Simulating 2018 USTA League Nationals - 18 & Over 5.0+ Women

My tenth simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 5.0+ women.  You can see all simulations here.

This event has the just 14 teams, Hawaii, Caribbean, and Missouri Valley seem to be missing.

Using my ratings' top-7 averages (5.0+ plays three courts so using top-7 rather than top-10), the actual schedules, and running things through a million iterations, the following is observed.

The range of top-7 averages is very large for this group, 4.49 to 4.99, and those aren't outliers as there are others in the 4.90s and 4.50s.  That is a big range!

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The chances are exceedingly slim though, just 0.01% for 5, and there is just a 3.1% chance for four to be undefeated.  With just 14 teams, it would have been possible to guarantee no more than four teams could be undefeated but that wasn't done.

In an ideal world, there are the 4 undefeated teams and there is no controversy, but as noted above, there is just a 3.1% of this.

This helps result in an 95% chance of having a tie for 4th place.  The most likely number of teams in the tie is 3 (30%), and the chances 4 are also very good at 28%, so there is a good chance of a good sized tie, 5 even has an 11% chance.  The largest number the simulation shows is "just" 8 with just 14 rather than 17 teams.

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 14% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker probably most of that 14% of the time.  The number of teams in this tie-breaker will usually be two 92% of the time.

It is also interesting to see how many wins the 4th place team has most often, and 97% of the time they'll have 3 and just 3.2% of the time they'll have 4, and in 15 of the million simulations, 4th place had just 2 wins!

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be Texas, SoCal, and Florida.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

Schedule strengths for USTA League Nationals - Should they be more even? They aren't as it stands

The USTA is rolling out a brand new format for USTA League Nationals this year, and I've written about the pros and cons of it, and subsequently have been doing a bunch of simulations to see how things would play out.

Something that has become clear is that with the new flightless random round-robin, the schedule strengths for each team can vary pretty wildly.  Using my ratings and calculating the average of the top-10 players for each team, the range can be pretty large.  Here are some examples from the 18 & Over events:

• 3.5 women - 3.43 to 3.69 (0.26)
• 3.5 men - 3.46 to 3.70 (0.24)
• 4.0 women - 3.84 to 4.04 (0.20)
• 4.0 men - 3.95 to 4.14 (0.19)

We see some pretty large ranges.  Even taking into account that my ratings are not a perfect predictor of results (but they've done pretty well), it appears some teams are clearly better than others.

These large ranges of top-10 averages coupled with a random round-robin format can lead to dramatically different schedule strengths too.  I'm calculating a schedule strength as the average of the top-10 averages for the teams scheduled to play.  Here are those ranges:

• 3.5 women - 3.51 to 3.63 (0.12)
• 3.5 men - 3.50 to 3.63 (0.13)
• 4.0 women - 3.89 to 4.00 (0.11)
• 4.0 men - 3.97 to 4.07 (0.10)

These are smaller than the ranges for the team strengths as you would expect, but are still pretty broad.  In some cases strong teams have tough schedules, but the "random" has worked and they often have easier schedules too.  Correspondingly, lower rated teams sometimes have easier schedules, but often tougher ones.

If you believe in random and luck of the draw, what has been done clearly provides that, but is that what is most fair or equitable?  Should some effort be made to have more balanced schedules so strong teams with easy schedules aren't virtually guaranteed to advance, and weaker teams with strong schedules are almost destined not to?

If you were to vote for something more balanced, I think it could be done pretty easily.

I've already written about one way to do the random round-robin by laying out the teams 1-17 and then having each team play the 2 teams ahead/behind them in the list, and if you just seed the teams at specific spots as you place them into the list, you can get more equitable schedule strengths.

For example, the 3.5 women using a standard placement I've come up with results in a schedule strength range of 3.52 to 3.60, just 0.08 vs the 0.12 the current schedule results in.  I didn't do this for each event, but I'd expect a similar shrinking of the gap for the others.

A drop of 30-40% is a big improvement, but we can do better.

