Monday, May 07, 2007
In defense of the "3 handler" zone offense
Note the title of this post is NOT "In defense of the traditional dump-swing zone offense". I am not arguing this. I am arguing that the traditional structure, with three handlers, two poppers, and two wings/deeps, has several advantages.
The "two handler zone offense", as explained in Jim & Zaz's book, has been steadily gaining popularity. There are plenty of good reasons for that. The best reason, really, is that the traditional dump-swing zone offense (where the middle handler just swings it back and forth and the side handlers look for a single continuation pass to a wing or popper) sucks. It's really not that hard to come up with an offense that's more effective than that, particularly when it's windy.
The other reason it has grown in popularity is that it has lots of good ideas, notably:
The key to the three-handler zone offense, in my opinion, is to have the handlers be extremely active. Basically, at any given time, one of the handlers should be acting more like a traditional popper than a handler. Let me give a handful of examples of handler action to show what I mean.
Basically, I am proposing a similar philosophy to the new two-handler set, but based out of the traditional three-handler formation. The fundamental difference is that in stead of having a full-time deep, you have an additional handler. I think that, for most teams, this is an advantage. I will give two main reasons:
The "two handler zone offense", as explained in Jim & Zaz's book, has been steadily gaining popularity. There are plenty of good reasons for that. The best reason, really, is that the traditional dump-swing zone offense (where the middle handler just swings it back and forth and the side handlers look for a single continuation pass to a wing or popper) sucks. It's really not that hard to come up with an offense that's more effective than that, particularly when it's windy.
The other reason it has grown in popularity is that it has lots of good ideas, notably:
- not losing yards on the dump
- crashing the cup and doing other things to keep the cup from staying set
- overloading an area of the defense to create an opening
- exploiting fast breaks downfield once you beat the cup
The key to the three-handler zone offense, in my opinion, is to have the handlers be extremely active. Basically, at any given time, one of the handlers should be acting more like a traditional popper than a handler. Let me give a handful of examples of handler action to show what I mean.
- One handler has the disc, forced middle, close to the middle of the field. Rather than play behind the disc for a dump, the break-side handler sets up slightly downfield, 5-10 yards directly behind the mark. This handler can move upfield to catch an inside break, or run back to catch a swing (outside break). The third handler sets up as a force-side swing, essentially playing the off-point in the same way the break-side handler is playing the mark. Late in the count, either of the handlers can move into or behind the cup for a crash or short dump.
- Forced middle on the sideline. Rather than set up dump/swing, the middle handler becomes a popper, moving downfield and attempting to find a hole in the cup or crash the cup. The far handler sets up in the middle of the field and trusts the thrower can swing it directly using an overhead or curving throw. If that pass is completed, you are in a fast break situation.
- Handler throws a short swing. The handler follows the throw and looks to receive a give-and-go pass, either inside or beyond the cup. If the give-and-go throw is completed, the cup will be out of position and the original thrower will now have an easy opportunity to hit a popper. If the cup crashes on the original thrower's cut, there should be options for a throw against the grain to a popper or a swing against the grain to the third handler.
- Wing or popper catches the disc, but doesn't have a good continuation look. Rather than throw the dump for a loss of yardage, the receiver fakes the dump, and then the handler quickly moves into the cup for a extremely short pass - often jumping just before the catch so that the handler catches the disc in the middle of the three defenders. The cup is now out of position and it's relatively easy to thread a pass to a popper or wing.
Basically, I am proposing a similar philosophy to the new two-handler set, but based out of the traditional three-handler formation. The fundamental difference is that in stead of having a full-time deep, you have an additional handler. I think that, for most teams, this is an advantage. I will give two main reasons:
- The three-handler structure leads to less reliance on overheads. Fundamentally, it requires a few more passes, with a slightly higher completion percentage per pass. It is not nearly as conservative as the traditional dump-swing zone offense, but it is closer to it than the two-handler set is. Most teams are not that great at overheads, and as such, for most teams I think this set is closer to optimal in a risk vs. reward sense.
