Are TLS Ratings that far off?

Discussion in 'Adult League & Tournament Talk' started by JJJ, Sep 27, 2013.

  1. JJJ

    JJJ New User

    Joined:
    Nov 28, 2012
    Messages:
    11
    I looked up some friends that TLS had rated 278 in 2012 they were 3.0's and TLS had them staying at 3.O (in 2012 they were 3.0 C rated players) yet USTA bumped them to 3.5 for 2013. Is TLS that far off? they DID NOT play tournaments that would have moved up their rating either. Just curios if the margin is that far off
     
    #1
  2. J_R_B

    J_R_B Hall of Fame

    Joined:
    Jan 27, 2010
    Messages:
    2,667
    Location:
    Newtown, PA
    In some cases they are. No one should ever use TLS to determine indvidual bumps up or down.
     
    #2
  3. gameboy

    gameboy Hall of Fame

    Joined:
    Jul 10, 2009
    Messages:
    1,620
    How off are we talking about here? If your friend wad rated by TLS as 2.9 or so and then got bumped, you should not be surprised.
     
    #3
  4. ncgator

    ncgator New User

    Joined:
    Oct 14, 2007
    Messages:
    58
    Location:
    Concord, NC
    In general, I've observed that TLS ratings are about lower than Early Start ratings. Several players I know have ES ratings that show them bumped from 4.0 to 4.5, but TLS still shows them as being 4.0 level. I think the ES ratings are more accurate, and TLS ratings are generally a little low...
     
    #4
  5. dizzlmcwizzl

    dizzlmcwizzl Hall of Fame

    Joined:
    Mar 24, 2010
    Messages:
    2,178
    Location:
    DE

    First, ES ratings are definitely more accurate since that is a snapshot of your actual dynamic rating. In fact your dynamic rating is what the TLS ratings are trying to mimic.

    TLS will never be perfect although I think they will do a better job with predictions this year than they did last year.

    And generally I think they do a fair job of showing relative ratings if the players have a large enough sample of matches.
     
    #5
  6. DrewRafter8

    DrewRafter8 Professional

    Joined:
    Jun 1, 2004
    Messages:
    831
    The problem with TLS ratings from what I have seen is that it does not take into account the trickle down from Nationals, Sectionals, and the State tournament. It also doesn't take into account what your rating was from last year. If you were a high 4.0 (3.9) from last year, then it would take less to get you bumped. I still like looking at them, but they are just an estimate.
     
    #6
  7. ace18

    ace18 Rookie

    Joined:
    Aug 7, 2009
    Messages:
    366
    I have a buddy that plays out of 2 districts, same city, one district his TLS for 55 and over is 3.92, his TLS for 18 and over is 3.75. Scratching my head...Does the mean he can have 2 ratings if his 55 up rating went above 4, he could be 4.5 for one league, 4.0 for the other?
     
    #7
  8. J_R_B

    J_R_B Hall of Fame

    Joined:
    Jan 27, 2010
    Messages:
    2,667
    Location:
    Newtown, PA
    No. TLS ratings are not USTA ratings. The USTA just gives you a single rating for all leagues.
     
    #8
  9. schmke

    schmke Professional

    Joined:
    Jan 20, 2010
    Messages:
    1,371
    And TLS calculates a unique rating for a player in each district they play in which is what you are seeing. This inherently reduces their accuracy a bit to a lot as you are seeing with the disparity for the same player.
     
    #9
  10. dizzlmcwizzl

    dizzlmcwizzl Hall of Fame

    Joined:
    Mar 24, 2010
    Messages:
    2,178
    Location:
    DE
    I think this could highlight some interesting sandbagging stories I have heard on this site.

    We do not really have the issue here, but I have read many stories about Houston (?) area players crushing folks during the adult season, getting an early start bump, and then going winless at tri level and getting and to stay on level at year end.

    I think we should have a TT contest to identify the player with the largest disparity in TLS ratings between one league and another. IE ... they are a 4.5 in the summer and a 3.0 in the fall (when it does not matter)..... = 4.0!
     
    #10
  11. schmke

    schmke Professional

    Joined:
    Jan 20, 2010
    Messages:
    1,371
    This might not quite work. You'd need players that play in different districts or areas, not just different leagues. I "think" TLS isn't so bad as to give a different rating for each league.

