Sunday, January 26, 2014

What would the men's NTRP ratings be from the Australian Open?

I wrote yesterday about what the NTRP ratings would be for the Women's Australian Open if the USTA's NTRP algorithm were to be applied to the tournament results, and I've also done it for the men now.  See the table below.

Like I did for the women's, I only used results from this tournament so I had to default players to start, but did go through an extra iteration to dampen out the effects of doing so.

It is no surprise that Stan heads the list, but it is perhaps a little surprising to see that Djokovic and Berdych are ahead of Nadal.  This is in part because Novak took Stan to a 5th set and Berdych took him to 4 sets, but also because the NTRP algorithm gives a lot of weight to the most recent match, and Nadal was expected to win so the loss was a poor result and really dropped his rating.

Have a look, and let me know what you think.

NameRatingRecord
NameRatingRecord
S.Wawrinka7.245-0
N.Djokovic7.144-1
T.Berdych7.105-1
R.Nadal7.095-1
A.Murray7.064-1
R.Federer7.065-1
D.Ferrer7.054-1
G.Dimitrov7.034-1
K.Nishikori7.023-1
R.BautistaAgut6.953-1
F.Verdasco6.931-1
R.Gasquet6.912-1
M.Raonic6.902-1
T.Robredo6.903-1
S.Robert6.903-1
M.Youzhny6.901-1
J.Tsonga6.893-1
V.Pospisil6.892-0
F.Mayer6.873-1
D.Istomin6.872-1
J.DelPotro6.861-1
E.Roger-Vasselin6.862-1
F.Lopez6.862-1
A.Dolgopolov6.841-1
A.Falla6.841-1
Y.Lu6.831-1
I.Dodig6.831-0
L.Mayer6.821-1
D.Tursunov6.821-1
D.Dzumhur6.821-1
P.Andujar6.821-1
M.Berrer6.811-1
S.Querrey6.812-1
J.Chardy6.812-1
M.Klizan6.801-1
F.Fognini6.792-1
K.Anderson6.793-1
V.Hanescu6.781-1
G.Simon6.782-1
B.Rola6.781-1
V.Millot6.771-1
B.Paire6.772-1
G.Monfils6.772-1
T.Gabashvili6.762-1
A.Mannarino6.761-1
N.Davydenko6.761-1
N.Mahut6.750-1
N.Kyrgios6.741-1
J.Benneteau6.741-1
J.Sock6.741-1
M.Przysiezny6.731-1
K.DeSchepper6.731-1
A.Seppi6.731-1
J.Vesely6.730-1
M.Cilic6.731-1
M.Ebden6.731-1
J.Janowicz6.732-1
T.Bellucci6.720-1
L.Kubot6.710-1
L.Hewitt6.710-1
J.Nieminen6.701-1
D.Thiem6.701-1
I.Sijsling6.700-1
L.Rosol6.700-1
D.Lajovic6.691-1
D.Brands6.690-1
B.Klahn6.680-1
G.Garcia-Lopez6.680-1
J.Monaco6.680-1
I.Karlovic6.670-1
E.Gulbis6.671-1
D.Young6.671-1
A.Gonzalez6.670-1
D.Sela6.670-1
S.Devvarman6.660-1
M.Matosevic6.660-1
J.Thompson6.660-1
A.Nedovyesov6.650-1
S.Giraldo6.650-1
R.Williams6.650-1
T.Kokkinakis6.651-1
C.Berlocq6.650-1
P.Gojowczyk6.640-1
J.Duckworth6.640-1
B.Kavcic6.640-1
L.Lacko6.640-1
M.Baghdatis6.640-1
J.Sousa6.640-1
L.Pouille6.630-1
S.Groth6.630-1
T.Kamke6.630-1
Z.Zhang6.630-1
D.Wu6.620-1
F.Dancevic6.620-1
R.Harrison6.620-1
D.Gimeno-Traver6.610-1
F.Volandri6.610-1
S.Johnson6.610-1
H.Zeballos6.610-1
M.Granollers6.600-1
D.Kudla6.600-1
R.Berankis6.590-1
W.Odesnik6.590-1
D.Guez6.590-1
A.Ramos6.580-1
B.Becker6.580-1
J.HutaGalung6.580-1
P.CarrenoBusta6.580-1
F.Delbonis6.570-1
A.Bedene6.570-1
M.Llodra6.560-1
J.Struff6.540-1
G.Soeda6.540-1
M.Kukushkin6.510-1
M.Russell6.510-1
S.Stakhovsky6.510-1
J.Hajek6.510-1
T.Smyczek6.490-1
J.Wang6.490-1
A.Montanes6.460-1

Saturday, January 25, 2014

What would the women's NTRP ratings be from the Australian Open?

