Monday, December 23, 2013

A Christmas gift to Northwest Washington - Full year-end Estimated Dynamic NTRP ratings list

I have just updated the Estimated Dynamic NTRP Rating list for Northwest Washington in the Pacific Northwest section of the USTA here.

As always, if you want more detail on how your rating was arrived at, an Estimated Dynamic NTRP Rating Report is the best way.  Contact me for more details.

Saturday, December 21, 2013

The perfect stocking stuffer - Estimated Dynamic NTRP Rating Reports are on sale

It is four days before Christmas and if you are still looking for a stocking stuffer for the tennis nut in your family or group of friends, I have a great one for you.

Tennis players that play in USTA leagues tend to obsess about their NTRP rating.  Year-end ratings came out earlier this month and many are wondering why they were/weren't bumped up/down.  A great way to find out is to get an Estimated Dynamic NTRP Rating report.  A report will show each match they played in a relevant USTA league, how it was individually rated, and how it affected their dynamic rating.  It will also show what their estimated year-end rating was and offer an explanation for why it ended up where it did.  See another example in this earlier blog post.

Thru Christmas, I'm offering 25% off all reports.  Contact me for more information or to order a report.

Tuesday, December 17, 2013

The 2014 USTA League season is upon us, Estimated Dynamic NTRP Reports are already being generated

It has been just two weeks since 2013 year-end NTRP ratings were published by the USTA, and people are already interested in what their 2014 dynamic rating is looking like.  This is because 2013 Fall leagues and/or 2014 early start leagues have had matches since the 2013 league year ended (10/27) and these matches will count towards the 2014 ratings.  In some cases players have played 5, 10, or more matches in "2014" already.

Given this, I've started generating Estimated Dynamic NTRP Rating reports for 2014 already and can do so for any USTA League player from any section.  Some are even signing up for periodic updates of their report and getting a discount over buying them individually which is a great way to track progress towards any goals that may have been set.

See the link above or contact me if you are interested or have any questions.

Sunday, December 15, 2013

USTA League Points Per Position Survey - Probably not as good an idea as it appears on the surface

The USTA sent out a survey last week asking for people's thoughts on a "Points Per Position" scoring proposal for USTA League team matches.

The short summary on the proposal is that rather than each court counting equally and there being a winning team based on who won the most courts in a match, the lower numbered courts will have more points allocated to them and team standings would be based on total points accumulated rather than team wins/losses.  The idea is that by doing this, teams will be encouraged to play their best players on the courts that are worth more points resulting in more consistently competitive matches.

Apparently, the USTA has received complaints from players about uncompetitive matches and/or "stacking".  What happens with stacking is that a team tries to steal a team win by playing weaker players on court 1 in order to try to assure that they get wins on courts 2 and 3.  In a normal 5 court match, by throwing courts 1 and winning courts 2 and 3, the team can get a 3-2 team win.

This is great unless you are the good player on the team playing things straight up.  You may be a strong 4.0 and have to face a so-so 3.5 playing up on court 1.  This isn't a competitive match and not a lot of fun for either player.  It can also wreak havoc with ratings as having players that are rated far apart play each other is hard to properly rate.

The issue originates in the misconception that there is meaning to the court number, e.g. court 1 is supposed to have the best player(s) and higher numbered courts are supposed to have the weaker players.  The USTA actually states that court numbers don't carry that meaning, they could just as easily be name red, white, and blue.  If there is no meaning to the court number, players shouldn't expect to play court 1 and get a tough opponent.  You might or might not.

But let's humor the USTA and consider what happens with a Points Per Position scoring system.

Given that some courts will be worth more, this results in two things.

First, there is even more pressure on captains to get the ringers and stud players that they can play on court 1 and be reasonably assured of a win.  There are already complaints about teams that recruit players to self-rate too low or manipulate ratings to get bumped down so they can be a ringer, and the pressure to do so would be even greater with this scoring system.

Second, the higher numbered courts will be worth less, effectively telling the weaker players on the team that they are less important.  One can do the math and see how a team might just have studs that always win courts 1 and could accumulate enough points that it doesn't matter if they ever win courts 2 or 3.  This effectively means teams don't need depth.

