Sunday, August 14, 2005


The Turnover Compact

The NBA All-Star game is all flashy plays and no D. Both teams
agree upon this arrangement and compete within these parameters.
The winning team is the one which does this best. It's fun to watch,
but the regular season is different: if you take wanton shots, your
opponent will punish you with precision offense. It might not *seem*
flashy... well, until you realize that title rings bling. Ask Tim Duncan
or Larry Bird.

In most of ultimate, a similar agreement prevails. It's called
The Turnover Compact. Both teams agree to make a certain
number of turnovers with exciting strategies and big throws.
Long throws are taken when there is a cut and open space,
rather than when there is a cut, open space, a clear shot past
the mark, a strong enough thrower and an 80%+ chance of success.
As a result, the fans get a kickass huckfest and the team which
does better wins. Good for all, right? In a way, yes! In another
way, we are vulnerable to an attack by a team of Duncans and Birds.

Let me drop some science: the team with fewer turnovers wins.
Do the math. It's true. (Exception: with hard caps decided by one point,
both teams may have an equal number of turns.) So you MUST
minimize your turns if you want to win. (This doesn't mean you must
always take the higher percentage throw. If you need fifty 99%
throws to score, your chances would be better with one 75% huck
to the endzone.)

What about the minions among us who know this and despise
turnovers? We're free to make our own team and prove them wrong,
right? The atheletic field is the perfect free market, and if Big Business
is colluding to sell subpar goods at inflated prices, we can
assemble our own product and beat them on their playing field
(to mix, then unmix a few metaphors), right? Well, not exactly.
It's hard to attract players to an unflashy idea, and since most
teams play without a coach, unpopular ideas may not have
an authority to promote them. All players need to buy in, but
there's no money in advertising to promote your idea, and the
governing party ain't listening -- so the status quo remains, and
The Turnover Compact cartel inhibits prudent playing (to mix a few

Here's the thing: this secret alliance is a secret even to its members!
People don't realize how inefficiently they play. Or what they
"know" on the sideline, they simply don't do on the field (see
previous post).

Tim, Larry, we need you!

That's why Godiva was so good--they tried to stay out of the Turnover Compact and emphasized low risk throws in order to take care of the disc. Of course, that type of offense has been called boring...

Which brings me to the question, do teams owe it to anyone to play a certain type of ultimate? Is winning enough in itself even if the strategy isn't exciting or interesting to any one else? In Sarasota in 2003 it was an effective strategy to huck and zone, but resulted in games that weren't necessarily of the highest asthetic quality. Does that matter?
Yes, 'cause it's not fair for fans to pay $50 a set just for the privilege of buying warm $6 beers and watching windy swillfests.

Oh, wait a second...

you being the mathematician...I would like to see how you explain this comment...

"(This doesn't mean you must
always take the higher percentage throw. If you need fifty 99%
throws to score, your chances would be better with one 75% huck
to the endzone.)"

Why, exactly would your chances be better with one 75% huck.

Are you basing it upon the notion that after each pass, the probability of a turnover increases?

I would think that is an errant assumption as each throw should be viewed as a seperate event, correct?

just asking... $

Each throw has a 99% of completion. The only way you don't turn it over is if every one of them is complete. (.99)^50 = 60.5%. So you only have a sixty percent chance of completing 50 99% passes in a row. That's less than 75%, so the huck is the better strategy.


The only people you owe anything to strategy-wise are your own teammates. If you're, say, coaching a new high school team, and you have one or two guy who can huck the disc far (but not accurately), the optimal strategy in the short-term at the high school level is probably huck and Z. But you might want to better prepare the players for success at the next level, by teaching them to play good high-percentage offense.
one thing that always makes this argument difficult is that you don't KNOW how many passes it's going to take to score. here it works since you know that 50 99% passes score and 1 75% pass scores. In practice, you throw that 75% pass you look bad 1/4 of the time (and probably get flak). The turnover on a 99% pass is bad, but much easier to look off (it's 99%, obviously the right decision).