    But I did look at a particular player from Texas awhile back that claims of sandbagging were made about and wrote this post. http://tt.tennis-warehouse.com/showpost.php?p=6981755&postcount=50

    It was a bit suspicious how low his match ratings were in the fall league matches after doing much better for the most part in normal league play.
     
    #11
  12. dizzlmcwizzl

    dizzlmcwizzl Hall of Fame

    Joined:
    Mar 24, 2010
    Messages:
    2,178
    Location:
    DE
    They give a rating broken down by league and also a summary by district
     
    #12
  13. Jack the Hack

    Jack the Hack Hall of Fame

    Joined:
    Jan 18, 2005
    Messages:
    1,803
    According to TLS, the #1 singles player on my 4.5 team this year has a 4.88 rating, which only puts him 0.13 away from being a 5.5. Yet he lost 1-6, 4-6 in the Local Area finals to somebody rated 4.48, and lost 1-6, 3-6 to somebody rated 4.50 in the Section final. Both matches, he played at the top of his abilities, but the opponents basically blew him off the court.

    In that Section final, one of our opponents is rated by TLS as a 4.40. That guy is in his mid-20s, played college tennis at a D1 school, and has a 24-3 record in 4.5 the past 3 years. His 3 losses were all at 4.5 Nationals, where he has a 5-3 record. By the way, this year will be his 3rd straight trip to 4.5 Nationals. How did he do this, despite also playing at 5.0 and winning matches there the past 2 years? That has to be a case study in the rating system flaws for the USTA to figure out... and I'm curious how TLS has him so low.

    TLS also shows that 5 guys from my 4.5 team are supposed to get bumped to 5.0 this year, and 3 more of us are on the border... yet the ESR only had 1 of us being bumped up. If the ratings that TLS showed were true, we should have killed everyone we played and gone to Nationals. However, we barely won our Local League 3-2 in the final, squeaked through our 1st round at Sectionals in third set breakers, and lost 2-3 in the Section final (which wasn't as close as it sounds because all of our losses were in straight sets and one of our wins was in a third set tiebreaker.)

    We'll see what happens when the Year End Ratings list comes out, but from what I've seen so far, TLS can give you a general sense of where you might be, but doesn't seem to be super accurate. Just guessing, I'd give it a margin of error around 0.25 to 0.35, which is pretty significant.
     
    #13
  14. J_R_B

    J_R_B Hall of Fame

    Joined:
    Jan 27, 2010
    Messages:
    2,667
    Location:
    Newtown, PA
    Playoff matches aren't included in TLS ratings.
     
    #14
  15. bethany2

    bethany2 New User

    Joined:
    Jan 29, 2013
    Messages:
    32
    If you play in multiple districts/sections or make it to any kind of playoff then I think the rankings are pretty inaccurate and not worth looking at. My ratings on TLS in different districts are almost a whole point different! And that doesn't even include playoff runs. Maybe it's different in regions where you can only play in one district and your team doesn't make it to sectionals, but around here it's more just funny to look at (like how did I do SO badly in x county? Or who did I beat there!??).
     
    #15
  16. schmke

    schmke Professional

    Joined:
    Jan 20, 2010
    Messages:
    1,371
    I just had another report customer tell me TLS had them as a 3.2 in one district and a 2.8 in another.
     
    #16
  17. Topaz

    Topaz Legend

    Joined:
    Jun 18, 2005
    Messages:
    6,806
    I played in three separate TLS 'areas' this year (2013).

    I did not advance to post season play in adult, only in mixed, so we can through that out and say I didn't have any post season at all, according to TLS.

    I am currently an NTRP 4.0 B, but it is my first year at 4.0. I play mostly singles, but there are a few doubles matches stuck in there.

    As of Sept 14:

    In area A, TLS has me at 3.26 (3.5)

    In area B, TLS has me at 3.45 (3.5)

    In area C, TLS has me at 3.37 (3.5)

    I did one NTRP tournament in area C, in mixed and women's dubs. The mixed won't count, but the women's dubs will, but I don't think it likely to budge anything much.

    So, it looks like I am on track to get bumped down. Deservedly so, I had a wretched singles season and lost several matches badly to 3.5s playing up.

    The only thing I have left that could affect my rating is trilevel in area C. Not sure it will be enough though.

    (I'm actually among the minority of people who want to move up, not down. Irony....its ironic.)
     