If instead I simulate a bunch of different schedules looking for the most equitable (smallest gap between hardest/easiest), here is what we could do:

• 3.5 women - 3.54 to 3.57 (0.03)
• 3.5 men - 3.57 to 3.60 (0.03)
• 4.0 women - 3.95 to 3.97 (0.02)
• 4.0 men - 4.02 to 4.04 (0.02)

Now this is a real improvement, it really makes the schedules more equitable and should, in theory at least, put each team on equal footing giving resulting in even closer competition.

You might ask, how would you seed the teams?  Well, if I can calculate top-10 averages for each team and do it, the USTA with the actual dynamic ratings for all the players certainly could too.

Would this really be better than the random round-robin they've done?  We'll never be able to say for sure as we can't have the actual matches played both ways, but if we see some really lopsided results at Nationals, we'll probably know some strong teams got easy schedules and weak teams got tough schedules, and I'm certain what I've outlined would address that.

Will this happen?   I doubt it.  But now that I've written about it it is out there and they could elect to in the future.

What do you think?

Simulating 2018 USTA League Nationals - 18 & Over 5.0+ Men

My ninth simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 5.0+ men.  You can see all simulations here.

This event has the just 16 teams, Hawaii seems to be missing.

Using my ratings' top-7 averages (5.0+ plays three courts so using top-7 rather than top-10), the actual schedules, and running things through a million iterations, the following is observed.

The range of top-7 averages is very large for this group, 4.62 to 5.05, but there six at 4.90 or higher so it isn't completely a runway for the top teams.

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The chances are exceedingly slim though, just once in the million simulations were there 6, and there is just a 2.4% chance for four to be undefeated.  Note that there is no way to guarantee just 4 with 16 teams but 5 could be guaranteed.

In an ideal world, there are the 4 undefeated teams and there is no controversy, but as noted above, there is just a 2.4% of this.

This helps result in an 89% chance of having a tie for 4th place.  The most likely number of teams in the tie is a good sized 4 (33%), and the chances 3 are also very good at 31%, so there is a good chance of a good sized tie, 5 even has a 14% chance.  The largest number the simulation shows is "just" 11.

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 11% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker probably most of that 11% of the time.  The number of teams in this tie-breaker will usually be two 88% of the time.

It is also interesting to see how many wins the 4th place team has most often, and 97% of the time they'll have 3 and just 2.4% of the time they'll have 4, but in over 4 thousand of the million simulations, 4th place had just 2 wins!

Since the highest rated team could have the toughest schedule, they won't necessarily be the team most likely to make the semis.  In this case, the top team has the 3rd easiest schedule so they are a near lock to advance to the semis, and the second highest rated team has an easier schedule too.  The team with the hardest schedule is a traditional power and has a pretty slim chance of advancing to the semis due to their schedule.

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be Intermountain and Pacific Northwest.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

2018 USTA League Nationals begins in less than 2 weeks!

I've been writing about the new format for USTA League Nationals and doing simulations to see how things may play out for several months now, but soon all my writing will be moot and play will begin and the players will decide things on the court.

The events starting October 5th are the 18 & Over 5.0+ at the National Campus in Orlando, and the 18 & Over 4.0 in Arlington, TX.

I will be writing additional previews and perhaps even some predictions so stay tuned, and if you are fortunate to be going to play at Nationals, there is still time to get scouting reports on other teams or flights.  Contact me if interested.

Simulating 2018 USTA League Nationals - 18 & Over 3.0 Women

My eighth simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 3.0 women.  You can see all simulations here.

This event has the full 17 teams.

Using my ratings' top-10 averages, the actual schedules, and running things through a million iterations, the following is observed.

The range of top-10 averages is very large for this group, 2.83 to 3.21, but there are only two over 3.11, the top-2 teams having a gap back to the rest.  Will this make them the favorites?

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The chances are actually pretty good, a 3.9% chance of 5 teams finishing undefeated, and even 2,046 of the million simulations had 6 teams undefeated.  Note that there is no way to guarantee just 4 with 17 teams but 5 could be guaranteed.