Those of us that play USTA League are well aware of the NTRP rating system.  The system has levels of play all the way from 1.5 to 7.0 (leagues typically start at 2.5) where 7.0 is defined as a world class player.  The fact that they even list what a world class player would be has always piqued my interest and so I've started to collect ATP and WTA results that I can calculate NTRP ratings from.

Yes, I know these are meaningless, the pros don't care what their NTRP rating is, but I thought it would be fun to look at to see the range of ratings and if the ratings pass the smell test.

Now, this is just the first of what I hope will be many posts on the subject, and this one is using limited data, so let me explain what I've done.

I have only used results from the Australian Open women's main draw and only those results that had completed matches.  Because I have no history, I have defaulted all players to be an "average" 7.0 (6.75) but have gone through an extra iteration of calculations to have the effect of everyone starting at the same rating be diminished.

Without further adieu, the table below shows how the ladies would rate.

There is no surprise that Na Li tops the list and by a wide margin.  Cibulkova, Radwanska, Bouchard, and Ivanovic being next are also no surprise.  It is also interesting to note that the top 4 would technically have an NTRP rating of "7.5".  What is interesting is the next three.

Safarova is the 6th highest rated player despite going out in the third round.  Why you may ask?  Because she took Na Li to a third set, arguably playing her tougher than any of her other opponents.  Because the NTRP algorithm takes into account who you play, she gets more credit for this than other players do for advancing farther against weaker opponents.

Similarly, Serena and Jankovic with 3-1 records are ahead of three others with 4-1 records.

I will be continuing to work on this, and will have the men's posted tomorrow after the final.  Let me know what you think.

