UTR Admits Change In Algorithm—Severely Hampers Ability To Raise Rating

CiscoPC600

Professional
As @texrunner feared, UTR has changed its methodology/algorithm. I don't understand the logic in excluding upsets or close matches with UTR differences more than two--especially in a tournament setting. I could understand the logic for self-reported matches in order to avoid match fixing but the likelihood of the same issue in a tournament setting is much lower. Personally, I believe this move severely hampers everyone across the board: rec players, juniors, and high school students competing for college scholarships/offers. As a rec player, I don't want to have to compete in substantially more lower level tournaments for marginal-to-no increases in UTR. It costs way too much money and time. I can't imagine how someone with more on the line feels.

The silver lining is UTR seems willing to communicate. I encourage everyone to leave a comment asking them to re-consider their position: https://support.myutr.com/support/tickets/new

 
Last edited:
Why is this bad? The earlier rule was 2.5 point difference, with close matches (rare indeed) possibly being counted. So how does this hurt us? The goal of utr is to get an accurate rating of match performance in the future, the algorithm must be adjusted as they learn more about it, and to counter any gamesmanship. I don't see we don't just play and let the utr rate accordingly.
 

MathGeek

Hall of Fame
I'd rather see all matches included. If needed a weighting factor could be used to reduce the impact of matches with larger UTR spreads.

But I also think the above claim that the change "severely hampers ability to raise rating" is exaggerated. In tournament play, it is rare to be paired with someone with a UTR more than 2 from one's own and it is even rarer to beat them or perform well enough against them to earn a significant UTR bump. There are plenty of opponents within 2 of one's UTR and these are the opponents one is most likely to be matched against in most UTR events and most USTA events.

Reviewing the tournaments my son and/or I have played in we've never had a match against an opponent with UTR more than 2 different from our own.
 

HunterST

Hall of Fame
I thought only including the result if it was close made a lot of sense. I can see how factoring a 6-0 6-1 loss to a much higher rated opponent might incorrectly bump the losing player up too much. If the loss is 6-4, 6-3, though, that indicates something about the abilities of the players.

The only explanation I can think of is they're concerned there must be some other factors at play for a match between two players of such disparate levels to be competitive. If a 10 loses to an 8, for example, it's not unlikely that the 10 could have had an injury, have been sick, etc. Then again, that could be true of any match.
 

CiscoPC600

Professional
Why is this bad? The earlier rule was 2.5 point difference, with close matches (rare indeed) possibly being counted. So how does this hurt us? The goal of utr is to get an accurate rating of match performance in the future, the algorithm must be adjusted as they learn more about it, and to counter any gamesmanship. I don't see we don't just play and let the utr rate accordingly.
The reason it hurts you is because you have to play more matches to make incremental changes. One example, I got back into competitive tennis last year and played a tournament in a lower division because I was out of shape and rusty. I lost in the second round. I have now been hitting/training and played in a higher division and won my first round then took the 5th seed to very close sets. However, that match won't count simply because currently his UTR is probably 2.0-2.5 higher than mine. In other words, I have to go back to the lower division and play tons of matches there, which isn't appealing because of time and money. I've seen other players at that range and they play every local tournament do well and still have the same UTR.

I could see the same for a junior or other player than makes a surprise and deep run in a tournament then the semifinals or finals he faces and upsets or has close matches with someone 2 UTR higher. It won't count and it's astonishing to me that it's give 0 weight.
 
My guess is the algorithm doesn't support weighing results between players of vastly different ratings (more than 2 is a great deal). Utr wouldn't delibaterly refuse to count valid results just to take credit away from you personally. Understanderably you want credit for beating the top seed but the algorithm would be rendered inaccurate by such results. The algorithm must come first, otherwise utr will then actually be worthless. I'm giving utr the benefit of the doubt since their final product so far seems highly predictive.

Despite the annoyance of coaches using utr as the sole proxy of skill, I think the real Customer is not you or I, the player, but rather the tournament organizers and coaches.

Utr may be faulty in some ways but it seems far more accurate than ntrp numbers or even rankings, and I myself am using it to look up players to get a sense of their skill level when needed.
 
The UTR is pretty much worthless, IMHO. It is very much like the old version of college football BCS system where the computer picked the two teams to play for college national championship.

It reminded me of the 2013 NBA draft where Gianis Adetokumpo was drafted in the 15th spot. There were 14 players drafted ahead of him and yet Adetokumpo is now a bonafide superstar. Most of the 14 players that were picked ahead of him are either not in the NBA anymore or benched players.
 

andfor

Legend
Don't like your own, your kids, or you students rating/rankings. Compare it to another non-tennis related highly criticized system, then site an aberration from basketball and call it flawed. Someone might want to sight some evidence based on real results and present it to UTR. Might also want to show the evidence of the flaw they imply and show us here to build some credibility.
 
Last edited:
Didn’t football replace the stupid BCS system where the computer used algorithm to spit out two teams that will play for the National championship game? The system was so flawed that they finally replaced it with the College Football Playoffs in 2014. Seem to me that the UTR also uses algorithm that no one has any ideas how it work.
 

CiscoPC600

Professional
My guess is the algorithm doesn't support weighing results between players of vastly different ratings (more than 2 is a great deal). Utr wouldn't delibaterly refuse to count valid results just to take credit away from you personally. Understanderably you want credit for beating the top seed but the algorithm would be rendered inaccurate by such results. The algorithm must come first, otherwise utr will then actually be worthless. I'm giving utr the benefit of the doubt since their final product so far seems highly predictive.

Despite the annoyance of coaches using utr as the sole proxy of skill, I think the real Customer is not you or I, the player, but rather the tournament organizers and coaches.

Utr may be faulty in some ways but it seems far more accurate than ntrp numbers or even rankings, and I myself am using it to look up players to get a sense of their skill level when needed.

How would it be rendered in accurate? It should be able to appropriately handle an upset or close match. The winning player still advances to the next round so the algorithm will have another match to figure out how to rate the player that struggled with a lower rated player.
 
Didn’t football replace the stupid BCS system where the computer used algorithm to spit out two teams that will play for the National championship game? The system was so flawed that they finally replaced it with the College Football Playoffs in 2014. Seem to me that the UTR also uses algorithm that no one has any ideas how it work.
What was it "stupid"? Because some people disagreed with it? It was system well-designed to deal with the problem we face in tennis as well: every player/team can't play every other one. So how do we know their true value?

Why would you think UTR doesn't know how their own algorithm works? They just aren't revealing everything, or the individual doesn't like the results.
 
How would it be rendered in accurate? It should be able to appropriately handle an upset or close match. The winning player still advances to the next round so the algorithm will have another match to figure out how to rate the player that struggled with a lower rated player.
Don't know the details but UTR has no incentive to make their algorithm inaccurate; or they want to discourage lopsided contests, or view them as without predictive value. (Would you like to have to play lower UTR ranked players in which you can only lose rating, never gain any?)

My other guess it may have to do with the purpose of encouraging competitive matches.

https://webcache.googleusercontent.com/search?q=cache:RFE36WCYKwwJ:https://blog.universaltennis.com/2017/07/25/three-ways-to-win-competitive-routine-decisive/+&cd=12&hl=en&ct=clnk&gl=us
 
Does UTR take into consideration that a UTR 10 player lost to a UTR 12.1 player 6-0 6-0 but the UTR 10 player had multiple break points on in all serving games by UTR 12.1 player but UTR 10 could not convert those points. Also, UTR 10 has all his serving games at either 40-15 or 40-30 but couldn’t close out the game. Yes on paper, it is a rout but UTR 12.1 player should have lost. He could have 6-0 6-0 in favor of the UTR 10 player.

Does UTR take that into consideration?
 

3kids

Rookie
Does UTR take into consideration that a UTR 10 player lost to a UTR 12.1 player 6-0 6-0 but the UTR 10 player had multiple break points on in all serving games by UTR 12.1 player but UTR 10 could not convert those points. Also, UTR 10 has all his serving games at either 40-15 or 40-30 but couldn’t close out the game. Yes on paper, it is a rout but UTR 12.1 player should have lost. He could have 6-0 6-0 in favor of the UTR 10 player.

Does UTR take that into consideration?
Just curious how do you propose UTR or any rating system take that into consideration?
 
The UTR idea is that over time these "I should have won" will balance out; eventually you will win sufficiently and your true rating revealed (and constantly updated). Even if your 10's UTR 2.10 daughter takes UTR 16.34 Nadal to 5 sets but can't convert and gets not a tenth of a point for the effort.

The problem is that I don't see a solution to the coaches using UTR as the only real qualification needed. Even I've started to UTR everyone to see their rating. So I see the problem and source of many of the complaints/objections, especially those trying to get into a college team. But the alternative might just be the same thing, just based on regional rankings, which has its own biases.
 

ClarkC

Hall of Fame
Just curious how do you propose UTR or any rating system take that into consideration?
I think he was being facetious in criticizing the current UTR system that counts how many games you won and not just who won and lost.
 
Last edited:
As @texrunner feared, UTR has changed its methodology/algorithm. I don't understand the logic in excluding upsets or close matches with UTR differences more than two--especially in a tournament setting. I could understand the logic for self-reported matches in order to avoid match fixing but the likelihood of the same issue in a tournament setting is much lower. Personally, I believe this move severely hampers everyone across the board: rec players, juniors, and high school students competing for college scholarships/offers. As a rec player, I don't want to have to compete in substantially more lower level tournaments for marginal-to-no increases in UTR. It costs way too much money and time. I can't imagine how someone with more on the line feels.

The silver lining is UTR seems willing to communicate. I encourage everyone to leave a comment asking them to re-consider their position: https://support.myutr.com/support/tickets/new

Nate Silver would not approve.

There is significant predictive value in blowouts. Removing them from the algorithm altogether just reduces the overall accuracy.

A better approach would be to use the ELO system 538 uses, which gives diminishing returns as margin of victory increases, but doesn’t toss out any data points.
 

jcgatennismom

Professional
I do find it interesting that twitter announcements of college team new freshmen or transfers now list players' "best " UTR instead of current UTR as many are down .7 or .8 from highest UTR. Case in point Justin Boulais new recruit for OSU with a best UTR of 13.69, currently 12.99. Obviously this year there have been several algorithm changes.

The 2.0 spread hurts players who play rec tennis as well as tournaments, fast-improving high school tennis players, former college players trying to get back in the sport. An ACC player who graduated in 2018 was ranked 13.1 in June '18. His college matches dropped off, he played some rec tennis at 5.0 level, and his UTR dropped to 11.06-2 UTR points down in 12 months. However, he played two tournaments vs current collegians in June, and now his UTR is back to 12.15. Clearly his level had not dropped 2 UTR points, just his level of competition. Now he wasnt back to UTR 13 as he only won 5 games vs a 13.2 but at least that match counted to help him get back to the 12 level.

Players who are unranked who want to get started playing UTR tourneys should practice a lot before their 1st one, choose an open with high level players to get a high project UTR. If an UR player plays UTR 4s and beats them easily, probably the best he would be rated is UTR 6. However an UR player who only get 2 games vs a 10 might be projected as an 8. If a player starts out ranked too low he can only move up its UTR incrementally since he wont get credit for matches 2.0 higher and will get less weight for matches vs opponents 1.0 higher.