- Having a set deep cutter makes the job of the defensive deep easier. I know that when I play the position, it is a relative relief to have one player back there who I can worry about. Yes, this keeps me from making as many plays underneath, but I am taking a cutter out in the process, and furthermore I am available to help on other deep cutters. By contrast, if I have four cutters in front of me, several of which may strike deep, it can be very hard to figure out who to play on. I may not be able to help deep on all cutters, which means the wings have to be more conservative. Furthermore, because I have to be available to help deep, I can't play in front of poppers, so I am less effective covering the underneath cutters than a short deep or wing can be.
Tuesday, May 01, 2007
The last rematch post EVER.
OK, the title is a lie. Unfortunately, this will not be the last time I write about why we play rematches is UPA formats. In fact, I will probably look back at this in a few months and feel like revising or adding to it. But this issue comes up so often, and I think it would be helpful for me to try to lay out all the main arguments in one place. So here it is.
---
Why do we play the final game to go in UPA formats even when this game is a rematch?
1) Teams should be able to plan ahead knowing that a specific sequence of wins and losses will effect a specific outcome. From both a psychological perspective and (more significantly) a physical one, this makes the tournament more fair. Teams can plan their rotations and strategy based on knowing what games they will be in depending on the results of the games they are playing. In my opinion, it makes the tournament seem more professional if the brackets are played out.
2) It introduces an element of uncertainty and randomness to the event if a team loses in bracket play, and then has to accept that their season may be over if things don't go "their way" in other games. I've often thought that it's a galvanizing moment when you lose your first game in double elimination, and someone says "OK, now we win the rest of our games". I don't want to have to respond "unless (the team that just beat us) loses in the finals… then we're fucked no matter what we do." Emotionally, having your season end with a win and being shut out of the game to go would be extrodinarily anticlimactic (and devastating).
3) Making the final game no-rematch conditional would introduce similar issues to the problems with three-way ties in pool play – namely, that something happening in another game on a different field could change where you finish.
In pool play, one loss usually puts you in second place, but if some other games go a certain way, you could end up third, and you have relatively little control over this. Similarly, if the final game to go is no-rematch conditional, then the results on another field can change whether you have a chance to advance or not. In both cases, yes, you could have avoided any potential issues just by winning all of your games. But unless we are talking about only running single elimination, one advance formats, such an argument is a non-starter.
Normally, bracket play avoids these issues – your results on the field and your results alone will decide where you finish. This is the main reason why almost every format uses brackets to determine which teams advance. Making the last game no-rematch conditional muddies the waters and, in my opinion, negates much of the benefit and simple elegance of using a bracket to figure out this spot.
4) If we make the final game no-rematch conditional, then we are introducing a fundamental inconsistency in our approach. We will be saying that rematches should be played UNLESS they are in the last game. Why is the last game different? The practical reason, of course, is that otherwise you are introducing all sorts of extra byes into the format, which is certainly more unfair than a rematch is. But if we're making this concession that avoiding rematches is a priority of the format, then in some sense we are saying, "this game is bad but you have to play it anyway".
This is not an argument against avoiding the final rematch, per se; I'm simply pointing out that pursuing the goal of avoiding rematches does not lead us to formats where we just make the final game no-rematch conditional. It leads us to single elimination brackets, or round robin formats with no brackets, or a few specialized formats with extremely abbreviated brackets. Those are the only formats where we can assure ourselves of avoiding rematches throughout.
5) "dead team/clinched team" scenarios and fairness issues.
There's several varieties of negative scenarios that can crop up if we make the last game no-rematch conditional. There are two basic variations. In both cases, AvB is the "upper" game, where the winner is done, and CvD is the "lower" game where the loser is eliminated, and the AvB loser plays the CvD winner in the game for the final spot.
a) D has lost to both A and B. D's season is over, they are a "dead team walking". They are likely to open up the rotation, or even worse, just forfeit the game. This unfairly benefits C and unfairly punishes A and B (ironically, they are being punished for their victories, not their defeats).
b) A has beaten both C and D, B has beaten neither. A has clinched advancement, and has no motivation for the next level except placement, while B is playing for a spot. B beats a relatively unmotivated A and ends the event.
