Feel free to read what I wrote earlier, but the summary is that unlike past years where the 17 sections were split into four flights, three with four teams and one with five, this year there will be no flights and instead, each team will play a random round-robin of four matches.
There will still be four teams advancing to the semi-finals, but without fixed flights, it will just come down to standings from the random round-robin play to determine the top-4 teams. The standings will be determined using the traditional team win/loss, then individual court win/loss, then head-to-head, then sets lost, and then games lost. While this method of determining the standings works fine for regular season flights and playoff flights where everyone plays each other, the big question, in my mind at least, is if this will be equitable in a format where all the teams don't play each other (more below so keep reading!).
The good news is that with this format, all teams will play the same number of matches, four, instead of most of the teams having just three matches. So this means more opportunity to get everyone a match or two if you have a large roster. It also means that, in theory at least, a team that lost a match in a tough flight and missed out on making the semis to the team they lost to now has a chance to still find their way in to the semis.
So how might this all play out? I discuss some possible scenarios below, and I've actually done simulations with millions of iterations to see how common some of the scenarios are.
Without going into too much detail, my simulations assign a representative strength (based on frequent top-8 averages I've observed) to each team and then have some built in variability of a team playing better/worse than their strength for a given match. All the matches are then simulated a million times and I take a look at the chances of different scenarios. On to some observations!
First, it is possible for there to be five teams that finish undefeated and in this scenario, one of those teams will be left out of the semis. Under the old format, going undefeated in your flight would guarantee the semis and now that isn't the case, so should this happen, that would be tough to swallow, especially given the questionably equitable tie-breakers (see below). What are the chances this happens? My simulations say there is a 1-2% chance, so it certainly won't happen often, but with nearly 40 different Nationals to be played each year, there it is likely to happen once every year or two. We'll have to see if my simulated chances come to fruition.
Second, the ideal scenario to avoid any controversy would be for there to be four 4-0 teams. The chances of this are actually reasonably high at 16%, so about 1 in 6 events we'll have a nice clean set of semi-finalists. That still leaves a lot of times where it won't be clean and tie-breakers will come into play.
Third, 84% of the time it appears there will be a tie for fourth. This will often be just a few teams, but there are scenarios where it is more than a few and even more than half the 17 teams! In fact, it is exceedingly unlikely, but my simulations show it is possible for there to be a 12 way tie for fourth! Won't that be a fun tie-breaker to figure out.
Here is the break out of how often the various number of tied teams may occur when there is a tie for fourth:
There will still be four teams advancing to the semi-finals, but without fixed flights, it will just come down to standings from the random round-robin play to determine the top-4 teams. The standings will be determined using the traditional team win/loss, then individual court win/loss, then head-to-head, then sets lost, and then games lost. While this method of determining the standings works fine for regular season flights and playoff flights where everyone plays each other, the big question, in my mind at least, is if this will be equitable in a format where all the teams don't play each other (more below so keep reading!).
The good news is that with this format, all teams will play the same number of matches, four, instead of most of the teams having just three matches. So this means more opportunity to get everyone a match or two if you have a large roster. It also means that, in theory at least, a team that lost a match in a tough flight and missed out on making the semis to the team they lost to now has a chance to still find their way in to the semis.
So how might this all play out? I discuss some possible scenarios below, and I've actually done simulations with millions of iterations to see how common some of the scenarios are.
Without going into too much detail, my simulations assign a representative strength (based on frequent top-8 averages I've observed) to each team and then have some built in variability of a team playing better/worse than their strength for a given match. All the matches are then simulated a million times and I take a look at the chances of different scenarios. On to some observations!
First, it is possible for there to be five teams that finish undefeated and in this scenario, one of those teams will be left out of the semis. Under the old format, going undefeated in your flight would guarantee the semis and now that isn't the case, so should this happen, that would be tough to swallow, especially given the questionably equitable tie-breakers (see below). What are the chances this happens? My simulations say there is a 1-2% chance, so it certainly won't happen often, but with nearly 40 different Nationals to be played each year, there it is likely to happen once every year or two. We'll have to see if my simulated chances come to fruition.