It is very rare for a player to beat an opponent 2.0 higher, but a large tourney may have several matches where a player 1.5-1.8 lower than his opponent wins or loses in 3 sets. Those matches should get full weight as the lower ranked opponent probably is underranked. In ITA summer circuit matches this past weekend a 5.63 beat a 7.56 (almost 2.0) in 3 sets, a 10.07 defeated an 11.82, That same 5.63 won 5 games vs an 8.77 about 3 higher but that match wont count. That 5.63 is underranked because he played 4.0 league vs UTR 4s and 5s. At the same tourney a 5.98 lost 0,1 to a 7.95 so that match counts but not the 5.63 vs the 8.77 which while not competitive was not a totally blow out either. It does not serve anyone to count blowout matches between opponents 2.0 apart, but matches that were competitive or not complete blow outs could help a player bump up his UTR. An unranked player lost in 3 to an 8.45 in the B draw so now he has a projected ranking of 8.26. He was supposed to play a 5.26 in backdraw but withdrew so he kept his high projected ranking. Now UTR by its algorithm is encouraging unranked players who lose a close match to one of the higher players in a draw to not play the backdraw if their next opponent is much lower.

UTR changes the algorithm to not count matches 2.0 UTR apart yet hosts on its platform the ITA summer circuits where some of the draws had wide ranges-one tourney had a draw with UTR 6.86 to 12.63. The top seed played 4 matches, and for 3 of the 4 he allowed 3 games or less total over 2 sets. That is not the level based play that UTR is always promoting.

I can see one argument for excluding matches 2.0+ in gap regardless of score. UTR wants to expand the high school teams reporting scores in their database. There are quite a few Texas high schools participating. Many high school teams though have to play much weaker teams and may not want those matches included in UTR to force the drop off of wins vs higher ranked players. Maybe their coaches pushed for 2.0 vs 2.5 exclusion. The solution-UTR should include 12 months of results. When my son was on his HS team, the team had the option to report scores in UTR but declined because the singles players wanted their tournaments wins counted in their 30 matches and not the wins vs lower HS players. Most of the singles players were already playing 60-80 USTA matches a year in addition to HS; they were just playing HS for fun and being part of the team. Some HS matches were vs players close in UTR but others were not. However, if UTR included all matches in a year and gave a lower weight to blowouts vs much lower ranked players, that would satisfy many HS players and coaches. If a player loses 0,0 to an opponent, how do you guess their ranking? Leave those matches out unless the players are close in range which means the player was overrated.
 
Last edited:

CiscoPC600

Professional
Nate Silver would not approve.

There is significant predictive value in blowouts. Removing them from the algorithm altogether just reduces the overall accuracy.

A better approach would be to use the ELO system 538 uses, which gives diminishing returns as margin of victory increases, but doesn’t toss out any data points.
Interesting. I'm advocating for keeping/including close matches (win or lose) with those that have a 2 or 2+ rating, but I'd be open to including blowouts if done well.
 

jcgatennismom

Professional
Does UTR take into consideration that a UTR 10 player lost to a UTR 12.1 player 6-0 6-0 but the UTR 10 player had multiple break points on in all serving games by UTR 12.1 player but UTR 10 could not convert those points. Also, UTR 10 has all his serving games at either 40-15 or 40-30 but couldn’t close out the game. Yes on paper, it is a rout but UTR 12.1 player should have lost. He could have 6-0 6-0 in favor of the UTR 10 player.

Does UTR take that into consideration?
If you want every point to count, play basketball. Tennis analysts who track results at ATP events often find that matches are won by the slimmest of margins. Sometimes players who won less overall points, win the match because they won the important points. Instead of expecting UTR to count points which I assume was in jest, instead aim for 3-5% improvement. With just a small amount of growth, more of those close games will go the losing player's way. Adjust game strategy in 2nd set if player is losing close game and start with clean slate mindset. One of my son's best matches was one he lost the 1st set 0-6. He adjusted and beat a higher ranked seed. His worst college match which he lost 0,1 he played the same player a few months later and won the first set 6-2 and was tied in the 2nd when match was called.