In practice I think it's very difficult to find when that riskier pass is needed and advantageous (heavy wind notwithstanding). People probably do take too many chances, and when the other team is making lots of mistakes (or playing normally?) they do subconsciously enter this turnover compact. In that case, is the question how to you avoid imitating your opponent?

to Gwen's comment: Godiva is a possible answer to this question. They have (had?) an extremely strict set of rules. No matter what happens, they do the same thing. Many teams (even very good teams) don't have near that level of identity, so when things might spiral a little out of control they lose track of what they're trying to do.

Another good example of this compact is 2002 open finals: Furious comes off a super clean semifinal against DoG to play Ring, who didn't play quite as clean an offense. Furious is initially sloppier than usual, but at about 7-6 or so they finally clean it up, and next time they turn the disc it's 12-6 or 12-7.
Mark makes a good point that the responsibility
lies partially on the thrower and partially on
the teams. That's why I mentioned "exciting
strategies" as being part of the Turnover Contract.
If a team's offense is designed as 1-2-3-huck,
then the team's captain shares the blame for some
of those low-percentage bombs.

On the flip side, you may not know if your 99%
throw is one of 5 or one of 50 to score. On the
other hand, you should. E.g., if your team plods
against zones with countless swings and
never progesses, then turns it over with
an unforced error, don't simply blame the
thrower. His/her error may have been a
statistical certainty. Blame the captain, too!
In this case, SOMEONE has to step up and tell the
team to go through or over the cup -- or simply
do so.

Let me stress that by and large the Turnover
Compact involves mainly overaggressive rather than
overconservative play (college is a bit different.)
I thought the main reason that Godiva was so good was that they had so many great players like Molly and Teens. They were lucky, in a sense, that none of their opponents fronted them wildly and adopted a slightly less huck-happy offense themselves. I think they were far from optimal (too conservative on both sides) in their choices.

Re: owing it to just your teammates. Yes, but also consider that "maximizing winning" isn't the only thing to consider, not just for a new high-school team but for every level. Distribution of playing time and roles, willingness to foul or play spiritedly, required training, seriousness, etc. That's why some teams break apart, because the basic goals of the leadership aren't aligned with that of the players.

One thing you can do to try to break the TC is to focus on the process rather than the outcome when commenting on choices. Although it may seem harsh, don't compliment a great huck if that pass would have been incomplete with anything but a great huck. Compliment or condone a good choice that was incomplete. Ask why a good-choice throw was not made.

Zas, it's "his" for offense and "her" for defense, remember?
the first year i coached my hs team it was bombs away all of the brief season...

in year 2, we played short field games, and man to man until right before the state games, then spent two weeks working on huck drills, and zones...

aside from finding out at the captains meeting that you had to alternate guy/girl hucks (i might, MIGHT have varied my overall training schedule for 2 of the girls), it was pretty successful... although it helped that on sunday we had this kid nate crabtree (10.62 100) and we would just bomb it to him. He would quickly erase the 15 yard cushion, and then make one of his own. then run around misreading it for a few seconds, and then sky the crap out of the defender.

now hopefully, we've gotten closer to critical mass, and i can cut practices to 3 days / week, and work on specific skills w/ specific kids, and mainly let them play...

so that i can have my feb and march afternoons open to ski or bike.

I'm invariably surprised at how few people have considered the ramifications of playing a game your team can't play. Where did it become important that you get a 'hack' everytime you score, or that every person on the team has to be Jerry Rice (uh, old Jerry Rice) and Joe Montana at the same time? Shaq's not allowed to shoot 3's...

I think it's demeaning to to new players to put them in unwinnable positions (picking up disc on trapped on own endzone... etc.) Practice that stuff in mini games, etc...