NameRatingRecord
NameRatingRecord
N.Li7.267-0
D.Cibulkova7.096-1
A.Radwanska7.025-1
E.Bouchard7.025-1
A.Ivanovic7.004-1
L.Safarova6.972-1
J.Jankovic6.963-1
S.Williams6.953-1
F.Pennetta6.944-1
S.Halep6.944-1
V.Azarenka6.944-1
M.Sharapova6.913-1
S.Stosur6.892-1
C.Dellacqua6.883-1
C.Wozniacki6.872-1
A.Kerber6.873-1
M.Keys6.871-1
J.Goerges6.861-1
A.Riske6.862-1
Ka.Pliskova6.861-1
K.Nara6.852-1
A.Pavlyuchenkova6.852-1
M.Niculescu6.842-1
G.Muguruza6.843-1
S.Lisicki6.841-1
Z.Diyas6.832-1
S.Stephens6.833-1
G.Voskoboeva6.811-1
K.Knapp6.811-1
M.Puig6.811-1
E.Svitolina6.812-1
E.Makarova6.803-1
D.Hantuchova6.792-1
O.Rogowska6.791-1
A.Beck6.791-1
B.Jovanovski6.781-1
C.McHale6.781-1
B.ZahlavovaStrycova6.781-1
J.Zheng6.782-1
O.Govortsova6.781-1
M.Rybarikova6.781-1
B.Bencic6.771-1
A.Cornet6.771-1
K.Flipkens6.771-1
M.Barthel6.772-1
V.Razzano6.761-1
L.Davis6.762-1
I.Falconi6.761-1
C.Giorgi6.751-1
J.Gajdosova6.750-1
L.Hradecka6.751-1
V.Williams6.750-1
C.SuarezNavarro6.752-1
T.Majeric6.750-1
L.Kumkhum6.741-1
A.Kudryavtseva6.731-1
A.Tomljanovic6.731-1
M.Minella6.731-1
I.Begu6.730-1
K.Bertens6.730-1
P.Kvitova6.720-1
Y.Meusburger6.722-1
A.Morita6.721-1
F.Schiavone6.720-1
H.Watson6.710-1
T.Pironkova6.711-1
L.Tsurenko6.710-1
A.Schmiedlova6.711-1
S.Peng6.710-1
J.Glushko6.710-1
A.VanUytvanck6.700-1
T.Pereira6.700-1
M.Erakovic6.701-1
V.Lepchenko6.701-1
V.Dolonc6.701-1
S.Voegele6.701-1
Y.Wickmayer6.701-1
T.Babos6.690-1
L.Arruabarrena6.690-1
J.Larsson6.680-1
K.Date-Krumm6.680-1
K.Kanepi6.680-1
Y.Shvedova6.680-1
C.Scheepers6.670-1
B.Mattek-Sands6.670-1
P.Mayr-Achleitner6.670-1
Y.Chan6.670-1
K.Zakopalova6.670-1
S.Cirstea6.670-1
S.Zhang6.660-1
A.Konjuh6.660-1
H.Tang6.660-1
R.Vinci6.650-1
D.Pfizenmaier6.640-1
Y.Putintseva6.640-1
S.Sanders6.630-1
K.Siniakova6.630-1
A.Tatishvili6.630-1
S.Kuznetsova6.630-1
K.Mladenovic6.630-1
P.Ormaechea6.630-1
M.Duque-Marino6.620-1
S.Errani6.620-1
C.Garcia6.620-1
N.Kichenok6.620-1
S.Peer6.620-1
S.Vickery6.620-1
M.Doi6.610-1
V.Zvonareva6.600-1
A.Barty6.600-1
E.Vesnina6.600-1
J.Cepelova6.600-1
A.Petkovic6.590-1
Y.Duan6.590-1
V.King6.590-1
C.Witthoeft6.580-1
A.Cadantu6.570-1
S.Soler-Espinosa6.570-1
M.Lucic-Baroni6.570-1
S.Hsieh6.570-1
A.MedinaGarrigues6.560-1
D.Vekic6.560-1
L.DominguezLino6.560-1
K.Piter6.550-1
L.Robson6.540-1
P.Parmentier6.530-1
P.Martic6.470-1

USTA League NTRP Benchmark calculations and why playoff results carry more weight

2013 USTA League year-end NTRP ratings came out nearly 2 months ago now and 2014 leagues have started in many areas, but I'm still getting questions about 2013 year-end ratings and specifically how playoff results get counted.  So here is a quick summary.

Dynamic NTRP ratings are calculated throughout the year as each match is played.  At year-end though, the final dynamic rating doesn't necessarily become the year-end rating, rather some additional calculations are done.  For one, in sections where tournaments are included in year-end ratings, they get included, but for purposes of this post, there are additional calculations done with playoff results, specifically what are called benchmark calculations.

To understand why there are benchmark calculations, it is important to remember the purpose of the NTRP system.  The main purpose is to rate players at appropriate levels to promote competitive play.  In leagues that advance to Sectionals and Nationals, players from different areas that likely have no history of playing each other come together to play which can make ensuring competitive play occurs difficult.  How can we expect that a 4.0 from Miami is of generally the same playing ability as a 4.0 from Seattle?  The answer is benchmark calculations.

The USTA does not publish details on how benchmark calculations are specifically done, but what follows is a high level understanding I've been able to glean from various sources.

When Nationals are complete, the results for each player that played there are given extra weight to calculate their ratings.  This is done to try to get the ratings of the Nationals players from each section to be accurate relative to each other.  Then using the ratings of the Nationals players, this is done at the section level to have how weak or strong the section's representative was at Nationals filter back in to the section.  This then gets repeated at the early stages of players, e.g. Districts, Regional, and Local, to continue to have the ratings be adjusted so that a 4.0 (and every other NTRP level) in New York City is relatively the same as in Iowa.

Make sense?  As you can imagine, doing all these calculations are not trivial, which is why it is roughly a month between when Nationals finish and the year-end ratings are published.

Hopefully this helps answer the question.  Leave a comment if you have more.