There are also crosses of these scenarios. For example, A has already beaten C and B has already beaten D. (This scenario would appear in 16 teams, 2 advance if the semifinalists meet in the game to reach the backdoor finals.) Any final result in the top game eliminates one of the bottom teams, and any result in the bottom game clinches for one of the top teams – effectively moving you into one of the two scenarios I outlined above. In these cases, both the upper and lower games become extremely uncertain affairs and you have a lot of scoreboard watching. Some people might consider this sort of uncertainty and randomness exciting, but I personally think we owe the players a more predictable experience.
6) Potential gamesmanhip
If we make the final game no-rematch conditional, then the desire to avoid being shut out of that game can actually lead a team to want to intentionally lose a game. If the final game is no-rematch conditional, you have one of two ways to advance to the next level:
a) reach the "upper game" (i.e. the game where the loser is in the final game to go)
b) make sure that you don't lose to the team you expect will lose the "upper game"
If you think your chances of reaching the "upper game" are low, for whatever reason, then your best strategy could become intentionally losing to avoid your most likely opponent in the final game to go. Consider the following scenario:
The tournament is the standard 16 team, 2 advance bracket format. There is a very large talent gap between the 1 and 2, and between the 3 and the remaining teams. The 3 has a very deep roster, while the 2 relies on a small core of players. If you are on the 3 seed, it may be in your interest to intentionally lose in the quarterfinals. This will allow you to get your (only) shot at the 2 seed in the last game of the weekend, rather than in the semifinals when they are relatively fresh.
It actually takes a fairly peculiar set of relative team strengths to make this strategy viable, but in my opinion it is more realistic than any gamesmanship scenario that results from our current no-rematch policy.
---
Now, allow me to address three of the more common arguments against playing the rematch:
"If we lose the rematch, then winning the first game was meaningless"
I consider this to be the weakest argument against playing the rematch. The only sense in which this is accurate is the same sense as the statement "all of our games were meaningless since we didn't qualify".
There are multiple, tangible benefits to winning the first meeting between two teams, even if you are fairly confident you will end up having to beat the same team a second time to qualify. Namely:
1) You save a loss for later. In all the bracket formats you have a finite number of losses. If the first meeting between the teams is the front-door semifinals, then winning means you don't have to win another game to reach the final game to go.
2) In many formats, you have to play one fewer game to reach the final game to go. This isn't true of the first "3v4" game in the 16 team, 3 advance bracket format – nor is it true in the four team double elimination that is played in a few formats. But in most formats it is true, and playing one fewer game can be very significant.
3) If you lose the first game, there is a non-trivial chance that you will be upset in a subsequent elimination game and not even reach the game to go. This is a variation on the first reason I cited.
4) Conversely, if you win the first game, you have at least some chance to win later games (perhaps in an upset), and therefore qualify without even having to play in the final game to go.
5) Even if you feel your chances of winning finals or losing backdoor games are trivially small, winning the first contest gives you the opportunity to rest your starters and save yourself for the game to go. This amounts to cashing in the loss you saved by winning the first game. If you really feel that first game is meaningless, let the other team be the one to tank.
"statistical argument – 1-1 proves nothing except ambiguity"
This argument at first appears quite strong, but upon further examination it is, in my opinion, something of a mirage.
Consider a final game to go that pits two teams against each other that had played earlier in the event, with the team from the "upper game" winning the first contest. This game can have one of two outcomes:
- the team from the upper game wins, and advances with a 2-0 record against the team from the lower game.
- the team from the lower game wins, and advances with a 1-1 record against the team from the upper game.
From the perspective of this "ambiguity" argument, the first scenario is irrelevant, so let's consider only the second case.
Imagine for a moment that you have an oracle with you, and you know that the lower team will win, even before the game starts. You have the choice of either allowing the game to go on or cancelling it, on account of it being a rematch. Should you cancel it? The first instinctive answer might be yes, because you want to avoid the ambiguity.
But are you really avoiding it? No, it already exists. You know, after all, that the lower team would win. No, what you are doing is avoiding the APPEARANCE of ambiguity. The ambiguity is there whether you like it or not.
