This user has no status.
This user has no status.
says
Shoo...nothing to see here. - zeio
Well, before the ban hammer comes, let me play the good but childish member again.
To get the uninitiated minds up to speed (spoken in a holier-than-thou tone), perhaps the better question is:
What's the purpose of a rating system?
To predict the outcome of a future match based on past results.
How do you do that?
Like a scientific experiment, you make some assumptions and come up with a model (hypothesis). This is where we get into the problem as OldUser pointed out - interpretations. For example, how much "memory" of the past do you keep track? How big a "pool" of players do you consider?
In practice, there are actually many variables that can determine a player's strength or affect a match outcome.
The level of the tournaments, bo5, bo7, match score, game score, matchup of different styles or even a particular pair or trio of players (a beats b, b beats c, a beats c?)...
Do you include them? If so, how do you treat and turn them into numbers? If not, how much of an impact does it have on your model?
At that point, you have to turn to approximation. That is, how close to reality can you get with the least amount of variables? And for that, you have to decide the point of diminishing returns.
To get the uninitiated minds up to speed (spoken in a holier-than-thou tone), perhaps the better question is:
What's the purpose of a rating system?
To predict the outcome of a future match based on past results.
How do you do that?
Like a scientific experiment, you make some assumptions and come up with a model (hypothesis). This is where we get into the problem as OldUser pointed out - interpretations. For example, how much "memory" of the past do you keep track? How big a "pool" of players do you consider?
In practice, there are actually many variables that can determine a player's strength or affect a match outcome.
The level of the tournaments, bo5, bo7, match score, game score, matchup of different styles or even a particular pair or trio of players (a beats b, b beats c, a beats c?)...
Do you include them? If so, how do you treat and turn them into numbers? If not, how much of an impact does it have on your model?
At that point, you have to turn to approximation. That is, how close to reality can you get with the least amount of variables? And for that, you have to decide the point of diminishing returns.
Last edited:
This user has no status.
This user has no status.
Well-Known Member
so saying it is obvious is childish?
Where were you? You didn't help at all. You post above doesn't explain anything that isn't already known.
Are you "piling on" after the fact? Now that is childish and even being a bully.
So how do you tern the data in to numbers? Have you ever done it before or are you just "talking big"?
This user has no status.
Well-Known Member
Yesterday I drunk a bit of Slivovica, I said what I said about the community and whatnot... I think NL and USDC understand it, and tried to cheer me by giving me a like to that post, which I actually wanted to delete the next day, even though I actually mean it. Today I was snowboarding with my son, not bad, but then he injured his right wrist a bit. Luckily not broken, will require 10 days rest.
@DukeGaGa Thanks. Levelled without words... But perhaps in another thread if you write something about TDE or compare with other rubber... :- )
@zeio I'm glad you're back. Who am I to be giving opinions? With brokenball, he keeps coming with moot points, and making big theatre about it. Imo, no need to waste too much energy. With USDC, it's different. You should clear things out as men. It will either work or not, but the result is secondary.
@DukeGaGa Thanks. Levelled without words... But perhaps in another thread if you write something about TDE or compare with other rubber... :- )
@zeio I'm glad you're back. Who am I to be giving opinions? With brokenball, he keeps coming with moot points, and making big theatre about it. Imo, no need to waste too much energy. With USDC, it's different. You should clear things out as men. It will either work or not, but the result is secondary.
says
Shoo...nothing to see here. - zeio
Mod threat still in effect: Avoid brokenball
I'll take up on your suggestion. It was short but I had fun. See you again sometime in the future! Take care!latej said:
Does all this number theorizing take in account that as an old offensive style player, that even at a decently high rating, by the end of a gruelling 6 match in an afternoon round robin, you don't have the energy to go against a kid half your rating and can only donate ratings points?
says
Looking for a bat that makes me faster
says
Looking for a bat that makes me faster
Active Member
No, it doesn't care if you break all the bones in your body. If your rating is >238 higher than your opponent then you are expected to win 100% of your game.
On the other hand, a 2000 player could probably beat a 1000 player without moving their feet.
This user has no status.
This user has no status.
Member
says
Spin and more spin.
says
Spin and more spin.
Well-Known Member
Super Moderator
I agree with your point about BrokenBall and I have nothing against Zeio even though I have seen his barbs specifically about me in this thread.
If he actually wanted to understand where I was coming from, I am still here.
This is part of Zeio's misunderstanding about things. I would like you guys not to goad BrokenBall into acting worse than normal. Back then, I was asking for Zeio to cool things down and avoid BrokenBall because as Zeio was instigating, BrokenBall was becoming more and more irrational and upset. Is there really a reason to push someone's buttons and making them irascible?
This is a group forum. There are posts BrokenBall makes that are excellent and others that are totally out of line. Why make him only post the latter because you are set on making him upset and angry. Why not just ignore? As Latej has said, isn't it a waste of time?
If Zeio does this, that will be fine.
This user has no status.
This user has no status.
Well-Known Member
Hopefully we can go back to proper content.
I won't necessary put a direct "no" to Wingtt.
