Handicap System based on UTR difference?

[sfpatrice]

Greetings all,

Has anyone heard of a handicap system for tennis based on the difference in UTR rating?

Given how UTR is computed (number of games won during the match basically), I think it's probably possible to compute an expected game total outcome given 2 players and their ratings. How to translate that into handicap games in a set and/or handicap points in a game, that's the tricky bit..

UTR - use to handicap a match like golf? basically suggests trial and error, I am hoping someone went further..

[Edit]: I found this dense statistical paper: https://www.columbia.edu/~rs3566/papers/2018tennis.pdf

From the conclusion: For a three-set match, we found that a one point difference in UTR between two players translated to 10-12 handicap points

 

[TennisOTM]

I found this dense statistical paper: https://www.columbia.edu/~rs3566/papers/2018tennis.pdf

From the conclusion: For a three-set match, we found that a one point difference in UTR between two players translated to 10-12 handicap points

Interesting - I was not aware that research existed.

It looks like the same authors had previously proposed a method for using what they call handicap points, which was to allocate them to the underdog to use whenever they want in the match. They use them by choosing to claim the next point of the match without playing it. So if you have 12 handicap points, you can claim any 12 points of the match you want. You could claim the first 12 points and start the match up 3-0, or save them for break points, or whatever strategy the underdog wishes.

So their result says that for opponents with a 1.0 UTR difference, the underdog should get around 12 handicap points for a three-set match, and that should make it an even 50/50 matchup for the match win. Do we believe it?

 

[sfpatrice]

Here is another bit of research I found: http://datagenetics.com/blog/august12018/index.html

Not directly applicable, but it basically is a way to map the difference in expected percentage of points win to expected percentage of games win, and then to matches win.

The headline which I think we already know is that a very small percentage of delta of winning points quickly results in close certainty of winning the match.

p	Game	Set	Match
0.40	0.264271	0.036567	0.000463
0.41	0.285501	0.053368	0.001401
0.42	0.307438	0.075782	0.003872
0.43	0.330020	0.104746	0.009763
0.44	0.353182	0.141002	0.022439
0.45	0.376851	0.184958	0.047018
0.46	0.400952	0.236570	0.089862
0.47	0.425403	0.295262	0.156868
0.48	0.450120	0.359895	0.250730
0.49	0.475015	0.428821	0.368331
0.50	0.500000	0.500000	0.500000
0.51	0.524985	0.571179	0.631669
0.52	0.549880	0.640105	0.749270
0.53	0.574597	0.704738	0.843132
0.54	0.599048	0.763430	0.910138
0.55	0.623149	0.815042	0.952982
0.56	0.646818	0.858998	0.977561
0.57	0.669980	0.895254	0.990237
0.58	0.692562	0.924218	0.996128
0.59	0.714499	0.946632	0.998599
0.60	0.735729	0.963433	0.999537

 

[sfpatrice]

Interesting - I was not aware that research existed.

It looks like the same authors had previously proposed a method for using what they call handicap points, which was to allocate them to the underdog to use whenever they want in the match. They use them by choosing to claim the next point of the match without playing it. So if you have 12 handicap points, you can claim any 12 points of the match you want. You could claim the first 12 points and start the match up 3-0, or save them for break points, or whatever strategy the underdog wishes.

So their result says that for opponents with a 1.0 UTR difference, the underdog should get around 12 handicap points for a three-set match, and that should make it an even 50/50 matchup for the match win. Do we believe it?

Intuitively sounds like too many joker points to me.. but I have no data to base this on.

 

[TennisOTM]

Intuitively sounds like too many joker points to me.. but I have no data to base this on.

Hmm my intuition was that it sounded like too few points. 1.0 UTR difference is pretty big - if I face someone 1.0 higher than me I'm probably losing 6-2, 6-2 or something like that. Would using 12 free points be enough to make me even? Maybe. I would guess that I'd need to use them all up to win a close first set, then I'd be toast. Seems like this assumes a full third set as well.