In an ideal world, there are the 4 undefeated teams and there is no controversy, and the simulation says there is a very good chance at 23%.

This leaves a nearly 77% chance of having a tie for 4th place.  The most likely number of teams in the tie is a whopping 5 (38%), and the chances of 6 or 7 is over 24%, so there is a good chance of a good sized tie, perhaps the largest of all the events so far.  The largest number the simulation shows is "just" 10.

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 9% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker probably most of that 9% of the time.  The number of teams in this tie-breaker will usually be two 93% of the time.

It is also interesting to see how many wins the 4th place team has most often, and just 72% of the time they'll have 3 and 27% of the time they'll have 4, but in 7 of the million simulations, 4th place had just 2 wins!

Since the highest rated team could have the toughest schedule, they won't necessarily be the team most likely to make the semis.  In this case, the top team has one of the easier schedules so they are a near lock to advance to the semis.  The team with the hardest schedule is the weakest team so they have no shot at the semis according to the simulation.  The team with the easiest schedule is the #10 team and even being #10, they have an 80% chance of finishing at least in a tie for 4th.

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be Southern to make the semis, and Florida may surprise.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

Another "feature" of the new format for USTA League Nationals - Have to beat the same opponent twice!

The USTA has come up with a new format for USTA League Nationals this year, and I've written a bunch of thoughts about it and simulations of how it might play out, but something just occurred to me that I hadn't written about yet.

It is entirely possible, and we'll see how often it happens, that a team could beat another team in round-robin play, and end up having to beat them again in the semis or final!

One of the perks of the new format is that a team that loses early and might have ended with one loss in their old format flight and lost out to the team that beat them still has a chance to advance to the semis.  This is certainly true and I think something that will keep teams engaged longer, but in this scenario, they would make the semis along with the team the lost to.  Or put another way, the team that beat them could have to play them again if they both advance.

In a situation where that team that won in the round-robin pulled an upset, this is kind of tough as they would be asked to do it all over again, but this time the other team knows they are coming.  This doesn't seem entirely fair, but I understand you can't preclude it if you are giving teams that lost a chance to still advance.  It sort of gives an advantage to the strong teams over the underdogs as a strong team is more likely to be able to pull off the double than the scrappy underdog.

Now, what will be interesting is if the USTA will just go by standings in seeding the semis (1st vs 4th, 2nd vs 3rd), or if they would go out of their way to try to preclude a rematch in the semis and only have it occur in the final?

We'll have to wait and see.  I have to say I'm hoping it happens at least once just so I can write about it more!

9/21 Update: I've learned that they will stick with 1st vs 4th and 2nd vs 3rd even if that means a rematch in a semi-final

Simulating 2018 USTA League Nationals - 18 & Over 3.0 Men

My seventh simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 3.0 men.  You can see all simulations here.

This event has the full 17 teams.

Using my ratings' top-10 averages, the actual schedules, and running things through a million iterations, the following is observed.

The range of top-10 averages is fairly large for this group, 2.92 to 3.19, but there are only two over 3.12, the top-2 teams having a gap back to the rest.  Will this make them the favorites?

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The chances are very slim, just 0.34% chance of 5 teams finishing undefeated, so it is unlikely.  Note that there is no way to guarantee just 4 with 17 teams.

In an ideal world, there are the 4 undefeated teams and there is no controversy, and the simulation says there is a decent chance at 9.3%.

That means though that there is an over 90% chance of having a tie for 4th place.  The most likely number of teams in the tie is a whopping 5 (31%), and the chances of 6 are even 21%, so there is a good chance of a good sized tie, perhaps the largest of all the events so far.  The largest number the simulation shows is "just" 11.

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 15% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker probably most of that 15% of the time.  The number of teams in this tie-breaker will usually be two, but 14% of the time it will be 3 or more.

