Friday, September 18, 2020

What will the effects of no 2020 year-end ratings be? And what might my dynamic rating be now?

The USTA announced yesterday that there will be no 2020 year-end ratings.  Due to the suspension of play earlier this year during the COVID-19 pandemic, they elected to just "punt" and make one long 2020/2021 season instead of figuring out a way to publish ratings for those players that were able to play.

In my opinion, this is unfortunate for a number of reasons, but the decision deserves a full analysis of all the pros and cons and potential effects it will have.

2020 teams can stay intact for 2021


One obvious effect is that players will stay at their current level, and thus be able to continue to play on the same teams they were on for 2020.  I believe a number of players and captains had the view that a team was formed for 2020 and the opportunity taken away to "fulfill the dream" of playing a full season together and seeing how far they could advance, so it is only fair to leave ratings alone and let those players and teams have another go at it.

This can be seen as a pro or a con depending on your perspective as this has other effects as noted below.  But this is arguably the only pro in the list of effects even in its limited form.

Self-rated players at wrong level


The effect on self-rated players will be significant as none of those that played in 2020 will get year-end ratings and thus will remain self-rated.  On the surface, that may not seem like that big a deal, but when you drill in there are several side-effects to it.

To understand the effects, we need to look at what typically happens with self-rates.  In 2019, there were over 41K self-rates that got a 2019 year-end rating.  Of these, 63% stayed the same level, 23% were bumped up, and 14% bumped down.  This means that had the USTA not published year-end ratings in 2019 and self-rated players stayed the same level, 37% or over 15,000 players would be playing at the wrong level the following year, and over 9,500 of those playing at too low a level.

Now, in 2020 with the suspension of play, self-rated players will have played less and a lower number of them gotten the three matches in required to get a year-end rating.  My stats show that just under 16K self-rated players have played enough in 2020 to have gotten a 2020 year-end rating were those to be published.  With another month and a half of play before the normal year-end, this number may go up a bit, but we'll use 16K.

If 37% of those 16K should have been bumped up or down, that is nearly 6K players that will be out of level for 2021.  Of those:
  • Over 2K will be "stuck" at too high a level and perhaps choose to not play anymore since they weren't bumped down
  • Nearly 4K will be able to play at too low a level for the ability they've demonstrated

This should all be "fixed" at the end of 2021, but for 2021 something close to the above will be the situation throughout the year.  Whether these number are alarming or not and represent a competitive (dis)advantage or not may be debatable, but I think they are an issue and will lead existing players to be unhappy about their league experience when they have to play these 4K players that are playing below what their level should be.

Note, the above numbers do not include any new self-rates for 2021 that may also be out of level one way or the other, at least later in the year as they improve.

In the end though, I don't think the situation this causes is a good thing and is arguably a con.

Existing players at the wrong level


Self-rated players are far more volatile with their ratings than established players, but there is still change from year to year for existing players so the effect of their not moving needs to be looked at.

In 2019, about 85% of established players stayed the same level, 5% were bumped down, and 10% bumped up.  With fewer matches played, fewer players would be in a position to be bumped up or down, but using my ratings, it appears about 5% of players would be in line to be bumped up and 2.5% bumped down.

This means that for 2021, these 7.5% of players will be out of level, and similar to the self-rates, those that should be lower may decide not to play where they aren't competitive, and those that should be higher get to unfairly prey on lower rated opponents.

If you don't believe this can be significant, here is a chart from my Estimated Dynamic Rating Report showing how a player has improved this year.


This is a player who started the year as a strong 4.0, but is now far closer to being a 5.0 than even a 4.5, but per the USTA's decision, they will remain a 4.0 for 2021.  Does that make sense?

Note that I picked a particularly egregious case to show, but there are a lot more that are still too high and should be bumped up.

I think leaving established players at their old level is a con, albeit perhaps not a huge one given the relatively small percentage, although the actual number is around 11K rated too low and 5.7K too high.  Add with the self-rated numbers, and that means about 15K players playing at a level below where they should be and 8K stuck at too high a level.  That is not insignificant.

Self-rated players managing their rating


As I understand it, a player's matches from 2020 will count towards their dynamic rating and 2021 year-end rating, and as noted above, nearly 4K of them will enter 2021 with a rating that is too high for their self-rate level.  It is possible they are even carrying strikes from 2020 (although I'm not certain these will carry over or not).

Now, this is probably a good thing in that if they really are out of level they are well on their way to getting a 3-strike DQ and being promoted up a level which "fixes" the problem noted above.  However, we all know there are captains that follow their self-rated player's performance closely and they may encourage their player to manage their rating to avoid getting disqualified.

Managing ratings is never a good thing as it keeps players at the wrong level and can make a mockery of matches that are played and reduce the satisfaction a player has from league play.  It can also wreak havoc with ratings giving a player too much credit for a result when they beat a higher rated player that is tanking the match.

So I think this effect is definitely a con.

Players trying to achieve a goal unfulfilled


The vast majority of USTA League players play for fun, to have competitive matches against similar skilled players and aren't fixated on Nationals or forming a super team, etc.  Many of these may be looking to improve and achieve a goal and be looking to be bumped up.  That goal will go unfulfilled for 2020 as their only official validation is the year-end rating that will now not be published.

The good news is that those that want to get an idea of where they stand, and at a minimum see what direction their rating is going, can do so by getting one of my Estimated Dynamic Rating Reports.  These reports given an estimate of where your dynamic rating is and gives a lot of insight into your rating and how it has changed and statistics to help you identify what is and isn't working.  I've continued to do these all year and these can be a great way to fill in the gap from no year-end ratings from the USTA.  Contact me if you are interested in getting a report.


By my analysis, the only good thing from the decision to not publish 2020 year-end ratings is that teams get to stick together for 2021.  I wrote this months ago and pointed out some of the above issues when I did my analysis and recommendation on what the USTA should do.  I also included a poll where just 22% voted for not publishing at all like the USTA did.

Is allowing 2020 teams to stick together for 2021 really a good idea given all the other issues?  Did the USTA make a mistake and just take the easy path in making this decision?


Update: Based on feedback to this blog and what I've heard, I wrote up more comments and reaction here.

2 comments:

  1. It's nice if your team can stick together, but should this really happen if there's players wrongly rated on it?

    There's so many factors that the computer can't take into account when calculating ratings such as: who you play, when you play them, surface, indoors/outdoors, weather, how many rated matches you play(seems like the fewer you play the more at risk you are of bumping), singles/doubles, age league(I've seen several 55+ players bump because they can dominate in the local 55 leagues but honestly have no business bumping). Lots of variables that can alter ratings quite a bit, and that's if everyone is playing honestly every match. And you could potentially win every match(losing more games than your opp) and bump down or vice versa; that just doesn't seem right. Overall, the computer probably gets it right around 85-90% of the time, but that's still a big % not getting it right.

    USTA leagues can be quite a shady business. I'm surprised Nationals wants to add to it and make it worse than a normal year next year by not publishing. Why not fix the current problems with ratings at the very least? A big problem I see is that districts/sections' offices do nothing about egregious S players or even C players during the year, but maybe they do in some areas. While we do have the strike system, this can be avoided by 'managing', etc. Also, you can't get strikes in mixed or tri level leagues it seems.

    ReplyDelete