Like I said, the ambiguity argument is, to me, a mirage. Making the game no-rematch conditional is not preventing ambiguity; it is merely hiding it by protecting the 1-0 team.
A slight variant of this argument is the argument that "the rematch does not consistently help us find the better team". Again, the 2-0 upper game winner case is irrelevant. So the question really becomes: who is more likely to be the better/more deserving team – the team that won the first game, or the team that won the second game? Here I don't have an answer, but my suspicion would be the team that won the second game. I have two reasons for this suspicion:
1) Everyone knows the format and knows that the second game is the one that truly settles your fate. While I reject the idea that the first game is meaningless, the stakes are even higher in the second game, and I would place a lot of emphasis on that high stakes contest.
2) Most people who I have discussed it with agree that nationals is about the most physically gruelling tournament you will ever play in. The winner of the second game is probably the team in better shape and/or or that relies on a deeper rotation, and that team is more likely to represent their region well at nationals.
"letdown/momentum/revenge argument – the team coming into the final game on a win has an edge"
I don't want to dismiss this argument out of hand. I am definitely a believer in sports psychology. I would simply state that I don't believe this mental edge argument to be consistently true, moreover I don't think it is true consistently enough to base policy on. All of these factors (upper game loser being deflated from a tight loss, lower game winner being confident and fresh, lower game winner motivated for revenge while upper game loser is anxious or complacent) are possible. But it could just as easily be the opposite factors in play (upper game loser well rested from a quick loss, lower game winner spent from a nailbiter, upper game loser confident but not cocky while the lower game winner is in their own heads and thinks they have to play past their level to win).
Which of these factors is dominant? I don't know. You don't know, either. Nobody knows. I would like to do a study of every rematch game-to-go that's ever happened, see the statistics on the upsets, and see how the teams did at the next level. I'd like to be able to draw some conclusions on the average physical/psychological state of the teams in the rematch games, but the data is just not there.
This is actually a good idea for a UPA grant in my opinion. Somebody should get a grant, go through every UPA event that has ever been played, and try to compile the available results into a database. Obviously the results would get increasingly sparse the further back you go, but the old newsletters do have much of the results from nationals at least, going back into the 70s. It would be really cool to be able to just pull up 1990 club nationals on the score reporter.
Until that happens, though, I consider this argument speculative at best and inaccurate at worst.
---
OK, that's all, for now. I consider this a work in progress.
---
Why do we play the final game to go in UPA formats even when this game is a rematch?
1) Teams should be able to plan ahead knowing that a specific sequence of wins and losses will effect a specific outcome. From both a psychological perspective and (more significantly) a physical one, this makes the tournament more fair. Teams can plan their rotations and strategy based on knowing what games they will be in depending on the results of the games they are playing. In my opinion, it makes the tournament seem more professional if the brackets are played out.
2) It introduces an element of uncertainty and randomness to the event if a team loses in bracket play, and then has to accept that their season may be over if things don't go "their way" in other games. I've often thought that it's a galvanizing moment when you lose your first game in double elimination, and someone says "OK, now we win the rest of our games". I don't want to have to respond "unless (the team that just beat us) loses in the finals… then we're fucked no matter what we do." Emotionally, having your season end with a win and being shut out of the game to go would be extrodinarily anticlimactic (and devastating).
3) Making the final game no-rematch conditional would introduce similar issues to the problems with three-way ties in pool play – namely, that something happening in another game on a different field could change where you finish.
In pool play, one loss usually puts you in second place, but if some other games go a certain way, you could end up third, and you have relatively little control over this. Similarly, if the final game to go is no-rematch conditional, then the results on another field can change whether you have a chance to advance or not. In both cases, yes, you could have avoided any potential issues just by winning all of your games. But unless we are talking about only running single elimination, one advance formats, such an argument is a non-starter.
Normally, bracket play avoids these issues – your results on the field and your results alone will decide where you finish. This is the main reason why almost every format uses brackets to determine which teams advance. Making the last game no-rematch conditional muddies the waters and, in my opinion, negates much of the benefit and simple elegance of using a bracket to figure out this spot.