View losses as learning opportunities.Little tweaks can have a big impact. I think UTR should put more weight on wins but put some weight on lower ranked opponents who keep it close with high ranked players. Ultimately the better player is the one who ekes out a win even if some sets werent pretty. No coach wants a player who wins more games but chokes on matchpoints. UTR will rate higher the guy who loses 6-1 6-7 6-7 because he won 18 vs 15 games, but the guy who should be rated higher is the guy who didnt let a terrible 1st set affect his mindset but instead adjusted his game to pull out 2 close sets.
 
Last edited:

CiscoPC600

Professional
After further thought, I don't see why UTR doesn't include all data. You can have tiers of how much weight you add to matches with major disparity in ranking and the tiers can be different for the lower ranked player than the higher ranked player. This way higher ranked players aren't overly negatively impacted by low ranked players in early rounds but lower ranked players still get credit for each and every game they get.


Like in my example, I gave a guy seeded in a tournament with roughly a 2.4 UTR higher than me a good run for his money but I lost 6-4, 6-3. Lots of deuces (not that that's counted) and could've been even closer. Certainly that merits some credit.
 
Last edited:

FIRETennis

New User
That's shocking news. I'm actually a fan of the UTR system in terms of promoting equal level play amongst amateurs compared to the NTRP system. This change to their algo will basically make it very difficult for someone who plays few matches and improves their skill to actually improve their ranking. Previously, an upset against a higher ranked player, even 2+ UTR above, would count. A lot of the local leagues, tournaments and even ITF events have much later spreads than 2.0 UTR. Playing matches against those opponents and maybe playing competitively, would not give the lower ranked player any benefit.
 
After further thought, I don't see why UTR doesn't include all data. You can have tiers of how much weight you add to matches with major disparity in ranking and the tiers can be different for the lower ranked player than the higher ranked player. This way higher ranked players aren't overly negatively impacted by low ranked players in early rounds but lower ranked players still get credit for each and every game they get.


Like in my example, I gave a guy seeded in a tournament with roughly a 2.4 UTR higher than me a good run for his money but I lost 6-4, 6-3. Lots of deuces (not that that's counted) and could've been even closer. Certainly that merits some credit.
But the credit may be so low as to do nothing significant to the rating.

Really it would be great if UTR explained why they limit unbalanced match ups from counting. Perhaps they cannot reveal more about their proprietary algorithm?

The big upset stories on the forum seem convincing until we realize they are rare, which is why they are remembered. Like that time I helped Rafi (UTR 16.34) warm up at the public courts downtown and got a "point" when he double faulted in our first practice supertiebreak.
 
Question: if the higher rated player only has a projected rating, though it is 2.0 higher, would a match count with a lower rated player? I would think so or there would be no way to get an accurate official rating
 

FIRETennis

New User
Their new algo makes sense for UTR sanctioned events where the maximum draw spread is < 2 UTR anyway. Many USTA tournaments and ITF tournaments feature players ranging from 4 UTR - 12+ UTR so I do feel it is unfair to not have any of the 2+ UTR matches count especially, the upsets.
 
Their new algo makes sense for UTR sanctioned events where the maximum draw spread is < 2 UTR anyway. Many USTA tournaments and ITF tournaments feature players ranging from 4 UTR - 12+ UTR so I do feel it is unfair to not have any of the 2+ UTR matches count especially, the upsets.
Which ones are utr sanctioned? You mean run by utr itself? I never saw a tournament with a narrow utr requirement.
 

FIRETennis

New User
UTR actually just confirmed matches in which the lower rated player wins, even if rating is more than 2.00 UTR lower than the opponent), then the match counts.