And if the original issue was uncompetitive matches, the problem may still not be solved.  Consider a team that has several exceptional players for their level.  Their opposition will probably know this and if they have several very good players that are still going to be underdogs, the prudent move may still be to play these players on the higher numbered courts rather than wasting your best players in losses on court 1.

If this is the case, we just changed the system to one that rewards top-heavy rosters and sandbagging and didn't in fact help reduce the stacking that was the original complaint.

So while I understand what they are trying to do, I don't think this is the right approach.

There are other approaches that might be better solutions.

First, while the USTA says the court number doesn't matter, it actually does a bit.  If there are defaults, they start at the highest numbered court, so there is some meaning to the courts today and you wouldn't want to put your best doubles team on court 3 only to have the other team default that court and your best players don't get to play.

An extension of this that gives a little more meaning to the courts would be to have a rule that players playing up are not allowed to play on a lowered number court than players at level.  This would avoid the scenario where a team throws a court by playing their player playing up on court 1 and avoid the biggest case of uncompetitive matches.

Another that the USTA will never do is to publish actual dynamic ratings and requiring that a team have their highest rated players on the lower numbered courts.  Or if they don't want to publish the ratings, have an app that lets you put in your line-up and it tells you which court the players must be on to have the right ordering.

Now, both of these would essentially ensure that the stronger team/line-up usually wins.  We'd be taking away the ability of a captain to manage their team into the best position to get the 3-2 win.  If the USTA wants to take the drama out of some of the matches and reduce the upsets, either of these could work.

What do you think?  Leave a comment and let me know.

Monday, December 9, 2013

Tips to help improve your USTA NTRP rating

The USTA NTRP year-end ratings are out and 2014 leagues will be starting soon if they haven't already.  Some players have been bumped up achieving what was perhaps a goal for 2013, but others with that same goal didn't quite make it.  Whatever the case may be, if your goal for 2014 is to move up a level, here are some tips on how to accomplish that.

First, the way the NTRP algorithm works, every game counts.  Winning a match 6-2,6-2 is going to do more to improve your rating than winning 6-4,6-4.  It is the couple of games here and there that get away, especially when playing lower rated opponents, that is the difference between the players that get moved up and those that don't.

But it doesn't just matter when you win.  We like to remember the good matches and ask how we didn't get moved up when we had such good results, often forgetting we had some real stinkers of a match too.  These count just as much as the good matches and while nearly everyone is going to lose some matches, limiting the damage and keeping them as close as possible is another important factor in improving your rating.

Second, playing up is a good way to have the opportunity to improve your rating, but playing up will not automatically result in a bump up.  If all you do is lose, and lose badly when playing up, your rating can actually go down.  This is because you may be playing opponents that are also playing up, or opponents may simply be rated on the lower end of their level, and poor results against them will have negative effects on your rating.  It is possible that you can do more to help your rating by winning convincingly at level rather than playing up.  Of course, if you play up and do well and get some wins, that will likely result in your rating going up.

Third, and this one may be counter-intuitive, but playing with a lower rated partner can present more opportunity to improve your rating.  The way the algorithm works is that it compares the expected result with the actual result for a match.  If you play with a strong partner, you are supposed to win, perhaps easily, and a close win could actually hurt your rating.  But if you play with a weak partner and pull off an upset win, that can really boost your rating.

Fourth, and perhaps most important, improve your game.  Your rating is simply an indication of your ability based on your match results.  The three items above are all about getting the most out of your opportunities, but in the end if your tennis skills don't change, your results won't change significantly and your rating isn't going to change very much.  So go get a few lessons.  Identify your strengths and weaknesses and improve your weaknesses and come up with a game plan that accentuates your strengths.

As you go through your new season with a rating improvement plan in place, you may start to see results in the form of more wins or more competitive matches when playing up.  But how will you know if you rating is really improving?  In some sections, there are Early Start ratings that give you an idea, but those often don't come out until July, August, or later.  Another option for sections without early start ratings or those that don't want to wait is to get a mid-year Estimated Dynamic NTRP Rating Report that will give an accurate picture of your current dynamic rating and how each match has contributed to it.

Contact me for more information or even to get a report for your 2013 season to see why your rating ended up where it did.

Friday, December 6, 2013

Interesting USTA League Stats - What levels do players self-rate at and is there deliberate underrating?