It is also interesting to see how many wins the 4th place team has most often, and 90% of the time they'll have 3 and 9.7% of the time they'll have 4, but in 31 of the million simulations, 4th place had just 2 wins!

Since the highest rated team could have the toughest schedule, they won't necessarily be the team most likely to make the semis.  In this case, the top teams are strong enough that even with reasonably hard schedules, they have over a 90% chance of at least being in a tie for 4th.  But the team with the best chance is actually the #6 team as they have the easiest schedule.  The #8 and #9 teams also have easier schedules and could sneak in if a favorite slips up.

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be Midwest to make the semis, but they may not be the favorite to win it all as they could face a higher rated team there. Southwest may also surprise.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

Simulating 2018 USTA League Nationals - 18 & Over 4.0 Women

My sixth simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 4.0 women.  You can see all simulations here.

This event has the full 17 teams.

Using my ratings' top-10 averages, the actual schedules, and running things through a million iterations, the following is observed.

The range of top-10 averages is actually modest for this group, 3.78 at the low end to 3.98 at the high and the distribution is pretty consistent in that range.  It is interesting no team is over 4.0.  This could make for a very competitive Nationals.

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The chances are larger than for some other events, 2.7% chance of 5 teams finishing undefeated, but still pretty unlikely.  But there is actually an extremely slim chance of 6 undefeated as 16 of the million simulations had this occur.  Note that there is no way to guarantee just 4 with 17 teams.

In an ideal world, there are the 4 undefeated teams and there is no controversy, and the simulation says there is just a pretty good chance of this happening at 17%.

That means that there is an over 82% chance of having a tie for 4th place.  The most likely number of teams in the tie is 4 (28%), but the chances of 5 or 6 is 42%, so there is a good chance of a good sized tie.  The largest number the simulation shows is 12!

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 17% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker probably most of that 17% of the time.  The number of teams in this tie-breaker will usually be two, but 20% of the time it will be 3 or more.

It is also interesting to see how many wins the 4th place team has most often, and 80% of the time they'll have 3, but in 235 of the million simulations, 4th place had just 2 wins!

Since the highest rated team could have the toughest schedule, they won't necessarily be the team most likely to make the semis.  In this case, two of the top . teams have easier schedules and make a tie for 4th over 90% of the time, but then there are two teams with over an 80% chance, one of those being the #8 team, and the top team has just a 56% chance due to a tough schedule.

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be Hawaii and New England as co-favorites with Midwest perhaps a surprise team.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

Update: I've now written a prediction of the semifinalists and winner.

Simulating 2018 USTA League Nationals - 18 & Over 4.0 Men

My fifth simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 4.0 men.  You can see all simulations here.

This event has the 17 teams.

Using my ratings' top-10 averages, the actual schedules, and running things through a million iterations, the following is observed.

The range of top-10 averages is actually not too large for this group, 3.89 at the low end to 4.06 at the high and the distribution is pretty consistent in that range.  This could make for a very competitive Nationals.

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The chances are pretty slim, just 0.07% chance of 5 teams finishing undefeated, so it is unlikely, but there is actually an extremely slim chance of 6 undefeated as 2 of the million simulations had this occur.  Note that there is no way to guarantee just 4 with 17 teams.

In an ideal world, there are the 4 undefeated teams and there is no controversy, but the simulation says there is just a 2.7% chance of this happening.

That means though that there is an over 92% chance of having a tie for 4th place.  The most likely number of teams in the tie is 4 (28%), but the chances of 3 to 6 is 85% respectively, so there is a good chance of a good sized tie.  The largest number the simulation shows is 14!

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 17% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker probably most of that 17% of the time.  The number of teams in this tie-breaker will usually be two, but 17% of the time it will be 3 or more.

It is also interesting to see how many wins the 4th place team has most often, and 97% of the time they'll have 3 and just 2.7% of the time they'll have 4, but in over 4,000 of the million simulations, 4th place had just 2 wins!