4) If we make the final game no-rematch conditional, then we are introducing a fundamental inconsistency in our approach. We will be saying that rematches should be played UNLESS they are in the last game. Why is the last game different? The practical reason, of course, is that otherwise you are introducing all sorts of extra byes into the format, which is certainly more unfair than a rematch is. But if we're making this concession that avoiding rematches is a priority of the format, then in some sense we are saying, "this game is bad but you have to play it anyway".
This is not an argument against avoiding the final rematch, per se; I'm simply pointing out that pursuing the goal of avoiding rematches does not lead us to formats where we just make the final game no-rematch conditional. It leads us to single elimination brackets, or round robin formats with no brackets, or a few specialized formats with extremely abbreviated brackets. Those are the only formats where we can assure ourselves of avoiding rematches throughout.
5) "dead team/clinched team" scenarios and fairness issues.
There's several varieties of negative scenarios that can crop up if we make the last game no-rematch conditional. There are two basic variations. In both cases, AvB is the "upper" game, where the winner is done, and CvD is the "lower" game where the loser is eliminated, and the AvB loser plays the CvD winner in the game for the final spot.
a) D has lost to both A and B. D's season is over, they are a "dead team walking". They are likely to open up the rotation, or even worse, just forfeit the game. This unfairly benefits C and unfairly punishes A and B (ironically, they are being punished for their victories, not their defeats).
b) A has beaten both C and D, B has beaten neither. A has clinched advancement, and has no motivation for the next level except placement, while B is playing for a spot. B beats a relatively unmotivated A and ends the event.
There are also crosses of these scenarios. For example, A has already beaten C and B has already beaten D. (This scenario would appear in 16 teams, 2 advance if the semifinalists meet in the game to reach the backdoor finals.) Any final result in the top game eliminates one of the bottom teams, and any result in the bottom game clinches for one of the top teams – effectively moving you into one of the two scenarios I outlined above. In these cases, both the upper and lower games become extremely uncertain affairs and you have a lot of scoreboard watching. Some people might consider this sort of uncertainty and randomness exciting, but I personally think we owe the players a more predictable experience.
6) Potential gamesmanhip
If we make the final game no-rematch conditional, then the desire to avoid being shut out of that game can actually lead a team to want to intentionally lose a game. If the final game is no-rematch conditional, you have one of two ways to advance to the next level:
a) reach the "upper game" (i.e. the game where the loser is in the final game to go)
b) make sure that you don't lose to the team you expect will lose the "upper game"
If you think your chances of reaching the "upper game" are low, for whatever reason, then your best strategy could become intentionally losing to avoid your most likely opponent in the final game to go. Consider the following scenario:
The tournament is the standard 16 team, 2 advance bracket format. There is a very large talent gap between the 1 and 2, and between the 3 and the remaining teams. The 3 has a very deep roster, while the 2 relies on a small core of players. If you are on the 3 seed, it may be in your interest to intentionally lose in the quarterfinals. This will allow you to get your (only) shot at the 2 seed in the last game of the weekend, rather than in the semifinals when they are relatively fresh.
It actually takes a fairly peculiar set of relative team strengths to make this strategy viable, but in my opinion it is more realistic than any gamesmanship scenario that results from our current no-rematch policy.
---
Now, allow me to address three of the more common arguments against playing the rematch:
"If we lose the rematch, then winning the first game was meaningless"
I consider this to be the weakest argument against playing the rematch. The only sense in which this is accurate is the same sense as the statement "all of our games were meaningless since we didn't qualify".
There are multiple, tangible benefits to winning the first meeting between two teams, even if you are fairly confident you will end up having to beat the same team a second time to qualify. Namely:
1) You save a loss for later. In all the bracket formats you have a finite number of losses. If the first meeting between the teams is the front-door semifinals, then winning means you don't have to win another game to reach the final game to go.
2) In many formats, you have to play one fewer game to reach the final game to go. This isn't true of the first "3v4" game in the 16 team, 3 advance bracket format – nor is it true in the four team double elimination that is played in a few formats. But in most formats it is true, and playing one fewer game can be very significant.