When new players register to join a USTA League team, they are required to self-rate.  There are a set of guidelines to be used to have a player rate themselves at the right level so that they are hopefully playing competitive matches when they begin play.

With the data I have from generating Estimated Dynamic NTRP Rating Reports, I thought it would be interesting to look at what levels players self-rate at.  Here is what I found for 2013 self-rates.

This probably isn't a big surprise, with the most common levels being 3.0 and 3.5.  Players that self-rate are usually new or newer to the game, but many of these having some athletic background, and so you'd expect the lower to middle levels to be the most common.

It is interesting that even though the lowest level league is typically 2.5 and the guidelines above only going down to 2.5, there are a few players that self-rate at 2.0.

Next, I thought it would be interesting to look at some different areas to see if the self-rate profile was similar or different.

First, California.

The profile is very similar with a little more bias towards 3.5 than 2.5 and 3.0.

Next, Florida.

Again, a similar looking profile.

Let's go to a state that might be more unique by being an island with perhaps a different cross-section of players, Hawaii.

The profile remains very similar.

Last, let's take a look at Puerto Rico.

Here we see a different profile, and not just subtly different either.  The most common level is 2.5, a full two levels "left" of the rest.

It would be interesting to understand the reason for this.  It is certainly possible that in Puerto Rico, players' first exposure to tennis is often through USTA League and so you get more that are truly beginners and this chart simply is evidence of that.  However, there are some that believe that there is systemic and deliberate underrating that takes place in order to create teams likely to do well at Nationals.  There are disqualification provisions in place to try to prevent this and I've written about these before, but they aren't perfect.

This chart certainly isn't proof that deliberate sandbagging is taking place in Puerto Rico, but it certainly supports that hypothesis.

Wednesday, December 4, 2013

Throwing matches in USTA League - Does it happen?

In sports where there are levels of play, a longstanding issue is how to have players playing at the right level.  This is a perennial challenge with USTA League play too.

By definition, there will be higher and lower rated players within a given level and while matches will generally be competitive between players at the same level, the higher rated players are going to win more than the lower rated players.  For some on the lower end of the range for a level, it is tempting to think "If I were only at the next lower level, I could win more".

The dark side of USTA League is that some players will go out of their way to try to make sure they get bumped down a level so they can win.  They may do this by deliberately losing games or matches in order to cause their rating to go down.  This is certainly not in the spirit of the rules and unfortunate, but it happens.

In many cases, a player that is right on the margin between levels can get bumped down and their level of play is not so high that there is a major issue with them playing at the lower level.  However, it does become a major issue when a player tanks matches so much that someone that can compete and win at the higher level manages to get bumped down to a level where they are clearly at the wrong level.

We'd ideally have a system that can detect when matches are being thrown and compensate accordingly.  The challenge is that every player's play varies from match to match so simply having good/bad matches doesn't necessarily indicate suspicious behavior.  In generating Estimated Dynamic NTRP Rating Reports, I regularly see players that have a range of 0.5 and a bit more than that is not uncommon, especially when a player is improving during a season.

I recently came across a player that seemingly goes way beyond natural variance from match to match and would seem to be a clear indication of suspicious activity.  Here is their chart from their 2013 matches.

We can see that this player has matches rated as high as 4.96 and as low as 3.64.  That is a range of 1.32, the largest I've ever seen.  A 4.96 is a rating it would be reasonable for a 5.5 rated player to post and a 3.64 could easily be posted by a 3.5.  We even see a 3.69 result exactly one week before the 4.96.  This is an awfully big range and not what one would normally see from normal variance of play.

We can also see a group of lower rated matches all in September.  These matches are from a Fall league that counts towards ratings but does not advance to any National championships and is rumored to be used by some players to manage their ratings down.

Was this player throwing matches to manage their rating?  I can't say for sure, I'm just looking at numbers and charts.  There could have been factors involved that explain why the results are what they are.  I can only say that the wide range is suspicious and makes one go hmmmm.

Tuesday, December 3, 2013

Appealing a USTA NTRP Rating - Successes and what it says about estimated ratings accuracy

I wrote about the new rules for appealing year-end ratings and I've received feedback from quite a few of you about your success in having appeals granted.  Some have wanted to be bumped up while others bumped down and both groups have been able to accomplish their goal.