Since the highest rated team could have the toughest schedule, they won't necessarily be the team most likely to make the semis.  In this case, the top teams have relatively tough schedules and they have a good shot at the semis with the top-4 all being between 65% and 85%, but the team with the best chance of making the semis is the #7 team due to them having an easier schedule.

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be Florida and Texas.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

Update: I've now posted a prediction of who will make the semis and win.

Simulating 2018 USTA League Nationals - 18 & Over 3.5 Women

My fourth simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 3.5 women.  You can see all simulations here.

This event has the full 17 teams.  Note that the schedule as I write this has an error with SoCal playing Texas twice and NorCal with just three matches, I'm assuming that gets corrected and NorCal plays Texas.

Using my ratings' top-10 averages, the actual schedules, and running things through a million iterations, the following is observed.

The range of top-10 averages is pretty large for this group, from 3.38 all the way to 3.66, but the highest rated 9 teams are all at 3.50 or higher so it shouldn't be a runaway for the top teams.

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The chances are pretty slim, just 0.25% chance of 5 teams finishing undefeated, so it is pretty unlikely.  Note that there is no way to guarantee just 4 with 17 teams.

In an ideal world, there are the 4 undefeated teams and there is no controversy, and the simulation says there is a pretty good 19% chance of this happening.

That means that there is an 81% chance of having a tie for 4th place.  The most likely number of teams in the tie is 3 (37%), and oddly the chances of a 5-way tie are better than 4-way (27% to 18%), but still a good chance of a good sized tie.  The largest number the simulation shows is 11.

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 18% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker almost all of that 11% of the time.  The number of teams in this tie-breaker will usually be two, but 20% of the time it will be 3 or more.

It is also interesting to see how many wins the 4th place team has most often, and 81% of the time they'll have 3 and 19% of the time they'll have 4, but in 16 of the million simulations, 4th place had just 2 wins!

Since the highest rated team could have the toughest schedule, they won't necessarily be the team most likely to make the semis.  However, for this event, the top-2 teams apparently don't have hard schedules or are that much better as they have a 99% and 98% chance of at least being in a tie for 4th.  However, the next most likely is the #9 team due to a very easy schedule, the #3 and #5 teams have a good shot at the semis, but the #4 team has a tough schedule and just a 23% chance.

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be Texas.  Who are teams outside the top-6 by the ratings but might surprise?  Look for Northern to have a good shot.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

Simulating 2018 USTA League Nationals - 18 & Over 3.5 Men

My third simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 3.5 men.  You can see all simulations here.

This event has only 16 teams, Southwest is missing from the schedule as I write this, so we'll simulate with 16 rather than the normal 17.

Using my ratings' top-10 averages, the actual schedules, and running things through a million iterations, the following is observed.

The range of top-10 averages is pretty large for this group, from 3.36 all the way to 3.64, but the highest rated 9 teams are all at 3.49 or higher.

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The chances are pretty slim, just 0.13% chance of 5 teams finishing undefeated, so it is pretty slim chance  Note that there is no way to guarantee just 4 with 16 teams.

In an ideal world, there are the 4 undefeated teams and there is no controversy, and the simulation says there is an 8.4% chance of this happening.

That means though that there is nearly a 91% chance of having a tie for 4th place.  The most likely number of teams in the tie is 3 (29%), but the chances of 4 or 5 is 27% and 22% respectively, so there is a good chance of a good sized tie.  The largest number the simulation shows is 9.

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 11% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker probably that 11% of the time.  The number of teams in this tie-breaker will usually be two, but 12% of the time it will be 3 or more.

It is also interesting to see how many wins the 4th place team has most often, and 91% of the time they'll have 3 and 8.5% of the time they'll have 4, but in 25 of the million simulations, 4th place had just 2 wins!

Since the highest rated team could have the toughest schedule, they won't necessarily be the team most likely to make the semis.  However, for this event, the top-2 teams apparently don't have hard schedules or are that much better as they have a 99% and 98% chance of at least being in a tie for 4th.  However, the next most likely is the #6 team, the #3 thru #5 teams have tougher schedules and have a 70%, 79%, and 57% chance of at least being in a tie for 4th.