I digress. From what, I'm not sure... I should go post on my own site, I just don't have any ultimate thoughts right now...
In a game, evaluating the higher chances of scoring with one 75% huck versus 50 99% passes also depends on an accurate assessment of the likely completion percentage for each pass.

After watching Disc 1, it seemed like the majority of elite club teams were content with taking risks on hucks that ended up looking closer to 50 or 60%. Granted, part of that is probably an aspect of the Turnover Compact (if both teams are taking those risks, it's okay), but it also seems to me that many teams and players didn't have an accurate assessment of what the chances of completion are on each pass. I bet many of those throwers putting up floaty hucks thought they were closer to 75% when they wound up.

How do you develop an accurate perception of the completion percentages of your passes?
To gambler's last question, the only way is
to show him the statistics. He thinks he's
throwing 80% passes but he's only completing
60% of them. He's overestimating himself
and needs to recalibrate his assessment.

Warning: he'll come back with, "Well, I didn't
see the poach on that one, and I usually do."
Or, "Well, I floated that one, but I usually don't."
People don't want to believe the bottom line,
or that statistics work even though *every*
case is a special case.
i like...
my throws are 100%.
your catches are 40%
I've never like the TC theory, because I feel like it invokes a mutualism that is non-competitive.

It makes more sense to me that, if one team is turning the disc over often, that the other team now realizes that it's own strategy can be optimized by ensuring that they never give their opponent a short field to work with. If your opponent can only go 20 yards at a time, you better make sure that your turnovers are more than 20 yards from the endzone you are defending.

If the TC were really true, we should see a random distribution of turnovers around the field. Do we see silly turnovers on difficult dumps and swings? In my estimation: no, we don't. What we actually see is many long, hucking turnovers. This is a change in strategy in which yardage has become relatively more valuable.

Team A proves they will not score often, especially given a long field. Team B, *recognizing* this, acts to maximize the distance of turnovers as a secondary goal. This decreases Team B's scoring rate, creating a situation where Team A can now optimize their strategy with a similar adjustment.

Now you have lots of hucking turnovers, without invoking mutualism. When teams score infrequently (through bad O or good D) it is relatively advantageous to insure yourself against short turnovers.

*recognize*, likely, does not mean "recognize consciously in the moment". it probably means "respond in a similar manner to how you responded and were then successful in the past".
Two thoughts on "The Turnover Compact"

1 - I think that teams are aware of the math, and that fewer turnovers than your opponent = more goals than your opponent. However, I think the number of turnovers in a game speaks more to the overall skill-level of the players on each team than to some underlying philosophy of what is better/safer/etc.
Take the Club Open finals this year. 13 turnovers total. That's pretty low. Yet the teams employ aggressive offensive strategies involving plenty of "risky" throws. How can they do this with so few turns relative to other teams? They have the best throwers and receivers. Would they have fewer turnovers (and thus win more games) with a less aggressive strategy? Well, it's hard to imagine too many fewer turns. And would less talented teams have fewer turnovers if they decided to work the disc up the field? I'm guessing it wouldn't be a lot different, b/c with a lower overall skill-level, it is more likely that those teams will make a mistake in their attempt to be conservative.

2 - The idea of comparing percentages for success by type of pass involves more than just figuring out the chances of scoring a goal in that possession. On any particular point, the object is not necessarily to score a goal in a single possession, but to score a goal before the other team scores a goal _during_that_point_.
Stringing together a bunch of 99% passes will at some point yield a turnover, and it is more likely that the turnover will occur closer to your own end zone than the turnovers that occur with the 75% huck. Playing defense near the end zone you are trying to score in increases your chances of capitalizing on errors by the other team with a goal.

Not that my own style of play isn't relatively conservative...mostly shorter, higher-percentage passes. But that has more to do with working with your strengths...which is exactly what teams should do with their strategy. If their strength is the long bomb (based on throwers and receivers) and defense, then the wide open strategy, even with more turnovers, might make sense for them.
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?