What would top players Usatt rating be?

[ThePongCommenter]

Well we have historical USATT rated matches. What does the data say?

Just from my experience of watching matches, I really don't believe its anywhere close to 2%. Maybe 10-15% is my guess

There is no way that properly rated 1800 players beat properly rated 2000 players 15% of the time - this is simply wrong.

 

[TensorBackhand]

There is no way that properly rated 1800 players beat properly rated 2000 players 15% of the time - this is simply wrong.

This is what Qwen says:

If a player has a ELO rating 200 points below his opponent, what is his mathematical probability of winning a match?

Qwen3-235B-A22B10:24 pm
To determine the mathematical probability of winning a match based on the ELO rating difference , we use the formula from the ELO rating system , which predicts the expected score (probability of winning) for each player.

ELO Probability Formula:​

For a player A with rating RA , facing an opponent B with rating RB , the expected score (probability of winning) for player A is:
EA=1+10(RB−RA)/4001

---

In Your Case:​

• Player's rating: RA
• Opponent's rating: RA+200

So the rating difference is −200 points for the lower-rated player.

Plug into the formula:
E=1+10(200)/4001=1+100.51

Now calculate:
100.5≈3.162
E=1+3.1621=4.1621≈0.24

---

Final Answer:​

The weaker player (200 points lower) has approximately a 24% chance of winning the match.

 

[TensorBackhand]

Gemini says:

Based on the Elo rating system, if a player has an Elo rating 200 points below his opponent, his mathematical probability of winning a match is approximately 24%.

This is derived from the standard Elo formula where a 200-point difference corresponds to a 76% winning probability for the higher-rated player, meaning the lower-rated player has a 100% - 76% = 24% chance of winning.

 

[TensorBackhand]

GPT says:

In USATT (USA Table Tennis) ratings, the probability of winning a match is also based on a rating difference, but the scoring formula is slightly different from the traditional Elo system.

✅ Approximate Winning Probabilities (USATT):​

According to USATT guidelines and empirical estimates:

Rating Difference	Lower-Rated Player Win Probability
0	50%
100	35%
200	20%
300	10%
400	5%

📌 So, if a table tennis player has a USATT rating 200 points lower than their opponent, their probability of winning is approximately 20%.​

This is slightly lower than the ~24% you'd expect using standard Elo formulas, reflecting the empirical nature of how USATT ratings are tuned based on actual match data.

 

[TensorBackhand]

Qwen gives a different answer than GPT:

Summary:​

System	Divisor	Win Prob for -200	Notes
Chess (FIDE/ELO)	400	~24%	Steeper curve
Table Tennis (USATT)	200	~9.1%	Sharper drop-off

So, in USATT table tennis , a 200-point underdog has about a 9.1% chance of winning.

 

[ThePongCommenter]

Please stop spamming the thread with what LLMs say, there isn't enough quality table tennis information out there for them to be accurate on this topic. Expected probabilities that a system was based on when it was created doesn't mean that that is the actual real life probability that has manifested decades down the line. Those percentages probably aren't factoring in that there are 3 out of 5 or 4 out of 7 games you need to win in a table tennis match.

Do you actually think that a 2000 player beats a 2400 player 1 out of 20 matches on average? Because that is what you just posted.

 

[TensorBackhand]

Please stop spamming the thread with what LLMs say, there isn't enough quality table tennis information out there for them to be accurate on this topic. Expected probabilities that a system was based on when it was created doesn't mean that that is the actual real life probability that has manifested decades down the line. Those percentages probably aren't factoring in that there are 3 out of 5 or 4 out of 7 games you need to win in a table tennis match.

Do you actually think that a 2000 player actually beats a 2400 player 5% of the time? Because that is what you just posted.

What is your rating? I would love to watch you win 50 matches in a row against 50 -200 rated players. I would be very impressed if you pulled that off.

I know I couldn't do that.

I posted a few LLM answers to provide a data point. I don't know if they are totally correct. You literally said 50 matches in a row without losing, suggesting a 2% win probability. I really doubt that. I guessed it might be 10-15%. The LLM is giving something in that range.