Second, the ideal scenario to avoid any controversy would be for there to be four 4-0 teams. The chances of this are actually reasonably high at 16%, so about 1 in 6 events we'll have a nice clean set of semi-finalists. That still leaves a lot of times where it won't be clean and tie-breakers will come into play.
Third, 84% of the time it appears there will be a tie for fourth. This will often be just a few teams, but there are scenarios where it is more than a few and even more than half the 17 teams! In fact, it is exceedingly unlikely, but my simulations show it is possible for there to be a 12 way tie for fourth! Won't that be a fun tie-breaker to figure out.
Here is the break out of how often the various number of tied teams may occur when there is a tie for fourth:
- 2 - 6%
- 3 - 21%
- 4 - 31%
- 5 - 25%
- 6 - 13%
- 7 - 4%
- 8 - 1%
- 9-12 - <1% (but still possible!)
- (yes, some numbers above are rounded up so the total is >100%)
So the most likely number of teams tied for fourth is 4, but we can expect to see 5 teams tied 25% of the time there is a tie for fourth.
Fourth, since it is a random four match round-robin, some teams will have tougher schedules and some will have easier schedules. In fact, my simulations show that the "best" team may not make the semis nearly 10% of the time because they have a tougher schedule and lose a match while other on paper weaker teams with easier schedules go undefeated or beat the stronger team out in a tie-breaker. Correspondingly, the weakest team there actually has a chance, albeit a very small one, of getting an easy schedule and making the semis. So just as the old format, there is still a "luck of the draw" factor that will significantly influence who will advance.
Fifth, we see that tie-breakers are going to come into play a lot of the time (84% for those paying attention). The first tie-breaker is individual courts, but with just four matches played, my simulations show that won't decide things a full 31% of the time there is a tie. It then goes to head-to-head, but with each team only playing a quarter of the opposition, it is more likely than not that there won't be a head-to-head match to break the tie. In fact, it is possible the tied teams may not have even played any common opponents! That means we'll go to the sets lost tie-breaker.
Because regular season flights play more matches, and because regular season flights and normal playoff flights all have head-to-head matches, the sets tie-breaker is rarely used and for that reason, I don't think people understand that it is fundamentally flawed.
Consider two teams, A and B, that both go 3-1 in their matches, and both go 13-7 on individual courts. Let's assume for simplicity that both teams have all their wins be in straight sets, but team A has all their losses be in straight sets while team B has all their losses be in third set tie-breaks. Both teams will have lost 14 sets so by the sets lost tie-breaker they are tied, but team B, in my opinion at least, did a lot better as they took all their losses to third sets and they won a lot more sets than team A did. But the sets lost tie-breaker gives them no credit for that.
Worse, if we modify the above scenario to have team B lose just one set in one of their wins, they now have lost 15 sets and lose the tie-breaker to team A despite team B winning 7 more sets than team A did. Does that seem right?
It seems pretty clear, and frankly pretty easy to do, that this tie-breaker should be sets won/lost differential instead of just sets lost. But because it is sets lost, and because this tie-breaker will come into play in the new format, it is not just possible but perhaps likely that the "wrong" team will win the tie-breaker and advance.
Sixth, taking it a step further, because the sets lost tie-breaker does not differentiate between our team A and B above in the first scenario, things will get to the games lost tie-breaker perhaps more often than you might expect (I haven't done a full simulation of this yet). Unfortunately, games lost suffers from the same plight as the sets lost tie-breaker, but it is even worse.
Consider a team A that loses a set 6-0 and team B loses a set 7-6. If you look at games lost, team A, that didn't win a game, will look better in this tie-breaker having lost just 6 games to team B's 7 games lost. In essence, team B is penalized for keeping the set close and forcing it to a tie-breaker.
Again, it seems pretty clear, and would be easy to do, that this tie-breaker should be games won/lost differential and that would make it a lot more equitable.
I certainly hope things go smoothly at all the Nationals this year and there is not controversy with undefeated teams missing semis or inequitable tie-breakers causing much angst and confusion over who should advance. I'll certainly be monitoring and look forward to hearing from anyone at Nationals how they think the new format works.
But what do you think now? Am I just too much of a nerd thinking about all the possibilities, and should anything strange happen it is no big deal? Or should we be worried that there will be confusion and controversy with the new format?
No comments:
Post a Comment