3) If you lose the first game, there is a non-trivial chance that you will be upset in a subsequent elimination game and not even reach the game to go. This is a variation on the first reason I cited.
4) Conversely, if you win the first game, you have at least some chance to win later games (perhaps in an upset), and therefore qualify without even having to play in the final game to go.
5) Even if you feel your chances of winning finals or losing backdoor games are trivially small, winning the first contest gives you the opportunity to rest your starters and save yourself for the game to go. This amounts to cashing in the loss you saved by winning the first game. If you really feel that first game is meaningless, let the other team be the one to tank.
"statistical argument – 1-1 proves nothing except ambiguity"
This argument at first appears quite strong, but upon further examination it is, in my opinion, something of a mirage.
Consider a final game to go that pits two teams against each other that had played earlier in the event, with the team from the "upper game" winning the first contest. This game can have one of two outcomes:
- the team from the upper game wins, and advances with a 2-0 record against the team from the lower game.
- the team from the lower game wins, and advances with a 1-1 record against the team from the upper game.
From the perspective of this "ambiguity" argument, the first scenario is irrelevant, so let's consider only the second case.
Imagine for a moment that you have an oracle with you, and you know that the lower team will win, even before the game starts. You have the choice of either allowing the game to go on or cancelling it, on account of it being a rematch. Should you cancel it? The first instinctive answer might be yes, because you want to avoid the ambiguity.
But are you really avoiding it? No, it already exists. You know, after all, that the lower team would win. No, what you are doing is avoiding the APPEARANCE of ambiguity. The ambiguity is there whether you like it or not.
Like I said, the ambiguity argument is, to me, a mirage. Making the game no-rematch conditional is not preventing ambiguity; it is merely hiding it by protecting the 1-0 team.
A slight variant of this argument is the argument that "the rematch does not consistently help us find the better team". Again, the 2-0 upper game winner case is irrelevant. So the question really becomes: who is more likely to be the better/more deserving team – the team that won the first game, or the team that won the second game? Here I don't have an answer, but my suspicion would be the team that won the second game. I have two reasons for this suspicion:
1) Everyone knows the format and knows that the second game is the one that truly settles your fate. While I reject the idea that the first game is meaningless, the stakes are even higher in the second game, and I would place a lot of emphasis on that high stakes contest.
2) Most people who I have discussed it with agree that nationals is about the most physically gruelling tournament you will ever play in. The winner of the second game is probably the team in better shape and/or or that relies on a deeper rotation, and that team is more likely to represent their region well at nationals.
"letdown/momentum/revenge argument – the team coming into the final game on a win has an edge"
I don't want to dismiss this argument out of hand. I am definitely a believer in sports psychology. I would simply state that I don't believe this mental edge argument to be consistently true, moreover I don't think it is true consistently enough to base policy on. All of these factors (upper game loser being deflated from a tight loss, lower game winner being confident and fresh, lower game winner motivated for revenge while upper game loser is anxious or complacent) are possible. But it could just as easily be the opposite factors in play (upper game loser well rested from a quick loss, lower game winner spent from a nailbiter, upper game loser confident but not cocky while the lower game winner is in their own heads and thinks they have to play past their level to win).
Which of these factors is dominant? I don't know. You don't know, either. Nobody knows. I would like to do a study of every rematch game-to-go that's ever happened, see the statistics on the upsets, and see how the teams did at the next level. I'd like to be able to draw some conclusions on the average physical/psychological state of the teams in the rematch games, but the data is just not there.
This is actually a good idea for a UPA grant in my opinion. Somebody should get a grant, go through every UPA event that has ever been played, and try to compile the available results into a database. Obviously the results would get increasingly sparse the further back you go, but the old newsletters do have much of the results from nationals at least, going back into the 70s. It would be really cool to be able to just pull up 1990 club nationals on the score reporter.
Until that happens, though, I consider this argument speculative at best and inaccurate at worst.
---
OK, that's all, for now. I consider this a work in progress.