Energy does run out, performance is only as good as the energy you have.
I'm not sure about different countries, but a lot of say, national trials in Taiwan, if you can't win the "first round", you need to grill in up to 20+ matches (best of 5s) in side 3 days to make the national team.
There isn't much rest or recover time, or even time for a proper lunch.
The challenge is beyond physical and mental. There is an energy factor and hunger factor in play.
These national trials to me, isn't ideal, but this is what they been doing for few decades and the fittest surive (while) someone like Feng who is ranked 6x in the world, didn't make the cut due to not peaking during the trials (6 rounds of failing).
Why is such grilling important? Well, for juniors, any one who went to world junior champs can tell you, playing 10 matches (best of 5s) in a day is NORMAL. Also, no time to eat or rest, it is one match done, and basically change your shirt, and you up again (they have doubles and singles on the same day)
Half your rating as Wingtt pointed is maybe far reach.
238 points is nothing though.
1600 vs 1850, or 1750 vs 2000
If the better player is playing at his worse with no energy, versus lower rated player at his best and full of energy, an upset could happen.
I have been involved with a kid, who is 1500, went to the recent USA U11 trials.
Group of 3, highest rated is 2nd seed, around 1900 (has wins against adults at 2100). this kid is 2 of the1500 in that group.
This kid beat the other 1500 convincingly. and against 1900, his mind kind of gave up before the match even began. Few months ago at US open, the 1900 cleaned this kid 3-0, very easily.
This time round, game 5, the kid was 5-0 up. at 5-3 the dad called time out (not a tt player, coach wasn't able to travel with, and only could text the dad "live"). fast forward, 9-6 up.... made a mistake with serve and that kind of killed his momentum and lost 9-11.
A bit sad, but that is life and a good lesson for the kid to come back strong next time.
Coach Gao Jun was pretty impressed seeing a barely 1450/1500 came within 2 points of beating a 1950.
I won't say this kid is 1900 level yet.
Some strokes, he can be 1800~1900, but overall, he is still around 1650~1750.
So this is almost a 400 gap (or 250~300) and it wasn't a walk in the park for the 1900.
Both are 11 years old and I have to say, U11 at 2000, U13 at 2200 is pretty impressive
This user has no status.
This user has no status.
Well-Known Member
This user has no status.
Active Member
I don't like the curve because to me there is no absolute 100% or 0% in winning probability. There should be a better approximation curve based on existing data from USATT.
This user has no status.
This user has no status.
Well-Known Member
There's always a difference between theoretical probability and real life probability, but that's not the reason to dislike the theoretical probability, you just need to keep in mind that it's more of a guide line, like pirate code.
This user has no status.
Active Member
This user has no status.
This user has no status.
Well-Known Member
You are correct! That is why the original curves that hcInnkng made are wrong and graph are wrong.
There is. I have shared it with hcInnkng and how to calculate it. I am waiting for hcInnkng to provide an updated graph and equation. Then all this sh!t will be worth it.
Look at OldUser's post shows the theoretical vs percent win graph. The theoretical graph goes from -1 to 1 but never quite reaches -1 and 1. A ratings difference of 500 should result in a win for the stronger player 99.99+% of the time. The is close to 100% but extremely rare events can occur.
The current USATT tables provide no upside for a strong player with a ratings difference of 238. There are no points to be gained.
David Marcus' Rating Central system is much better.
This user has no status.
This user has no status.
Well-Known Member
It is, but once the data is analyzed and processed it's not real life anymore. For example, we all know a fair coin flips 50/50 heads or tails, but in real life it never is. But how did we get the 50% heads in the first place? We took a coin and flipped it, a lot, then we observed and recorded the data. Up till this point it is still real life data, but then we looked at the data and found it is really close to 50% heads, then we thought "hey, the coin has two sides, if the coin is made so perfectly then we can just say 1 side facing up after a flip is 1 out of 2 possible outcomes, so that is just 50%."
This user has no status.
Active Member
I think you misunderstood my comment. I understand all about real life versus approximation. My comment is just that "the curve predicted 100% win for player with 238 points higher" is not a good prediction because real data will not say that. Somebody tells me he is going to beat me 100% in a match because he is 238 pts above me will be laughed at, not by me but by others, for being too arrogant. Real data will show that such players still can lose a very small percentage, say 1% or 0.5% .. so on. So a curve with polynomial equation may be better and it may exist by curve-fitting the data better.
This user has no status.
This user has no status.
Well-Known Member
I didn't misunderstood, and like I said it's how you interpret the 100%. It's like seeing 100% means you should be saying "you should win all the time" vs. "you'll likely win most of the time". But then "should win all the time " doesn't translate to actually winning all the time.
This user has no status.
Active Member
Ok. I got your point. But I believe there is a better curve for us more number-oriented people.
This user has no status.
This user has no status.
Well-Known Member
There’s always a better graph.
And I’m also a number oriented person, that’s why I have a science and an engineering degree, but I know when to not be number oriented, that’s the key, life isn’t always about numbers.
Similar threads