    #17
  18. Topaz

    Topaz Legend

    Joined:
    Jun 18, 2005
    Messages:
    6,806
    I also noticed that TLS has people assigned an age division, though it seems to be their 'upper' cap? For instance, a friend of mine played 18+ and 40+, but her age division is listed as 40+. Are her 18+ results not counted?
     
    #18
  19. J_R_B

    J_R_B Hall of Fame

    Joined:
    Jan 27, 2010
    Messages:
    2,667
    Location:
    Newtown, PA
    All results are counted as far as I can tell, but you are just listed at your highest age level since they can't list multiple.
     
    #19
  20. emilyhex

    emilyhex Rookie

    Joined:
    Apr 30, 2012
    Messages:
    211
    They included them last year, but it took them a while to incorporate it into the ratings. So, it's likely they haven't been added in for 2013 yet.
     
    #20
  21. tennisjon

    tennisjon Semi-Pro

    Joined:
    Jan 8, 2007
    Messages:
    764
    Location:
    West Orange, NJ
    At one point this year, I was the second highest rated player in Delaware, but nearly down to 4.0 level in Eastern because I lost a bad match to two 4.0s. The TLS ratings are per district and you can have completely different results in each district. USTA combines all of that.
     
    #21
  22. 86golf

    86golf Semi-Pro

    Joined:
    Aug 30, 2009
    Messages:
    607
    Location:
    Carolinas
    IMO it's not accurate enough to use for setting line ups...some ppl are pretty far out of place. There is an ex D1 player that has only lost one match at 5.0 in two years and his dynamic is 4.71. I don't think so. I'd like to know how they came up with the algorithm and how they initially assigned someone's self rate?

    It is interesting to look at the statistics/players rankings across your area, state etc. clearly league tennis is dominated by women. You would think more would post in this section....oh right, they're too busy actually playing tennis.

    One observation on the men's side, it appears in some areas that "average" tennis is 4.0 level. What happened? I remember about five years ago I was playing line 2 DBLS and the line 1 guys playing next to me looked like college players. It was actually some pretty good tennis. Now 4.0 is pretty dumbed down.
     
    #22
  23. schmke

    schmke Professional

    Joined:
    Jan 20, 2010
    Messages:
    1,371
    E-mail me who the player is and I'll tell you if my ratings agree or disagree with the 4.71. computerratings@techrunning.com.

    This is looking at all genders, but I did a quick analysis on the number of players at each level by section a bit ago. See http://computerratings.blogspot.com/2013/08/updated-interesting-usta-league-stats.html The mid-point seems to be a mid-3.5 to low 4.0 in most sections.
     
    #23
  24. schmke

    schmke Professional

    Joined:
    Jan 20, 2010
    Messages:
    1,371
    He e-mailed me the 4.71 along with a couple others he thought were wrong on TLS.

    My ratings have the 4.71 expected to be bumped up to 5.0, so if he is right TLS is way off and I have this one right.

    He gave me another that TLS has at 3.84 that he thinks will be bumped to 4.5 and I agree, albeit narrowly, so again if his expectation is right, I have this one correct.

    The third he gave he expects to be bumped to 5.5, and while my ratings don't agree, this player also played with/against self-rated players quite a bit so has 4 matches that I'm not calculating ratings for yet. These are all convincing wins so at year-end when these are included, he could very well be bumped. So I may not have this one right (but TLS doesn't either), but it is because I'm not replicating the year-end calculations in the rating I currently have.

    So we won't know if these 3 are or aren't bumped for just under 2 months, but if his expectation is right, I have 2 of the 3 right that TLS is wrong on and the 3rd has an asterisk.
     
    #24
  25. J_R_B

    J_R_B Hall of Fame

    Joined:
    Jan 27, 2010
    Messages:
    2,667
    Location:
    Newtown, PA
    In last year's ratings, there was a self-rated high school kid on my 4.0 team that TLS had rated at 4.46 (or nearly 5.0!). He wasn't DQ'd or even bumped to 4.5 byt the USTA. He had one great result where he beat some other kid 6-0 6-2 who played both 4.0 and 4.5 (generally won his 4.0 matches and lost every 4.5 match). The rest of his matches were just regular competitive 4.0 results. For whatever reason, TLS blew that one result way out of proportion.
     