 

[ThePongCommenter]

What is your rating? I would love to watch you win 50 matches in a row against 50 -200 rated players. I would be very impressed if you pulled that off.

I know I couldn't do that.

I posted a few LLM answers to provide a data point. I don't know if they are totally correct. You literally said 50 matches in a row without losing, suggesting a 2% win probability. I really doubt that. I guessed it might be 10-15%. The LLM is giving something in that range.

A lot higher than yours. You are missing the point and focusing on this 2% number - say the actual number is 4% instead of 2% so something that happens 1 out of 25 matches approximately. If I played 25 matches against people 250 points lower than me, would I win 24 of them? Yeah, probably.

One thing both you and Lightspin missed is that it is actually 0 points you win if you beat someone 238 points or more lower than you, and if they win, they get 50. So, that means, it is fully expected that you would beat somone 250 points lower than you to the point that there is not even a single point of reward. What does that tell you?

The point is that it is nowhere near 15%, and that the difference in level as you get higher up increases even more. If you want to sit hit and obsess over pinning down whether it is 2, 3, or 4% I'll leave that to you.

 

[ThePongCommenter]

We saw your clip against Victor who is probably around 400 points better than you. To be clear, you think you would win 1 out of 20 matches against him if he is fully trying? Not games. Matches.

 

[lightspin]

There is absolutely no way a 2200 player has a 20% chance of winning against a 2400 player. According to the USATT's own website, if the 2200 player wins, he gets 40 points. If the 2400 player wins he gets precisely 1 point. If both players level's stay static, one would expect that the 2400 player wins 40 times for each time the 2200 player wins. That means the 2200 player has about a 2% chance of winning. Having practiced and played with tons of people who are 2200 and 2400 (and even some who are 2750+), there is absolutely no way a 2200 player can beat a 2400 player 20% of the time if their ratings are real. None. Zero. In fact, I don't think they can't even win 2% of the time. The gap is that massive. The higher rated you are, the more skewed the ratings are.

Sure a 1200 might beat a 1400 and an 1800 might beat a 2000 from time to time. But a 2200 player rarely will beat a 2400. It just isn't going to happen even 1/40 times. The higher the rating, the less variance in their playing level.

 

[TensorBackhand]

We saw your clip against Victor who is probably around 400 points better than you. To be clear, you think you would win 1 out of 20 matches against him if he is fully trying? Not games. Matches.

I cited a 200 point difference and now suddenly you pulled 400 points out of nowhere?

To answer your question, if I played a player 200 points below me, I don't think I could win 20 matches in a row.

And yes, I regularly play 2100 players, which is 200 points above me and I do win here and there when I execute my best game or if they are slightly off. That's the point, nobody can be at their best all the time. That's why the concept of probability exists.

You didn't answer my question. Do you think you could win 50 matches in a row against a player 200 points less than you?

 

[JJ Ng]

To be fair, this is TB's thread so he can carry the conversation to whatever direction he wants. Plus kudos to TB for uploading a practice match he had with Victor.

I think the issues are, every day in clubs, people talk about their ratings and often talk about their "peak" ratings. What is their true ratings now? No one knows unless that person enters the tournament again. For all we know, he could have been talking about his rating from 8 years ago when he last played in a sanctioned tournament. That happens in US clubs all the time. Take those discussions with a grain of salt. It is like people in the gym trying to see who has the bigger muscles. It is a lot of testosterone distorted trash talks.

If you really want to know your true ratings, play in at least 5-6 tournaments a year and KEEP your ratings. It is harder than you might think. For example, if you are, say, rated 1992 and you enter a U2000 event, you better win it! Because you will lose a lot of points if you lose to most of those players in that event. For example, in your round robin group, U2000 event, you should be a top seed at 1992 and the second seed might 1850? Third seed might be 1650? Fourth seed might be 1400? So to protect that 1992, you better sweep your round robin group and do that for 5-6 tournaments a year, then you are truly 1992 or 2000 rated at the end. If you lose one single match to someone rated 1800 or below, you lose 40 points right there. No way you can protect your close to 2000 ratings if you cannot beat any player rated 1800 below 90+% of the time.