@Pro Kenneth
Yes, UTR has what they call "Verified Events" which can be restricted to a certain range of UTRs, however I do not seem to find any that have such a narrow spread of UTRs.
 

texrunner

New User
Haven't read through all the comments so sorry if im repeating what others have said. Anyway, if I'm understanding this correctly, as the distance between opponents' ratings increases, less weight is placed on the score to the point where 2.0 difference or higher doesn't count at all.
 

texrunner

New User
Whoa wait a minute...

At a big junior tournament my daughter recently participated in there were some surprising scores so i just wanted to check something. Ok...there were two players, one with a 5.6 and the other with about an 8.02 (so a 2.42 difference). The lower rated player won and the match DOES count (there's that little grey ball next to the check mark.). However, next round the 5.6 loses to a 7.9 and the match does NOT count.

Another example from the same tournament, a 6.4 beat a 9.1 in a tiebreak and the match DOES count.

So i suppose matches with a 2.0 or greater spread only count if the lower rated player wins.

Examples like these are one of the many reasons i cannot stand UTR. Everyone knows that tiebreaks can be a toss up so where's the accuracy or fairness in my above examples? They either should count or not regardless of who wins.

I wonder if there are any college coaches on this forum and if so i would love their thoughts on UTR. The more i read and talk to other parents of college tennis prospects, the more UTR seems to be losing its relevancy. I hope that really is the case!
 

texrunner

New User
Pro Kenneth, no relevancy in this case is not social credit - it’s a flawed system that a lot of coaches are realizing it’s not everything it was promised to me. And I assure you tennis moms have no pull with the folks at utr
 

CiscoPC600

Professional
Pro Kenneth, no relevancy in this case is not social credit - it’s a flawed system that a lot of coaches are realizing it’s not everything it was promised to me. And I assure you tennis moms have no pull with the folks at utr
I'm not sure when they're gonna implement the algorithm. For my case it seems it did. For yours, it should've as well as the email makes it clear even upsets won't count if its a difference of 2.0 or more.

You should contact UTR. Them being transparent and getting feedback is important.
 

texrunner

New User
Who think the algorithm should be published on their site? It’s be like the SAT not giving students any idea of how their score was calculated!
 

andfor

Legend
Doubtful they’ll post their algorithm. That’s what makes UTR unique. Others will copy it, then tweak it and claim superiority. Just practice more and win more matches, problem solved.


Sent from my iPhone using Tapatalk
 
Whoa wait a minute...


So i suppose matches with a 2.0 or greater spread only count if the lower rated player wins.
Back when UTR was first getting started, they counted all matches. This negatively impacted many of the junior players who played high school tennis and had their results reported to UTR. My son was a 5-star who saw his UTR drop by more than 1 point during high school tennis season because many times he was playing an opponent with more than a 4.0 spread between them. Even when he won one of these matches, 6-0,6-0, his UTR took a hit. I think it is great that they no longer include these matches if the higher rated player wins.
 
Pro Kenneth, no relevancy in this case is not social credit - it’s a flawed system that a lot of coaches are realizing it’s not everything it was promised to me. And I assure you tennis moms have no pull with the folks at utr
Talking about SAT skills: read again what I wrote, the point is what you or tennis moms think is not what makes UTR relevant; it is what coaches think.

And "a lot of coaches are realizing" sounds like wishful thinking you are making up. If that were the case, you wouldn't care about UTR one way or the other.
 

MarTennis

Semi-Pro
Pro Kenneth, no relevancy in this case is not social credit - it’s a flawed system that a lot of coaches are realizing it’s not everything it was promised to me. And I assure you tennis moms have no pull with the folks at utr
Oh contraire mon frere. Universal Tennis is hella lurking. I'm pretty sure that are monitoring JCGATENNISMOM.
 

MarTennis

Semi-Pro
Talking about SAT skills: read again what I wrote, the point is what you or tennis moms think is not what makes UTR relevant; it is what coaches think.