Interestingly, the #11 team has a 60% chance of at least tying for 4th, so they seem to have a pretty easy schedule.

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be Texas.  Who are teams outside the top-6 by the ratings but might surprise?  Look for Southern to have a good shot.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

Update on USTA League Nationals schedule changes - An error to be corrected for the 3.5 women?

I wrote last week about some schedule changes made for Nationals.  After looking some more, I've deduced a few more things.

First, several of the events now have just 16 teams as it appears a few teams have elected to not go to Nationals so no team from their sections is there, and that likely necessitated the schedule changes.

But, in at least one case, there is still an error as on TennisLink at least, the 18 & Over 3.5 women has NorCal playing just 3 matches while SoCal plays 5!  Given that SoCal is playing Texas twice, I'm guessing there was a data entry error and NorCal should be Texas' opponent for one of those.  I expect this will be corrected soon, especially since I've written about it now!

USTA League 40 & Over Sectionals affected by weather and not complete yet

Several sections held 40 & Over Sectionals this weekend, or were scheduled to, but weather got in the way.

Mid-Atlantic appears to have delayed theirs due to fears that Florence would move north and rain everything out (it appears that didn't happen), and Texas played theirs but despite going to short sets, exceptionally short in some cases, was not able to get everything completed.

I believe Mid-Atlantic has rescheduled theirs for the end of the month which doesn't give teams much time between Sectionals and Nationals.

I was told Texas is still working on what they will do.  Perhaps they try to pick a team based on the matches played, or find some way to finish the matches still to be played.  Either is pretty tough, but to my knowledge, to send a team they need to pick one.

I'll write more as I learn more.

Simulating 2018 USTA League Nationals - 18 & Over 4.5 Women

My second simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 4.5 women.  You can see all simulations here.

This event has the full set of 17 teams.

Using my ratings' top-10 averages, the actual schedules, and running things through a million iterations, the following is observed.

The range of top-10 averages is pretty large for this group, from 4.23 all the way to 4.48, but there is a big group in the middle with 7 between 4.33 and 4.37.

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The ratings are top heavy enough that the chances are pretty good of 5 teams finishing undefeated, 11%, but there is actually a chance for 6 to finish without a loss!  In the million simulations, it only happened 393 times so not a huge chance, but a lot higher than the 4.5 men.  Note that it would have been possible to have a schedule that precluded 6 teams from being undefeated, but with 17 teams, there is no way to guarantee just 4.

In an ideal world, there are the 4 undefeated teams and there is no controversy, and the simulation says there is a 38% chance of this happening.

That means though that there is a 62% chance of having a tie for 4th place.  The most likely number of teams in the tie is 5 (44%), but there is a 16% chance there is 6 or 7 in the tie.  In some really rare cases, the simulation showed there could be as many as 10 teams tied for 4th!

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 9% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker probably that 9% of the time.  The number of teams in this tie-breaker will usually be two, but 16% of the time it will be 3 or more.

It is also interesting to see how many wins the 4th place team has most often, and just 51% of the time they'll have 3 and 49% of the time they'll have 4, this is a top-heavy event with favorites expected to advance.  And in just 2 of the million simulations, did the 4th place team have just 2 wins.

Since the highest rated team could have the toughest schedule, they won't necessarily be the team most likely to make the semis.  But in this case the top rated team does have the best shot at being at least in a tie for 4th at 98%.  The surprise is the 3rd highest rated team has a really tough schedule, they have to play both the #1 and #2 teams, and only a 36% chance of being at least 4th.  The 6th and 7th rated teams are much higher at 76% and 69% respectively.

Similarly, lower rated teams aren't completely out of the running.  The 13th rated team actually has a 5% chance of at least tying for 4th.

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be Southern, and the team outside the top-5 with the best shot would be Missouri Valley.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

Simulating 2018 USTA League Nationals - 18 & Over 4.5 Men

My first simulation of USTA League Nationals using the new format for 2018 is the 18 & Over 4.5 men.  You can see all simulations here.