 

[JJ Ng]

I cited a 200 point difference and now suddenly you pulled 400 points out of nowhere?

To answer your question, if I played a player 200 points below me, I don't think I could win 20 matches in a row.

And yes, I regularly play 2100 players, which is 200 points above me and I do win here and there when I execute my best game or if they are slightly off. That's the point, nobody can be at their best all the time. That's why the concept of probability exists.

You didn't answer my question. Do you think you could win 50 matches in a row against a player 200 points less than you?

I think the discussion has been all over the place. One time we are talking about 200 point difference. Another time we are talking about 250 point difference.

For me, I know my ratings because it is pretty solid by now. I am very confident I can beat players rated 200 points below me 19 out of 20 times.

 

[ThePongCommenter]

I cited a 200 point difference and now suddenly you pulled 400 points out of nowhere?

Rating Difference	Lower-Rated Player Win Probability
0	50%
100	35%
200	20%
300	10%
400	5%

​

Pulled out of nowhere? Did you even read what you posted?

And yes, I regularly play 2100 players, which is 200 points above me and I do win here and there when I execute my best game or if they are slightly off. That's the point, nobody can be at their best all the time. That's why the concept of probability exists.

And that probability changes greatly in tournament play versus casual Wednesday night informal matches at the club. As has been told to you repeatedly on multiple threads, but which you seem reluctant to accept.

You didn't answer my question.

Yes I did.

 

[TensorBackhand]

Pulled out of nowhere? Did you even read what you posted?


And that probability changes greatly in tournament play versus casual Wednesday night informal matches at the club. As has been told to you repeatedly on multiple threads, but which you seem reluctant to accept.


Yes I did.

Listen, we are not going to see eye to eye on this. That is fine. You are free to your opinion, that a 200 point difference is massive and the odds of winning are near 2%.

Obviously, 200 points is a big step below, and I don't think they are likely to win with a 200 pts deficit. What is the statistical probability? I still haven't found the correct answer. You say 2%. I posted 2 answers from LLM as a reference, not a definitive truth. It doesn't mean I believe them fully, but they offer a datapoint. I guessed 10-15% for a 200 pt deficit. GPT suggested 20%. Qwen said 9%. So even the LLM's don't agree. So I still dont know the true answer.

Some people have suggested that the point difference doesn't scale up linearly. Is that mathematically true? I don't know because nobody has posted sources.

Based on my own experience, I see 200 pts leaders lose matches they shouldn't lose. This is quite regular. It's called choking. Pros do it in international tournaments. Being in an official tournament doesn't automatically protect you. My opinion is that you are over-estimating the consistency that humans can perform at. As I said, I can't perform to my normal level 20 matches in a row, and I definitely would lose sometimes to 200 lower players.

Now I have 3 data points. You think there is a 2% chance. Fine.
In my own experience, I estimate something like 10-15%, I could be wrong.
LLM suggested 9-20%.

 

[TensorBackhand]

We saw your clip against Victor who is probably around 400 points better than you. To be clear, you think you would win 1 out of 20 matches against him if he is fully trying? Not games. Matches.

Where did I say this?

I haven't played many players 400 pts better than me, so I don't have any sense of my odds. Obviously 400 pts is a huge deficit, so the odds wouldn't be high. I have played many players 200 pts above me, so I have a better sense. As I said, I think it would be in the 10-15% range.

GPT estimated a 5% chance for a 400 pts deficit. I have no idea if that is correct or not. I merely provided GPT's answer.
Qwen cited a much lower probability. I also provided Qwen's answer. I think Qwen is closer to the truth. Qwen also said that a 200 pt deficit is bigger in USATT than it is in Chess. I don't understand the math behind that. I don't know why that would be true, or if that is true.

You cherry-picked GPT's answer, and ignored both my own estimate and Qwen's estimate.

 

[Tony's Table Tennis]

i'm waiting to see who tb suggest that can beat Ma Long at 2800 level.

