Wednesday, June 12, 2019

NorCal USTA Un-flighted Round Robin - What should have been expected?

I just wrote about NorCal using the new un-flighted round-robin format for 40 & Over Districts last month where a 4.0 mens team went undefeated and is not going to Sectionals as there were four others and they finished 5th.

That seems like a bad thing, so I went about researching if this was a freak occurrence, or if it could have been known in advance that it was likely to happen.

I'll say up front, just looking at a situation where there are 22 teams, each only playing 3 matches against random opponents, it screams to me that it was likely for there to be a lot of undefeated teams, but let's do some real analysis to see just how likely.

To do this, I'm using my Flight Simulation Report that uses my Estimated Dynamic NTRP Ratings for the players on each team to go through one million simulations of each team's schedule ,with some random variability into how each team will do, tallying up what the most likely record will be for each team and what the standings will be.  Also, as part of doing this, I can look at how often specific situations occur, like what percentage of the time there are varying numbers of undefeated teams.

So what did it all say?


The simulation said the most likely resulting records would have 4 teams all 3-0 which would have been nice and clean.  The 4 teams in this most likely scenario were San Carlos, Bay Club, Seascape, and Walnut Creek and it turns out 3 of these teams did all finish undefeated.  So the simulation isn't too bad!

The chances of any 4 teams finishing undefeated was 41%, so pretty likely, but a lot of room for other scenarios.  Here is how the others were predicted:
  • Three - 20%
  • Four - 41%
  • Five - 30%
  • Six - 5%
  • Seven - 0.3%
  • Eight - 0.01%
Yes folks, with this format and the schedule they had, there was the possibility of eight (8) teams ending up undefeated!  Yes, it is a very slim chance, and similarly for seven, but still technically possible.

The bigger thing is that a still substantial six undefeated had a 5% chance, meaning if you hold this event 20 times it is likely that you'd have six undefeated teams once.  Having a format where that is reasonably possible is just silly!

The chance of five undefeated was an even higher 30%, nearly as likely as the most likely scenario of four undefeated.  This is in fact what happened and clearly it was not a freak occurrence but should have been reasonably expected it could happen.  Add up the chances of five or more and you are over 33%, so this is saying one of these was expected to occur in one of every three times the event would be played.

With this many teams playing just three matches, the chances of a tie for fourth place (the last spot advancing to Sectionals) is very high, and when it occurs, it is more often that not going to be very big.  Basically any time there aren't exactly four undefeated, you are going to have a large tie at 2-1.  The number of teams tied could in theory be as large as 16, but is more typically expected to be between 5 and 10 teams, still very big and highlighting the inadequacies of the current tie-breakers, especially when just three matches are played as that gives very little information to differentiate with.

What about other flights?

The women's 4.0 flight had 20 teams and fortunately only had three undefeated teams.  But what did the simulation say was likely?

Here are the chances:
  • Three - 36%
  • Four - 38%
  • Five - 11%
  • Six - 1.3%
  • Seven - 0.05%
  • Eight - 3 of the million scenarios
We see the chances of of "bad" situations is quite a bit lower, this is in part to having two fewer teams, but also the strength of the teams and how the schedule was laid out.  Still, there was a significant chance, 12+%, of five or more teams being undefeated even if one of the more likely three actually occurred.  Hold this even eight times it is likely that once you will have a "situation".

Another 20 team flight was the 3.5 women.  This ended up having a nice and clean four undefeated teams, but what were the possibilities?
  • Three - 38%
  • Four - 35%
  • Five - 10%
  • Six - 0.7%
  • Seven - 0.01%
  • Eight - 1 of the million scenarios
This is quite similar to the women's 4.0 flight, and one of the most likely scenarios occurred.  But there was still an 11+% chance of five or more, so it is likely to happen one of every nine times.

In fact, there was a team that was 2-1 in 5th place that lost a team match 3-2 on a super tie-break loss.  Had they won that super tie-break, there would have been five undefeated teams and controversy would ensue again.

I did not go through the other flights, there were fewer teams there so less chance of problems occurring, although the chances were by no means zero.

I think the above shows though that it was a little short-sighted to implement this format the way they did.  So many teams playing un-flighted round-robin against only three opponents is ripe for this to happen and it seems having a format planned with these problems built in is not really fair to the participants.

Simply playing four matches per team would have significantly improved things and reduced, but not eliminated, the chances.  I wonder why that wasn't done?

What do you think?

1 comment: