New format for USTA League Nationals for 2018 - What might happen?
BeyondTheTape
Semi-Pro
It will be interesting with the new format this year with the randomized schedule. In previous years, did the teams with the highest top-10 players make the semis at a higher % than teams this year with the new format. That is what I am curious about, how big of a factor the strength of schedule will be to eliminate top teams who would have made the semis in previous flighted years.
J011yroger
Talk Tennis Guru
schmke
Legend
To be honest, I didn't simulate prior years like I'm doing this year as frankly, it was a lot more boring to simulate a 4 team flight with little chance for controversial ties.
But, the team I identified as the top team in each flight would win the flight anywhere from 60-80% of the time, which, depending on how much higher they were than the #2 team is probably about what you'd expect. There were surprises for sure as some teams/players would be underrated coming in to Nationals.
The difference this year is that in the past, one upset could keep a top team out as the loss would usually mean they don't win the flight. Now, since it is generally pretty rare that there are 4 (or more) undefeated teams (generally <10% for all, much lower for many), one loss teams will still have a shot and strength of schedule is key now. For example, a 3-1 team with an easy schedule that wins most of their matches 5-0 or 4-1 and loses 3-2 in their one tough match is going to win the tie-breaker over a "better" team with a tough schedule that ekes out 3-2 wins against the top opponents to go with that one 3-2 loss.
J011yroger
Talk Tennis Guru
Vox Rationis
Professional
Schmke, your thread about two teams playing twice made me wonder, how do they plan on seeding the semifinals? Random draw? Seeded based on the tiebreakers? Also how did they do it in the past? I'm assuming it was random.
schmke
Legend
Good question, I'll try to find out.
In the past it was predetermined that flight 1 and 3 would play one semi and flight 2 and 4 would play the other, so it was known in advance, and to the degree flight assignments were random, the semi assignments were too.
Vox Rationis
Professional
Thanks!
schmke
Legend
From what I've learned, it appears they will take #1 vs #4 and #2 vs #3 in the standings regardless of if they played each other in round-robin.
BeyondTheTape
Semi-Pro
J011yroger
Talk Tennis Guru
Just to be clear, it doesn't matter how many sections send teams, everyone plays 4 matches?
J
schmke
Legend
Yep, that appears to be the case. That is an added benefit of this format, no matter the number of teams attending, everyone plays 4 matches.
schmke
Legend
So the answer is yes!
Just wrote about it and looked at the 3.5 and 4.0 men and women. Go read the blog for all the details, but in some cases the schedule strengths (average top-10 average for each opponent) have ranges of at least 0.10 and as much as 0.13. That is a pretty big difference and some teams will have a lot easier path to the semis.
If you apply a more systematic approach to the "random" and use a little knowledge of who the stronger/weaker teams are, it is easy to reduce that 30-40%, I did it for the 3.5 women and dropped the schedule strength range from 0.12 to 0.08.
But if you let me at it, I ran a bunch of different random schedules to see which produced the most equitable and got every event's range down to 0.02 or 0.03. That would be a lot more fair IMHO.
CHtennis
Rookie
Here is the link to the relevant blog post: http://computerratings.blogspot.com/2018/09/schedule-strengths-for-usta-league.html
CHtennis
Rookie
I would love to see some seeding of some sort. You could theoretically play the top 4 teams and another team could play the bottom 4 teams now and that seems like an inequality that should not be possible. I did not mind that randomness when one team had to come out of a pod but this way you could go 3-1 versus the top four teams and miss out because of your bad tiebreakers.
J011yroger
Talk Tennis Guru
NumbersGuy
Semi-Pro
So far there have been 11 Nationals under the new format, by my count.
23 teams finished 4-0, and all went through to the semis.
52 teams finished 3-1 and 40% went to the semis.
No team that finished 2-2 went to the semis.
23 teams finished 4-0, and all went through to the semis.
52 teams finished 3-1 and 40% went to the semis.
No team that finished 2-2 went to the semis.
schmke
Legend
Yep, pretty much as expected. Four 4-0 teams has not happened yet (but could, not a microscopic chance), let alone five (around 1% depending on the event), and the 2-2 is possible in theory but very slim chance too. The ties at 3-1 have usually been around 5 teams which means the tie-breakers used are a big factor. But we've only had the flawed tie-breaker procedure come into play once so far.
NumbersGuy
Semi-Pro
And a little more statistics for those headed to Nationals and wondering "how well should we expect we have to do to make the semis?":
% of lines lost.........% of 3-1 teams making semis
<=20%..........................100%
25%.................................80%
30%.................................60%
33-40%...........................33%
42%+.................................0%
For example, if you were 3-1 and were 14-6 in lines won-lost (30% lost), you had a 60% chance of making the semis, all else being equal. Lose one line fewer and that goes up to 80%.
% of lines lost.........% of 3-1 teams making semis
<=20%..........................100%
25%.................................80%
30%.................................60%
33-40%...........................33%
42%+.................................0%
For example, if you were 3-1 and were 14-6 in lines won-lost (30% lost), you had a 60% chance of making the semis, all else being equal. Lose one line fewer and that goes up to 80%.
Similar threads