What is especially interesting is finding out when someone successfully appeals, as that provides an opportunity to validate the estimated ratings I have for a player.  Since there are ranges for appeals to be granted, finding out when one is lets me know if my rating was close or not.

Here are a few I heard about the past couple of days:

  • A self-rated 3.5 was bumped up to 4.0 but was able to appeal down to 3.5.  They'd played 6 matches so their range was 0.06 and my rating was 3.56 which was within that range.
  • A self-rated 3.5 wasn't bumped up to 4.0 but was able to appeal up.  The range for this is 0.04 and I had their rating at 3.45 so just 0.01 off from being right.
  • A 3.5 wasn't bumped up to 4.0 but was able to appeal up.  I had their rating at 3.52 so just 0.02 off from being right.
  • A 3.0 wasn't bumped up to 3.5 but was able to appeal up.  I had their rating at 2.99 which is within the range.
I generally aim to have my estimates within 0.05 of the actual rating, so all of these are within that.  And potentially each of these ratings are no more than 0.02 off from the actual rating.

Seeing that my ratings are this accurate great.  See this earlier post on other observations on accuracy.

And if you think you are close and tried to appeal, let me know if you did or didn't have it granted.

Monday, December 2, 2013

Year end ratings are out, how accurate are my Estimated Dynamic NTRP Ratings?

With year end ratings now being published, I took a quick look at some players to see how accurate my Estimated Dynamic NTRP Ratings are.  I plan to do a more in depth analysis later, but here is how I did on some initial checks.

I played on a 4.0 team that went to local playoffs this year and included a 3.5 playing up and several players that played up at 4.5 too.  In our early start ratings, two were bumped up to 4.5 while the rest stayed where they were and I had all of these correct, and in the year-end ratings both early start bump ups came back down as I predicted.  So 14 for 14 here.

I did reports for a 4.5+ team that went to Nationals and won it all, and as expected they had quite a few players bumped up.  I got every player on this roster correct except for one.  This included a few self-rated players as well as a 5.0 that won every match all year and did not get bumped up to 5.5.  11 for 11 here.

I just heard back from someone I did a report for that I nailed her rating too.  She was a 3.0 that I had at 2.99, just below the threshold to be bumped but close enough to appeal if I was correct.  She wasn't bumped in the year-end ratings but was able to successfully appeal.  So my 2.99 was correct or at most just a few hundredths off.

From this sample, I am 25 out of 26 correct, or 96%.  I don't know that when I do a more complete analysis I'll maintain this high percentage, but I've also gotten other feedback and checked a few others and found my ratings to be right now for most.

Stay tuned for more.

2013 USTA NTRP ratings are out - Why didn't I get bumped up (or down)?

With the publishing of the 2013 year-end NTRP ratings by the USTA, the inevitable questions begin, the main ones being "Why wasn't I bumped up (or down)?"

The algorithm used to calculate the ratings works off of the scores of your matches and the ratings of the players involved.  Your win/loss record actually isn't taken into account at all.  This can lead to players going undefeated but not getting bumped up, or having poor records and getting bumped up.

If you are baffled by why you weren't bumped up/down, an Estimated Dynamic NTRP Rating Report is a great way to find out and understand why your rating is where it is.  And while my ratings aren't perfect, of the 30 or so players I've checked so far today, 29 have been correct.

No matter what section you are from, I can generate a report.  So contact me if you are interested.

2013 Year-End USTA NTRP Ratings Released

It appears that the year-end ratings for the 2013 season have been released today as expected and can be accessed on the TennisLink Web-site.  All those I've checked have a date of 12/31/2013 and if any section hasn't been updated yet I'm sure it will be shortly.

Sunday, December 1, 2013

People really do shop on Black Friday rather than thinking about USTA NTRP ratings

Thanksgiving weekend is known for big sales as everyone looks for bargains as they begin their Christmas shopping in earnest.  For USTA League tennis players, it is also known as the weekend that precedes the Monday that year-end NTRP ratings come out.

With both of those in mind, it is interesting to take a look at the traffic to this blog.

As you might expect, my post on when year-end ratings will be out has garnered quite a bit of traffic, in fact resulting in near to new recent highs for page views on Wednesday, Thursday, Saturday, and today (Sunday).

What about Friday?  Page views on Friday were less than half of the other four days, so yes, even tennis players go shopping on Black Friday.