimes 10 teams are within 50 rating points, and sometimes one team is 100 points above the next. The rating difference is what actually matters for a strength evaluation.
No ranking will ever be perfect. This one tends to move around quite a bit because that gives the best average prediction on the past dataset. If you come back a few times, you may find that one day it looks a bit crazy, and one event later, you agree with it.
What do the ratings mean?
They're a relative measure of skill. The absolute value has no meaning. Only the difference to another team. The approximate odds in a best of five are:
50 rating difference: ~57% win
100: ~64% win
200: ~76% win
300: ~84% win
400: ~91% win
The transparent portion of the bar shows the uncertainty that the AI has about a team. If that bar is really large, then you should take the teams rating with an extra grain of salt.
Why can't I filter?
The page is just getting started. More to come, including player rankings. You're always free to message me on Discord if you think you have a great novel idea for this page.
03/23/2023, 9:00 PM