 

[esgee48]

How did the hypothesis make adjustments for data that is bad, less than 100%? What happens if is not current? BITD, at a club we use to use something similar but was based on NTRP. It did not work out well for the lower level player. Higher level player went all out to assure that you could only even out a game score. So they quickly jumped out to a 3-0 or 4-0 lead. So we had 8 points max to claim. Wasn't pretty to have a B level (4.5/5.0) vs a C (3.5/4.0). A levels were excluded. But UTR difference of 1 point is not a full NTRP level higher.

 

[nyta2]

Hmm my intuition was that it sounded like too few points. 1.0 UTR difference is pretty big - if I face someone 1.0 higher than me I'm probably losing 6-2, 6-2 or something like that. Would using 12 free points be enough to make me even? Maybe. I would guess that I'd need to use them all up to win a close first set, then I'd be toast. Seems like this assumes a full third set as well.

How did the hypothesis make adjustments for data that is bad, less than 100%? What happens if is not current? BITD, at a club we use to use something similar but was based on NTRP. It did not work out well for the lower level player. Higher level player went all out to assure that you could only even out a game score. So they quickly jumped out to a 3-0 or 4-0 lead. So we had 8 points max to claim. Wasn't pretty to have a B level (4.5/5.0) vs a C (3.5/4.0). A levels were excluded. But UTR difference of 1 point is not a full NTRP level higher.

i'm not a fan of the "N free points to allocate anywhere" as it doesn't simulate the score board pressure you feel in tennis...

there's a drill i've done with say 8 people...
* you're matched against an opponent
* everyone else (6) gets to vote on what the score will be
* then the avg is what the "goal posts" are..
eg. so let's say i'm favored to win by 4... in a typical game to 11... i win if i reach eleven before my opponent reaches 7...
and then we play for something (pushups, laps, etc...)

the closest i've seen to somthing like that drill, is the french/belguium systems... where you get allocated games/points based on the level difference...
eg. if a few levels above, maybe i always start 0-2 (games) 0-40 on my games
which to me, simulates scoreboard pressure,...
but i've never figured out the exact rules to be able to implement something like that...

 

[sfpatrice]

i'm not a fan of the "N free points to allocate anywhere" as it doesn't simulate the score board pressure you feel in tennis...

I agree - you get to use it on high pressure points and the ability to resist pressure is a major component of tennis ability..

For what it's worth, Even though the ranks in the French system are those that came from the old handicap calculation (e.g 30/2 meant you start at 30-love for 2 games - https://www.team-tennis.fr/classeme...ent-au-tennis/origine-du-classement-au-tennis) the actual handicap hasn't been used in a long while

Given the UTR system is based on expected game percentage wins, I think it would make sense to have whatever handicap based on a number of games per set, so instead of points in a game as in the French system, you would start with a number of games in the set

 

[nyta2]

I agree - you get to use it on high pressure points and the ability to resist pressure is a major component of tennis ability..

For what it's worth, Even though the ranks in the French system are those that came from the old handicap calculation (e.g 30/2 meant you start at 30-love for 2 games - https://www.team-tennis.fr/classeme...ent-au-tennis/origine-du-classement-au-tennis) the actual handicap hasn't been used in a long while

i know, but if had a complete chart of levelX vs. levelY, i would have tried to run toruney that way...

Given the UTR system is based on expected game percentage wins, I think it would make sense to have whatever handicap based on a number of games per set, so instead of points in a game as in the French system, you would start with a number of games in the set

but i think for sufficiently large gaps in levels, it will be impossible for say a ntrp3.0 to win 4 poitns in row against an ntrp4.5...
but if say, the ntrp4.5 always had to start -30 to +40... that's alot of pressure (eg. the ntrp has 5 chances to win 1 point to win the game)... i'm thinking i'd naturally play more cautiously,... maybe enough where it's competitive for both?

 

[TennisOTM]

i'm not a fan of the "N free points to allocate anywhere" as it doesn't simulate the score board pressure you feel in tennis...

Have you tried it? I have not, but I would imagine it could create pretty big score board pressure. If I was the favorite and my opponent had some free points to use, then I would feel a lot of pressure to not allow them to get to break point on my serve, as they would likely cash in a freebie to complete the break. So any time my opponent gets to 30 on my serve then that's basically a break point. Same thing towards the end of a close set - if I let them get within a few points of winning a clinching game or tiebreak, then they'll likely cash in. That's a lot of pressure trying to keep them from getting close.