Sunday, January 19, 2014

Early Super Bowl XLVIII Preview

With the Conference Championships complete, the ratings don't change much, Seattle and Denver stay #1 and #2 and we get the rare 1 vs 2 match-up in the Super Bowl.

RankTeamRatingRecordScheduleChange
RankTeamRatingRecordScheduleChange
1Seattle90.45715-382.008+0, +0.141
2Denver88.90215-379.926+0, +0.324
3San Francisco88.35114-582.745+0, -0.125
4New Orleans86.47912-682.573+0, -0.011
5Arizona86.03710-682.574+0, +0.005
6Carolina85.68112-581.869+0, -0.021
7New England85.39213-580.480+0, -0.292
8Kansas City83.77211-679.982+0, +0.049
9San Diego83.63710-881.316+1, +0.061
10Cincinnati83.60311-679.077-1, -0.012
11Indianapolis83.20812-681.501+0, +0.011
12St Louis81.8337-982.989+0, +0.004
13Philadelphia81.75110-779.839+0, +0.032
14Dallas80.2388-880.065+0, +0.031
15Pittsburgh79.9518-878.681+0, -0.017
16Tennessee79.7727-981.183+0, +0.022
17Miami79.4928-880.740+0, -0.040
18NY Giants78.7157-980.745+1, +0.034
19Baltimore78.6878-879.374-1, -0.017
20Detroit78.5587-978.851+0, +0.006
21Chicago78.3148-879.145+0, +0.008
22Green Bay78.1648-8-179.400+0, +0.002
23Tampa Bay78.0454-1282.920+0, -0.025
24Atlanta77.7834-1282.472+0, -0.024
25Buffalo77.6256-1080.418+0, -0.031
26NY Jets77.3418-880.254+0, -0.031
27Minnesota76.5655-10-179.853+0, +0.008
28Oakland74.9934-1281.041+0, +0.046
29Washington74.4473-1380.653+1, +0.031
30Houston74.4312-1482.275-1, +0.014
31Jacksonville74.1894-1281.530+0, +0.015
32Cleveland74.0494-1279.618+0, -0.011

Who is the pick in the Super Bowl? Based on the ratings alone, Seattle would be a 1.5 point pick. But stay tuned for a more in depth preview later.

2014 AFC and NFC Championship Game Previews

I haven't been posting my NFL ratings and thought regularly this year, but it is Conference Championship time so I thought it was time to do so.

First, here are my computer's ratings through the Divisional Round of the playoffs.

RankTeamRatingRecordScheduleChange
RankTeamRatingRecordScheduleChange
1Seattle90.31614-381.824+0, +0.185
2Denver88.57814-379.766+0, -0.015
3San Francisco88.47614-482.133+0, +0.837
4New Orleans86.49012-682.589+0, -0.140
5Arizona86.03210-682.570+1, +0.013
6Carolina85.70212-581.903-1, -0.784
7New England85.68413-479.808+0, +0.872
8Kansas City83.72311-679.922+1, -0.120
9Cincinnati83.61511-679.096+2, +0.022
10San Diego83.57610-881.244+0, -0.069
11Indianapolis83.19712-681.484-3, -0.781
12St Louis81.8297-982.985+0, +0.010
13Philadelphia81.71910-779.800+0, -0.011
14Dallas80.2078-880.022+0, -0.006
15Pittsburgh79.9688-878.705+0, +0.082
16Tennessee79.7507-981.150+0, -0.058
17Miami79.5328-880.788+0, +0.033
18Baltimore78.7048-879.381+1, +0.101
19NY Giants78.6817-980.694-1, -0.021
20Detroit78.5527-978.845+0, +0.028
21Chicago78.3068-879.139+0, +0.020
22Green Bay78.1628-8-179.409+0, +0.084
23Tampa Bay78.0704-1282.948+0, -0.000
24Atlanta77.8074-1282.500+0, +0.005
25Buffalo77.6566-1080.468+0, +0.045
26NY Jets77.3728-880.303+0, +0.045
27Minnesota76.5575-10-179.839+0, -0.009
28Oakland74.9474-1280.979+0, -0.043
29Houston74.4172-1482.257+0, -0.023
30Washington74.4163-1380.619+0, +0.033
31Jacksonville74.1744-1281.495+0, -0.058
32Cleveland74.0604-1279.643+0, +0.051

We can see the 4 teams in the championship games are ranked 1 thru 3 and 7.  So not a lot of surprises here.  We can also see that the home teams are rated higher, so in theory when you add in home field advantage they should be favored as they are.

