There are 2 sites calculating NTRP ratings currently and I am seeing different rating results for the same player. Which site's numbers do you think is more accurate?
Can you elaborate why do you think that?
I will try. Have no proof at all but anecdotally I will tell you Tennisrecord.com seems to have far fewer ratings that make me scratch my head than TLS. If you want hard data schmke is probably the only one here that could give you that.I could but. I'm just too lazy. Someone else will come along and explain it.
But for my rating Tennis record is actually 14 points higher (.14) than on TLS, this is why I am confused.Just an observation that is probably known already, but Tennisrecord seems to have lower ratings in general.
Disclaimer: I am not affiliated with either site, I don't know anything about them beyond what anyone can see on their sites, and I do my own ratings so these sites are arguably my competition, so take this all for what it is worth. I share it because folks seem to be interested.I will try. Have no proof at all but anecdotally I will tell you Tennisrecord.com seems to have far fewer ratings that make me scratch my head than TLS. If you want hard data schmke is probably the only one here that could give you that.
Will try to do that next time I do a similar analysis.I'd be interested to know in your area what % of folks TLS And TR are within +/- .03 of your ratings.
Again anecdotally I think TR would be a higher %.
Disclaimer: I am not affiliated with either site, I don't know anything about them beyond what anyone can see on their sites, and I do my own ratings so these sites are arguably my competition, so take this all for what it is worth. I share it because folks seem to be interested.
However, I have done some analysis and have made every effort to ensure it is objective and so I believe these are facts about those sites that anyone could verify if they were as tennis ratings crazy as I am and put the time in to do it.
First, one has to define "accurate". Accurate could be defined to be the most accurate representation of a player's ability, or it could be the most accurate accounting of a rating reflecting their accomplishments, or it could be the most accurate predictor of the USTA's dynamic rating and/or year-end levels. Think of college football, should the #1 team be the team most likely to win on a neutral field against other opponents? Or the one that has managed to compile the more impressive resume to date? Those are not necessarily the same team. Similarly, all tennis ratings are not created equal as they may have different goals.
However, for purposes of this discussion, I'll assume "accurate" to mean most closely matching the USTA's rating and level. Note that I do not know if either site's motivation or goal is to match this so keep that in mind.
The challenge is the USTA only discloses year-end levels so that is all we can really check. And it should be pretty easy to predict year-end levels as the the range for each level is a full 0.5. Throw in that about 80-85% of players stay the same level each year so one could simply predict that every player would stay the same level and they'd be at least 80% correct. So 80% is sort of the baseline.
Now, on to some numbers. And again, let me reiterate that with respect to these two sites, I'm just a guy looking at the numbers and doing my best to do objective analysis and calculations. I have also not done comprehensive checks, but just done spot checks or checked specific areas so these numbers may not represent their accuracy overall. And while I don't think there are errors in my analysis, I am human.
Using TLS's posted lists, I have done some analysis of accuracy and generally found them to be accurate in predicting the USTA year-end levels in the low-70's percent of the time. Specifically, for my area (Northwest Washington), in 2015 it was 72% and 2016 74%. Looking by level, this statistic is fairly consistent, in the same area in 2016 ranging from a low of 71% for 2.5s to 77% for 4.0s and 4.5s, and for the typically most common level of 3.5, they are 75%.
I have not done a similar check for my area for TennisRecord, but have done a bunch of spot checking, probably a couple hundred players in a variety of areas, and when I've done this have found them to be accurate in the 70's% of the time as well. I may check a player there and also check their partner and 6-8 opponents and they usually miss on 2-4 of the 10 or so I check. This is far less scientific so take it worth a grain of salt, but that has been my observation. I encourage others to do their own checks.
So none of this is conclusive or comprehensive, but it is what I've observed. Make whatever conclusions you wish.
The natural question may be what about my ratings? Overall in 2016 they were at 85% and for that most popular 3.5 level were at 89%. But I know my ratings aren't perfect and aren't going to be exactly right, I generally figure I'm within a few hundredths for the majority of players, so if one allows a few hundredth margin of error my accuracy goes up to 93% overall and 95% for the 3.5 level. 2016 was a pretty representative year but it has been higher in some years, but also lower in years where the USTA made big adjustments that I had no way of predicting.
There you have it, use and consider as you see fit.
Now that you've identified yourself, I say kudos for playing so many tennis matches!TLS has me at 4.43 on LI 4.60 in Manhattan, TR has me at 4.46.
Not including playoffs or sectionals.
J
Now that you've identified yourself, I say kudos for playing so many tennis matches!
Another way to check accuracy and one that I do fairly regularly is see how well DQ's are predicted.Disclaimer: I am not affiliated with either site, I don't know anything about them beyond what anyone can see on their sites, and I do my own ratings so these sites are arguably my competition, so take this all for what it is worth. I share it because folks seem to be interested.
However, I have done some analysis and have made every effort to ensure it is objective and so I believe these are facts about those sites that anyone could verify if they were as tennis ratings crazy as I am and put the time in to do it.
First, one has to define "accurate". Accurate could be defined to be the most accurate representation of a player's ability, or it could be the most accurate accounting of a rating reflecting their accomplishments, or it could be the most accurate predictor of the USTA's dynamic rating and/or year-end levels. Think of college football, should the #1 team be the team most likely to win on a neutral field against other opponents? Or the one that has managed to compile the more impressive resume to date? Those are not necessarily the same team. Similarly, all tennis ratings are not created equal as they may have different goals.
However, for purposes of this discussion, I'll assume "accurate" to mean most closely matching the USTA's rating and level. Note that I do not know if either site's motivation or goal is to match this so keep that in mind.
The challenge is the USTA only discloses year-end levels so that is all we can really check. And it should be pretty easy to predict year-end levels as the the range for each level is a full 0.5. Throw in that about 80-85% of players stay the same level each year so one could simply predict that every player would stay the same level and they'd be at least 80% correct. So 80% is sort of the baseline.
Now, on to some numbers. And again, let me reiterate that with respect to these two sites, I'm just a guy looking at the numbers and doing my best to do objective analysis and calculations. I have also not done comprehensive checks, but just done spot checks or checked specific areas so these numbers may not represent their accuracy overall. And while I don't think there are errors in my analysis, I am human.
Using TLS's posted lists, I have done some analysis of accuracy and generally found them to be accurate in predicting the USTA year-end levels in the low-70's percent of the time. Specifically, for my area (Northwest Washington), in 2015 it was 72% and 2016 74%. Looking by level, this statistic is fairly consistent, in the same area in 2016 ranging from a low of 71% for 2.5s to 77% for 4.0s and 4.5s, and for the typically most common level of 3.5, they are 75%.
I have not done a similar check for my area for TennisRecord, but have done a bunch of spot checking, probably a couple hundred players in a variety of areas, and when I've done this have found them to be accurate in the 70's% of the time as well. I may check a player there and also check their partner and 6-8 opponents and they usually miss on 2-4 of the 10 or so I check. This is far less scientific so take it worth a grain of salt, but that has been my observation. I encourage others to do their own checks.
So none of this is conclusive or comprehensive, but it is what I've observed. Make whatever conclusions you wish.
The natural question may be what about my ratings? Overall in 2016 they were at 85% and for that most popular 3.5 level were at 89%. But I know my ratings aren't perfect and aren't going to be exactly right, I generally figure I'm within a few hundredths for the majority of players, so if one allows a few hundredth margin of error my accuracy goes up to 93% overall and 95% for the 3.5 level. 2016 was a pretty representative year but it has been higher in some years, but also lower in years where the USTA made big adjustments that I had no way of predicting.
There you have it, use and consider as you see fit.
@schmke has the ratings that are closest to the actual USTA computer, but he doesn't put them out en mass for free on a website.
@schmke has the ratings that are closest to the actual USTA computer, but he doesn't put them out en mass for free on a website.
Rightfully so!
J
There are 2 sites calculating NTRP ratings currently and I am seeing different rating results for the same player. Which site's numbers do you think is more accurate?
Both are wrong eek, you aren't ok. Well, to be fair, 4.43 is certainly low IMHO, but you aren't way over the threshold either. Which means year-end calculations could drop you below, but right now I'd lean towards you being bumped up.It's interesting to see the differences. TLS has me at a 4.43, but doesn't have my district or sectional results in there. Of which I went 5-0. So, I would assume it'll go up. TR has my rating at 4.427 (pretty close to TLS), but it does include my district/sectional matches.
I'm concerned that I'm on the cusp of getting bumped based on my record for 2017 (15-2). And the results from post-season play doesn't help. However, the 4.43(ish) prediction makes me think I'm okay.
Sucky. Thanks for the input though.Both are wrong eek, you aren't ok. Well, to be fair, 4.43 is certainly low IMHO, but you aren't way over the threshold either. Which means year-end calculations could drop you below, but right now I'd lean towards you being bumped up.
