Provisional Rating Calculator
Estimate an early Elo-style rating from opponent ratings, match score, provisional weight, confidence band, and remaining games.
| Rating Difference | Expected Score | Meaning | Use In Calculator |
|---|---|---|---|
| -400 | 9.1% | Large underdog | Big reward for scoring |
| -200 | 24.0% | Clear underdog | Upsets move rating up |
| 0 | 50.0% | Even pairing | Draw matches expectation |
| +200 | 76.0% | Clear favorite | Loss costs more points |
| +400 | 90.9% | Large favorite | Score must be strong |
| K Setting | Typical Use | Rating Feel | Example Swing |
|---|---|---|---|
| 60 | First placement games | Very fast | 30 pts per +0.5 |
| 40 | Provisional lists | Fast | 20 pts per +0.5 |
| 32 | Club ladders | Responsive | 16 pts per +0.5 |
| 24 | Settling players | Moderate | 12 pts per +0.5 |
| 16 | Stable pools | Slow | 8 pts per +0.5 |
| Opponent Type | Weight | Why It Matters | Calculator Setting |
|---|---|---|---|
| Established | 100% | Opponent rating is trusted | Full weight |
| Mostly rated | 85% | Some ratings still moving | Default setting |
| Mixed provisional | 70% | More uncertainty in field | Wider band |
| Unrated estimates | 55% | Seed ratings are guesses | Lowest weight |
| Score Pattern | Performance Idea | Confidence Effect | Common Note |
|---|---|---|---|
| Perfect score | Above field average | Still capped here | Needs more games |
| About 50% | Near field average | Moderate band | Good anchor point |
| Draw heavy | Stable signal | Slightly tighter | Useful vs peers |
| All losses | Below field average | Still capped here | Do not overread |
Early in the process, then, you’ll probably think of your rated games as less like tests of skill than a bit of a lottery. On paper you might be facing someone who looks half your strength, yet you crush them. Or maybe you’re facing someone who seems stronger, but you win. That’s okay… This is what happens with provisional ratings.
Your rating after three or four matches isn’t an accurate representation of your strength. At best it’s a crude coordinate in a terrain you’ve hardly started exploring. Once you know how that coordinate is calculated, it changes how you see every upset, draw, and loss.
Why Your Rating Changes at First
Rating updates has a throttle called the K-factor. If your K-factor is high, then each game will cause your rating to swing wildly. That’s good; it’s how the system gets data aggressively, and new players need that. Low K-factor keep established players stable. It protects against a single bad day wiping out years of consistency.
When you choose a weight (using the calculator above), the tool runs all those numbers for you. But it helps if you know why you made the choice. In a mixed field, there may be other new player, and their ratings is floating too. This is noise. The tool lets you account for that by adjusting opponent confidence. That says to the algorithm: “Trust this data coming in, but not so much,” because accurate expectations matter more than inaccurate ones. You don’t want your rating to get dragged down based off false hope.
By contrast, performance rating has nothing to do with who you are or where you came from. It’s just about how you performed relative to the people you were playing. Your rating shoots through the roof when you outplay a field of tough competitors. Even if your raw provisional number only inches upward. Performance rating is great for judging one ladder run or one tournament.
What was your skill level like at that time? Did I score well against the specific group I faced? Did I lose to the person who’s really good? That’s the question performance rating asks.
The table on that page show how much your expected results change as your ratings differ. Four hundred points is huge. Outplaying someone four hundred points higher than you is an odd occurrence that should of been rewarded heavily. Losing to them matters very little. The math take into account how hard they are to play.
If you’re obsessing over the exact number after five games, then you made a common error. After playing five games, most people obsess about the specific count. That’s an error. Uncertainty is baked into the design. It’s supposed to be a wide confidence band. That’s a feature, not a bug. It tells you that the rating system isn’t yet sure what to think of you.
You’re crossing the line from unknown to established status (typically around the 30-game mark). At this point, the band begins to tighten. The system starts to feel confident enough to treat you less as a statistical mystery. Until that moment, expect some volatility, it’s the price of admission. You pay with rating points to get more information.
So what’s the trick? It’s all about sample size. In a three game event, getting a perfect score doesn’t tell you much at all about your ceiling. Getting a poor score in a fifty game stretch tells you everything. That’s where the calculator comes into play; it helps you visualize the number of games before your rating should stabilize.
With that in mind, you want to focus less on precision and more on volume. Go play some more games against rated players and let the algorithm do its work. The early swings will smooth out naturaly. You don’t have to force your rating to meet your perception of yourself. Give it time and enough room to breathe and the data will catch up.
Don’t trust the momentary spike. Trust the process. Your rating is a trailing indicator of your improvement. It is not a live feed of your confidence. You don’t want to look good today, you want to be accurately placed in six months.