This event has only 16 teams, Hawaii is missing from the schedule as I write this, so we'll simulate with 16 rather than the normal 17.

Using my ratings' top-10 averages, the actual schedules, and running things through a million iterations, the following is observed.

The range of top-10 averages is pretty large for this group, from 4.31 all the way to 4.53, but the highest rated 7 teams are pretty close, 4.47 to 4.53.

With the actual schedules the USTA set up, there are ways for more than 4 teams to finish undefeated, meaning a team that doesn't lose could be sent home without making the semis.  The chances are pretty slim, just 0.05% chance of 5 teams finishing undefeated, but there is actually a chance of 6 to finish without a loss!  In the million simulations, it only happened 3 times, but still, it is possible.  Note that it would have been possible to have a schedule that precluded 6 teams from being undefeated, but with 16 teams, there is no way to guarantee just 4.

In an ideal world, there are the 4 undefeated teams and there is no controversy, and the simulation says there is a 10.5% chance of this happening.

That means though that there is nearly a 90% chance of having a tie for 4th place.  The most likely number of teams in the tie is 3 (32%), but there is a 54% chance there are 4 to 6 teams in the tie.  In some really rare cases, the simulation showed there could be as many as 11 teams tied for 4th!

When there is a tie for 4th, the first tie-breaker is courts won/lost and in 15% of the cases the 4th and 5th teams will be tied on courts meaning it will go to head to head, but the teams are unlikely to have played, so it will go to the sets tie-breaker probably that 15% of the time.  The number of teams in this tie-breaker will usually be two, but 13% of the time it will be 3 or more.

It is also interesting to see how many wins the 4th place team has most often, and 90% of the time they'll have 3 and 10% of the time they'll have 4, but in 226 of the million simulations, 4th place had just 2 wins!

Since the highest rated team could have the toughest schedule, they won't necessarily be the team most likely to make the semis.  And in this case, the 3rd highest rated team and the highest rated team have basically the same chance of at least being tied for 4th at 97%.  The 2nd highest rated team has a tough schedule and only has a 78% chance of finishing 4th.  The 4th highest rated team is even worse though, just a 44% chance while the 5th place team has an 87% chance.

Similarly, lower rated teams aren't completely out of the running.  The 13th rated team actually has an 8% chance of at least tying for 4th, and the 10th rated team a 19% chance.

That is a lot of interesting but unspecific information, so lets just get to it to see who the simulation says the favorite is.  That would be co-favorites in this case, Southern Cal and Florida.  Who are teams outside the top-5 by the ratings but might surprise?  Look for Southern and Caribbean to have a good shot.

If you'd like to get more details of the simulation and each team's chances, or are interested in a flight or team report to help scout your opponents, contact me.

Simulating USTA League Nationals - Methodology and list of events

2018 USTA League Nationals will be starting in less than a month, and with it we will see the new format in action.  As a reminder, the new format has no flights like past years, instead each team plays 4 random opponents and the top-4 teams in the standings across all the teams advance to the semis.  Read what I wrote earlier with thoughts on the format in general.

But with the new format, without a clear cut four flight winners, it will certainly be a lot more interesting to see what happens.  I did some general 'what if' type of simulations using fictitious teams/ratings that mirrored what we often see at Nationals and that had some interesting results, but with flights and schedules showing up on TennisLink, I can ditch the fictitious simulations and do some real ones.

My methodology for the simulations is as follows.  Using my ratings, I calculate the top-10 average for each team and then play out each team's schedule picking a winner for each match.  I do a million iterations of this, randomly varying a team's top-10 average up or down a bit each match to simulate players' normal variation in play from match to match and allow each iteration to be randomly different.

The result is I can look at how often different scenarios occur like how often there will be 5 (or more!) undefeated teams and a team will get left out of the semis that doesn't lose a match, or how often there is a tie for 4th and how many teams will see their fate decided by questionable tie-breakers.