But let's take a deeper look!

Here is Seattle's performance chart.



We can see they had their worst game of the year against Arizona, but recovered pretty well with good results against St. Louis and New Orleans.  We can also see that while they do well at home, they have some very good results on the road too so they aren't just a home team.  Even if we ignore their loss to Arizona, their last three games are right around 94, but with it their last four are 89.5.




San Francisco we can see has been on a slow but pretty steady improvement since mid-year.  Their last four, three on the road, average 90.6, but take their three best of their last four and they are 92.4.

So you can make the argument that the 49ers are playing at the same level Seattle is, but given the game is in Seattle, you have to give the advantage to Seattle.  But San Francisco just proved they can throw up a great result, 99.1, on the road last week so if they can do that again, they have a shot.




Denver took a small drop late in the year, and the San Diego loss was their worst result of the year, but they avenged that loss last week.  Their last three average is 89.9 and with the San Diego loss it is 85.8.




New England improved slightly late in the year with 3 good results.  Their average from those three is a very good 93.6 but throw in the loss and they are at 89.6.

So based on their late season results, New England would seem to be the stronger team, but the game is in Denver so there is some advantage to that.  By the numbers, this game should be closer and has the better chance of the upset.

What do you think?

Wednesday, January 15, 2014

More Interesting USTA League Stats - Matches played in 2013

As we begin the 2014 USTA League year, it can be interesting to look back at 2013.  I've written a bunch of posts on other stats, so this is another in that sequence.

In total, I show 637,310 individual matches/courts played in the 18 & over and 40 & over leagues.  This was the first year of the introduction of the 40 & over league and it was very popular with 192,014 individual matches played.

Where were the most matches played?  Digging in to the 18 & over counts, the Southern section leads the way with over 128K individual matches played.  Texas and the Midwest follow both over 50K and then both NorCal and Inter-Mountain are over 30K.

The least?  Hawaii tallied just over 1,200 and the Caribbean just over 3,500.

Tuesday, January 14, 2014

Are there USTA League teams with an entire roster of self-rated players? More Interesting USTA League Stats

The short answer to the question in the title is yes, and many of you may have seen one before in your flight or area.  But how common is it?

With the data I have from creating Estimated Dynamic NTRP Rating Reports, I went about trying to find out and here is what I discovered.

For the 2013 season in the 18 & over and 40 & over leagues, I found 163 teams that were comprised of entirely self-rated players.  The next obvious question is, what levels are these teams at?  As you might expect, most are at the levels where new players typically join the USTA, those being 2.5 and 3.0.  However, I did find both a 3.5 team and 4.0 team that were entirely self-rated.

I then relaxed the criteria a bit and looked for teams that had just one or two players that weren't self-rated.  This count was 209 and here there was also just one 3.5 team but two 4.0 teams.

Are there other USTA League stats you are interested in?  Leave a comment!

Tuesday, January 7, 2014

It is 2014, let USTA League tennis begin

With the start of the new year, a new USTA League season begins.  Ok, a bunch of sections have early start leagues and so a bunch of "2014" matches have already been played.  But in many sections, the main leagues don't start until January or later and that is now.

In my district (Northwest Washington in the Pacific Northwest section) our schedules for our 18 & over and 40 & over leagues just came out and some teams play as early as this weekend, so our season will be off and running.

And based on some of the inquiries and requests for Estimated Dynamic NTRP Rating reports (ok, and looking at schedules too), many other districts are getting started soon too.

I've had a number of people get team reports in preparation for the season to get a good handle on their own team or to begin scouting opponents.  Another great way to scout opponents is to get a head-to-head report showing historical matches between the players on two different teams.  During league play, these can be very useful in looking for the best match-ups and establishing line-ups.

And the reports are not limited to my district or section.  Reports can be generated for any USTA League player or team.

As always, contact me if you are interested in any of the above reports or anything custom as well.