Sucky. Thanks for the input though.
3-0 with a three setter in there. (2-0 at the district playoff, two three setters). Other two matches I got lucky and got their weaker player/team. I know I'm on the verge, but I probably wouldn't win a match at 5.0 unless it's against another bumped-up guy.You went 5-0 at sectionals and want to stay at your current level???
I know I'm on the verge, but I probably wouldn't win a match at 5.0 unless it's against another bumped-up guy.
And that guy almost always gets bumped back down. I play this game with the USTA each year. I get bumped every few years, appeal, get denied, go play a few 5.0 matches, lose, come back down.At every level, someone has to be the guy that won't win many matches.
And that guy almost always gets bumped back down. I play this game with the USTA each year. I get bumped every few years, appeal, get denied, go play a few 5.0 matches, lose, come back down.
If I do get bumped, I'll just sit out. As mentioned, the only reason play is for the competitive matches. While I'm a good 4.5, all my matches are competitive. I like all the guys I play. We almost always have close, well-played matches. That's what makes it fun. I have little interest in going out every week to play against guys I don't really know just to lose 1 and 2.
^^ This is me.There are two kinds of 4.5/5.0 players imo.
The first kind is very solid and consistent and all but the other elite 4.5 players just can't do anything against them, but at 5.0 they simply don't have enough game.
^^ This is me.
A lopsided win for me at 4.5 is probably 2 and 3. Most of my matches are more like 4 and 5, and looking this year, 30% of my matches went 3 sets. So, even though I win more often than not, almost all of them are competitive. Which is what the USTA should be striving for. As you mentioned, since I struggle to make matches better than 2 and 2 at the 5.0 level, what's the point of playing?
I have a number of friends that have fallen into this same category. They win 9/10 matches. All are competitive. They get bumped to 5.0. They never play USTA again.
If the USTA would allow for some overlap (which would require transparency in ratings), then you could have a guy who is a 4.592 playing at 4.5, but his court position would be dictated by his rating and/or doubles partner evening him out. This would, theoretically, ensure even more competitive matches. They sort of do it in 40+, where they allow two 5.0 players to play, but those players HAVE to play on court 1 (if they play together in dubs), or on 1 and 2 in singles. It's kind of a watered down version of what I would suggest. But, while I feel it's a good step in the right direction, it's only at 40+.
Anyway, I digress. Bottom line is I agree.
Ahhh, yes. You're right. Only one 5.0 can play singles. My bad. I'm not 40 for another 6 months, so I'm not terribly well-versed in that leagueAre you 100% sure two 5.0s can play 1&2 singles in 40+ 4.5+? Here they definitely can't. 5.0s have to be on ct. 1 so you could put both in 1 dubs or one in 1 singles and the other at 1 dubs but never can a 5.0 be on 2.
J
Ahhh, yes. You're right. Only one 5.0 can play singles. My bad. I'm not 40 for another 6 months, so I'm not terribly well-versed in that league
...I just drink beer and heckle from the sidelines.
J
Just an observation that is probably known already, but Tennisrecord seems to have lower ratings in general.
Yeah. Just for comparison, there's a team I know that went to a Sectional (and lost there). TennisRecord has 3 of their players getting bumped up a level. TLS has 5.I was surprised to see that most of the teams we faced in 3.5 districts fielded entire lineups filled with guys in the 3.1-3.3 range according to that site. Many of them looked like strong 3.5s to me. It's hard to believe you could make districts let alone win with lineups made of below-average to average 3.5s.
I know folks are fond of calculating "4.5fh, 3.5bh, 3.5serve == 3.8333"
we should be considering "1.0 conditioning" of most folks
lol, pet peeve of mine...+1
A simple clinic drill showed me that I'm not in as good a shape as I thought: instructor faces student while the student is on the BL. Instructor tosses ball out near the alley. Student hits a GS. Do 10 times. If I have to move to the alley [or beyond] every time, I get tired and my technique [as it were] starts to break down.
lol, pet peeve of mine...
when doing clinic drills, we do king of the hill style drills, starting with a feed.
many times, the drill calls for you to start at or even in the alley (ie. one foot in)... (ie. and a ball is fed to the opposite corner)
really irritates when folks start getting tired, they start creaping toward the the center hash
ie. winning the drill, is more important then getting the full benefits of the drill (ie. running fh/bh wide)