How does this different from my flight reports which many teams have used to get to and win Nationals?  First, it is using real schedules and who plays who so strength of schedule is factored in and the team with the highest ratings may not be the most likely to make the semis, or a mid-pack team may have a great shot at making the semis due to a weaker schedule.

Also, my flight reports provide a full roster average and a top-8 average to give a feel for a teams overall strength as well as how they rate when they run their best out.  For these simulations I'm using a top-10 average as few if any teams will only play their top-8.

So that explains things, and I think it is going to be fun to see what the simulations reveal.  I'm going to try to get to every division/gender/level, but will be starting with those that begin first and/or that people show interest in (hint hint, leave a comment and share with your friends so they can too), and as I complete them I'll add links to them here so this blog entry can serve as the roadmap on the methodology and all the simulations done.

Simulations and predictions written:

• 18 & Over 3.0 men - prediction
• 18 & Over 3.0 women - prediction
• 18 & Over 3.5 men
• 18 & Over 3.5 women - prediction
• 18 & Over 4.0 men - prediction
• 18 & Over 4.0 women - prediction
• 18 & Over 4.5 men
• 18 & Over 4.5 women
• 18 & Over 5.0+ men - prediction
• 18 & Over 5.0+ women - prediction

Also, see where I've written about schedule strengths and a proposal for making them more equitable.

For those fortunate to be headed to Nationals, should you be interested in doing some scouting with a flight report, team report, or more details from my simulations than I write up, contact me and I'd love to see how I can help make your Nationals experience a winning one.

More observations on the 2018 USTA League Nationals format - Which gender/level should I simulate first?

With some of the Nationals events having their schedules updated, it appears things really are getting finalized, and that means actual teams and schedules are showing up on TennisLink.

That means I have details to be able to do some real and not just hypothetical simulations of who has the best chance of making the semis based on not just strength of roster, but also who has the most difficult or easiest schedule.

I'm doing some trial simulations using top-8 averages from my ratings right now and am seeing a few things that are interesting.  These include:

• Not only is it possible for 5 teams to finish undefeated and one team be left out of the semis on a tie-breaker, but for some of the events, 6 can finish undefeated!  It seems while it was possible to ensure at most 5 could finish undefeated, the schedules were not done this way.  To be fair, what I've seen so far is the chances of it happening are no more than 1 in 100,000, but it is still possible.
• The chances of a tie for 4th can be pretty high, I've seen greater than 90%.
• The chances of a 6-way tie for a semi spot can be pretty high, I've seen 10%.
• The chances of a large tie for 4th is not zero, I've seen 10 or 11 way ties show up occasionally.
• The highest rated team is not always the favorite to make the semis, and correspondingly I've seen some mid-pack teams have very good chances at semis.  This is all due to "luck of the draw" and top teams having hard schedules, or mid-pack teams having easy schedules.

I'll be posting more details on my simulations soon, and I'll start with the gender/level that has the most interest, so leave a comment here or on Facebook and let your voice be heard!

USTA make changes to 2018 Nationals schedules and "flights"

I wrote last month about the 2018 USTA League Nationals schedules being available.  Well, despite "final" being in the same of the PDFs the schedules were in, it appears changes have been made to at least some of the events.

To get the latest, go to the USTA's Nationals page and click on the "Schedule" link for your event.  If the page that comes up still shows "Final" in the URL, I believe it is unchanged.  But if it has "updated" with a date, it has changed since the first release.

My quick check as of this writing shows the following have changed:

• 18 & Over 5.0+ (8/18 update)
• 18 & Over 2.5 (9/6 update)
• 18 & Over 3.5 (9/6 update)

I have not checked to see what specifically changed, but I believe it was originally possible for 7 teams in one of the men's events to finish undefeated, so perhaps that was changed.  Or sometimes a section will elect to not send a team so changes are required for that.  I will try and check more closely shortly.

If you did happen to contact me and get reports to scout opponents for Nationals already and the schedule changes have affected the teams you will play, let me know and we'll work something out.