I know a heck loads of 2800 players, and a normal day for Ma Long (not on his lowest), 50 attempts might not even be enough for 1 win.

 

[Tony's Table Tennis]

Here is one of my players that will appear in MLTT next season and Pro A.
he is about 2800 and the opponent is a top U17 in Taiwan, easily 2600.
How 2600 gets pushed around by 2800..... will be the same when Ma Long toys around with a 2800.
the question is, if they want to use that energy or not.

 

[ThePongCommenter]

Some people have suggested that the point difference doesn't scale up linearly. Is that mathematically true? I don't know because nobody has posted sources.

The only source you need for this are your own eyes - 1000s beating 1200s? Happens all the time. 2300s beating 2500s? Almost never.

Based on my own experience, I see 200 pts leaders lose matches they shouldn't lose. This is quite regular. It's called choking. Pros do it in international tournaments. Being in an official tournament doesn't automatically protect you. My opinion is that you are over-estimating the consistency that humans can perform at.

Did anyone say that being in an official tournament "automatically protects you" from bad losses? I don't think so. But again, the point that has been made to you over and over, is that how a player plays someone far below their level in practice oftentimes isn't the same way they'd play them in a tournament. Maybe some players try their hardest to win every informal practice match but many don't - they use them as opportunities to practice other skillsets, or go on cruise control effort mode.

I will give an example that also goes back to your 49/50 question. I have one friend I play with regularly, at least once a week or sometimes more, who is about 250 points lower than me. Each session, we will typically play several practice matches. I feel absolutely zero threat that he can ever beat me. Even if he were up 2-0, I would still be sure I would win. I'll typically practice a variety of different playstyles, receives, patterns that I need work on etc. just to make the matches somewhat competitive. I can't remember the last time I lost to him and remember we play weekly. Sometimes I will even intentionally put myself in holes to simulate more pressure. Even if it goes to a fifth game, I know that as soon as I start trying 100%, I will win comfortably. That is what it is like. This might sound cocky but it is 100% true and more representative of these situations than you seem to think.

 

[PoppaChubby]

The only source you need for this are your own eyes - 1000s beating 1200s? Happens all the time. 2300s beating 2500s? Almost never.


Did anyone say that being in an official tournament "automatically protects you" from bad losses? I don't think so. But again, the point that has been made to you over and over, is that how a player plays someone far below their level in practice oftentimes isn't the same way they'd play them in a tournament. Maybe some players try their hardest to win every informal practice match but many don't - they use them as opportunities to practice other skillsets, or go on cruise control effort mode.

I will give an example that also goes back to your 49/50 question. I have one friend I play with regularly, at least once a week or sometimes more, who is about 250 points lower than me. Each session, we will typically play several practice matches. I feel absolutely zero threat that he can ever beat me. Even if he were up 2-0, I would still be sure I would win. I'll typically practice a variety of different playstyles, receives, patterns that I need work on etc. just to make the matches somewhat competitive. I can't remember the last time I lost to him and remember we play weekly. Sometimes I will even intentionally put myself in holes to simulate more pressure. Even if it goes to a fifth game, I know that as soon as I start trying 100%, I will win comfortably. That is what it is like. This might sound cocky but it is 100% true and more representative of these situations than you seem to think.

I've played countless of chess tournaments, and indeed a 2300 player is most of the time a FM, FIDE Master, and a 2500 one is sometimes a GM, Grand Master, IM International Master being in between, there are some 2300 IM but most of the time it's players that have lost games and went down from 2400 to 2300. You have to score at least 3 norms at 2400 to be IM, then 3 norms at 2500 (beating 3 guys rated 2500). 3 norms, it does not seem to be a lot, but it shows how difficult it is. I've seen young talents at 2200 beating some old 2500 GMs, exactly like in table tennis, but that's the only times you see those performances with so much differences in ratings.

Being a 1900 at my best it was possible to beat 2000s guys, made it only twice, 2100... never. When I was starting at 1400 I've been beating countless of 1600 guys in my first season, and even 2 1700s. When arriving at 1800 it started being hard, really hard.