And "a lot of coaches are realizing" sounds like wishful thinking you are making up. If that were the case, you wouldn't care about UTR one way or the other.
College coaches do question the pinpoint accuracy of UTR. They don't quarrel with the broad assessments. They've been burnt for sure in the recent past.
 

CiscoPC600

Professional
Looks like the community feedback at least made them include upsets..

"Do some matches not count towards my UTR?
The algorithm excludes matches in which a) a player withdraws before the match starts, b) the match starts but neither player wins at least four games (due to a withdrawal/retirement), or c) there is a UTR difference of more than 2.00 and the higher-rated player wins as expected. These excluded matches still show up on the player profile but are not used in the rating calculation.

Why does the algorithm exclude matches with a UTR difference of more than 2.00?
As the difference in UTR increases, so does the likelihood the higher-rated player wins the match easily. Our data indicates matches with a UTR difference of more than 2.00 are almost certain to be a blowout. Results like these are not indicative of either player’s skill level and are excluded by the algorithm.

Do matches with a UTR difference of more than 2.00 ever count towards my UTR?
Yes, there is one scenario in which this occurs. If the lower-rated player wins the match, this result will count towards each player’s rating. For example, if a 5.20 wins a match against a 7.61, this match will count. However, if a 7.61 wins a match against a 5.20, this match will not count."
 

MarTennis

Semi-Pro
Do matches with a UTR difference of more than 2.00 ever count towards my UTR?
Yes, there is one scenario in which this occurs. If the lower-rated player wins the match, this result will count towards each player’s rating. For example, if a 5.20 wins a match against a 7.61, this match will count. However, if a 7.61 wins a match against a 5.20, this match will not count.
 

MarTennis

Semi-Pro
Looks like the community feedback at least made them include upsets..

"Do some matches not count towards my UTR?
The algorithm excludes matches in which a) a player withdraws before the match starts, b) the match starts but neither player wins at least four games (due to a withdrawal/retirement), or c) there is a UTR difference of more than 2.00 and the higher-rated player wins as expected. These excluded matches still show up on the player profile but are not used in the rating calculation.

Why does the algorithm exclude matches with a UTR difference of more than 2.00?
As the difference in UTR increases, so does the likelihood the higher-rated player wins the match easily. Our data indicates matches with a UTR difference of more than 2.00 are almost certain to be a blowout. Results like these are not indicative of either player’s skill level and are excluded by the algorithm.

Do matches with a UTR difference of more than 2.00 ever count towards my UTR?
Yes, there is one scenario in which this occurs. If the lower-rated player wins the match, this result will count towards each player’s rating. For example, if a 5.20 wins a match against a 7.61, this match will count. However, if a 7.61 wins a match against a 5.20, this match will not count."
They were listening/lurking.
 

MarTennis

Semi-Pro
Looks like the community feedback at least made them include upsets..

"Do some matches not count towards my UTR?
The algorithm excludes matches in which a) a player withdraws before the match starts, b) the match starts but neither player wins at least four games (due to a withdrawal/retirement), or c) there is a UTR difference of more than 2.00 and the higher-rated player wins as expected. These excluded matches still show up on the player profile but are not used in the rating calculation.

Why does the algorithm exclude matches with a UTR difference of more than 2.00?
As the difference in UTR increases, so does the likelihood the higher-rated player wins the match easily. Our data indicates matches with a UTR difference of more than 2.00 are almost certain to be a blowout. Results like these are not indicative of either player’s skill level and are excluded by the algorithm.

Do matches with a UTR difference of more than 2.00 ever count towards my UTR?
Yes, there is one scenario in which this occurs. If the lower-rated player wins the match, this result will count towards each player’s rating. For example, if a 5.20 wins a match against a 7.61, this match will count. However, if a 7.61 wins a match against a 5.20, this match will not count."
Including the upsets is BIG. Will this work retroactively to the current 30 matches? That seems a relevant question.
 
Top