But again, I haven't tried it so I don't know - if you did, then why was it different than what I described?

 

[TennisOTM]

How did the hypothesis make adjustments for data that is bad, less than 100%? What happens if is not current?

In the research paper that used UTR, they said they only used rated players that had the 100% confidence on UTR. They used pretty high rated rec players only within a reasonably small range (UTR 7-9). If you needed to handicap someone with bad or small amounts of rating data than there's only so much you can do - that's gonna be a problem for any system.

 

[JK208]

Fundamentally, a handicap system will not make for more enjoyable matches. When I play against some guys who are UTR 10.5+, our rally ball is different enough that it isn't enjoyable for either of us. Granted, if I were to get 15ish points to use whenever over the course of the match, I'm pretty confident that I would at least take them 3 sets, if not win. The way I see it is game 1, they serve, we get to 30-30 (pretty common for us, especially if they make a random error or two), I use 2 points to seal it, then I keep using 1-2 points every service game to hold it out for the win. Did the match look competitive? Yes it did. Did either of us enjoy the match? Not really- Waiting for a few lucky errors then using my handicap is no fun. I'd rather play a guy closer to my level, lose 2 and 2, but have some fun rallies and go for some big shots.

 

[sfpatrice]

How far apart are you from your opponent then? No handicap can bridge a big difference.. but if you are say a point apart it could make the match more interesting IMO. 2 points or more that's no fun no matter what probably..

 

[TennisOTM]

Fundamentally, a handicap system will not make for more enjoyable matches. When I play against some guys who are UTR 10.5+, our rally ball is different enough that it isn't enjoyable for either of us. Granted, if I were to get 15ish points to use whenever over the course of the match, I'm pretty confident that I would at least take them 3 sets, if not win. The way I see it is game 1, they serve, we get to 30-30 (pretty common for us, especially if they make a random error or two), I use 2 points to seal it, then I keep using 1-2 points every service game to hold it out for the win. Did the match look competitive? Yes it did. Did either of us enjoy the match? Not really- Waiting for a few lucky errors then using my handicap is no fun. I'd rather play a guy closer to my level, lose 2 and 2, but have some fun rallies and go for some big shots.

I might be wrong but I wonder, if you actually tried it, your opponent would be forced to change up his shotmaking to avoid letting you get to 30 as often as you do now. It could force your opponent to slow down or simplify his rally ball to avoid making the errors that commonly get you to 30. Then you'd perhaps have more fun rallies. Not sure it'd work out that way, just food for thought.

 

[JK208]

I might be wrong but I wonder, if you actually tried it, your opponent would be forced to change up his shotmaking to avoid letting you get to 30 as often as you do now. It could force your opponent to slow down or simplify his rally ball to avoid making the errors that commonly get you to 30. Then you'd perhaps have more fun rallies. Not sure it'd work out that way, just food for thought.

There's only one way to know. I'll see if any of my buddies are back home for spring break and try to play a match with this setup. Not sure they'd enjoy it- I know I've tried something like this going in the opposite direction and I didn't really like it.

 

[TennisOTM]

There's only one way to know. I'll see if any of my buddies are back home for spring break and try to play a match with this setup. Not sure they'd enjoy it- I know I've tried something like this going in the opposite direction and I didn't really like it.

Well I suppose the fact handicapping in tennis has never caught on anywhere (as far as I know) is decent evidence that it doesn't work very well. I do feel like it could work, but it seems tough to get it right.

 

[Roforot]

free points can make the score look closer, but its not fun for either player since its more like a cards game strategy. If you use your points at 30-30, do you really feel the same as breaking someones serve?

Other things Ive done are things like on points to close out a game (40-anything) the favored player gets only 1 serve. The underdog on the other hands gets an extra serve in those situations. This works well if theres just o.5 NTRP difference but the favored player always won, but it felt closer.

Other ideas would be something like the favored player cant hit a winner on his first two touches. Or the favored player cant hit drop shots unless he grts drop shotted first.