    #25
  26. Topaz

    Topaz Legend

    Joined:
    Jun 18, 2005
    Messages:
    6,806
    schmke, if I sent you my info, would you be able to give me an idea of whether or not I'll be bumped down or stay?
     
    #26
  27. schmke

    schmke Professional

    Joined:
    Jan 20, 2010
    Messages:
    1,371
    E-mail me and I'll see what I can do. computerratings@techrunning.com.
     
    #27
  28. Topaz

    Topaz Legend

    Joined:
    Jun 18, 2005
    Messages:
    6,806
    Cool, thanks! Email sent!
     
    #28
  29. BeyondTheTape

    BeyondTheTape New User

    Joined:
    Aug 15, 2013
    Messages:
    38
    Location:
    Southern Section
    On my team, many believe that each local league has a certain quota for players to get bumped up based off the sectional trends. I have some players that i have found their ESR by plugging in their USTA # into an early start league and found that they were not bumped. Some think....well maybe locally they will be bumped.
     
    #29
  30. J_R_B

    J_R_B Hall of Fame

    Joined:
    Jan 27, 2010
    Messages:
    2,667
    Location:
    Newtown, PA
    There's no quota per se but a district's sectional results and a section's national results will filter back to the local level, so the local leagues that do better in playoffs will naturally have more people bumped just the way the system works.
     
    #30
  31. BeyondTheTape

    BeyondTheTape New User

    Joined:
    Aug 15, 2013
    Messages:
    38
    Location:
    Southern Section
    We went to Districts and tied for second. And ESR only has 1 of our guys bumped up...maybe we overachieved hmmm.
     
    #31
  32. J_R_B

    J_R_B Hall of Fame

    Joined:
    Jan 27, 2010
    Messages:
    2,667
    Location:
    Newtown, PA
    Good onya. Next year, sectionals...
     
    #32
  33. BeyondTheTape

    BeyondTheTape New User

    Joined:
    Aug 15, 2013
    Messages:
    38
    Location:
    Southern Section
    We also went to Sectionals and errr got smoked.
     
    #33
  34. BeyondTheTape

    BeyondTheTape New User

    Joined:
    Aug 15, 2013
    Messages:
    38
    Location:
    Southern Section
    I'd like to point out, according to TLS, both a 3.27 and a 3.29 Computer Rated player was bumped up to 4.0. That was their most current rating on TLS. They now rated a 4.0 in an Early Start League.
     
    #34
  35. Orange

    Orange Rookie

    Joined:
    Sep 14, 2011
    Messages:
    209
    It seems that TLS does not include championship rounds in its calculations, but the USTA does include championship rounds when calculating year-end NTRPs.

    It also seems that players who do particularly well in championship rounds have a good chance of being bumped because their opponents are likely to have NTRP ratings that are higher than opponents in regular season play.

    Is this correct?

    If so, why doesn't TLS include playoffs?
     
    #35
  36. J_R_B

    J_R_B Hall of Fame

    Joined:
    Jan 27, 2010
    Messages:
    2,667
    Location:
    Newtown, PA
    That is definitely correct about the USTA. As for TLS, I think someone said they eventually incorporated the 2012 playoff matches, but not until well after the year. I don't know why they have a problem using this information.
     
    #36
  37. catfish

    catfish Semi-Pro

    Joined:
    Jun 22, 2007
    Messages:
    792
    I wonder who does TLS and what they used as a starting rating for each player? There's little info on the TLS website. Is TLS simply a TLS rating, meaning it's not meant to correspond to a person's exact USTA NTRP dynamic rating? Was TLS created to mess with the minds of league players obsessed with ratings? :)
     
    #37
  38. J_R_B

    J_R_B Hall of Fame

    Joined:
    Jan 27, 2010
    Messages:
    2,667
    Location:
    Newtown, PA
    TLS ratings, as far as I know, are meant to replicate the USTA DNTRP calculation, but just like everyone else, they don't have the exact formula, so they can't claim it's an exact match. How well they do varies from person to person.
     
    #38
  39. NLBwell

    NLBwell Legend

    Joined:
    Jun 18, 2004
    Messages:
    7,166
    Remember, there is a cutoff date for when the TLS ratings are calculated. Looks like some people are wondering about matches that occurred after the last TLS posted ratings. I've found them to be a reasonable approximation but certainly not an exact match.
     
    #39

Share This Page