Testing the Tanking Theory

Amidst all the media coverage of Tony Liberatore’s claims about Carlton ‘tanking’, it’s surprising that no-one in the media seems to have quoted from our own expert on this, Melbourne University Professor Jeff Borland. Here’s his summary, from a 2006 paper (gated, sorry):

research on the AFL National Draft appears to suggest that adverse incentive effects are minimal. Borland, Chicu and Macdonald (2006) compare team performance in the draft and pre-draft eras (1968-1985 against 1986-2005) to test whether either the reverse-order draft or priority pick system has affected the performance of teams which are unable to make the finals or which are eligible for a priority choice. Their strong finding is that for teams unable to make the finals, or eligible for a priority selection, there is no evidence of a negative effect on the probability of winning matches. It is speculated that the difference between findings for Australian Rules football and the NBA may be due to several factors. One factor is the more muted incentives to obtain high-order draft selections in Australian Rules football due to larger team size, difficulties in assessing player talent in the early years of the AFL National Draft, and the more extended upper tail of the distribution of talent in the United States with a larger population.

With regard to the effects of priority selections, the main explanation would appear to be the relatively small number of teams that have ever been eligible for priority pick selections in the AFL National Draft compared to the larger number of teams eligible for top selections in the NBA Draft so that there is just much less scope for teams in the AFL to ‘lose to teams they should have beaten’.

This entry was posted in Sport. Bookmark the permalink.

8 Responses to Testing the Tanking Theory

  1. cba says:

    I’d agree that most seasons there’s little incentive to drop games, but in this case the incentives were a lot stronger than usual. Chris Judd is a proven champion that a club can build a team around.

  2. Mark Picton says:

    Jeff’s point as I saw it was that rubbish teams had little opportunity to tank because they just didn’t have many winning chances to throw. And that’s a compelling argument for why you wouldn’t expect to see much tanking.

    But of course it’s no argument at all as to whether _if_ there were an opportunity to tank there would be tanking. A team with an incentive to tank playing its last eight games against top eight teams or those with a good chance at the time could not be said to be tanking. But a team with that incentive that played mainly other rubbish teams might.

    For what it’s worth, Carlton’s last eight opponents last year (to all of which they lost, ladder positions in brackets):

    Sydney [7]
    Brisbane [10]
    St Kilda [9]
    Collingwood [6]
    Pt Adelaide [2]
    Geelong [1]
    Kangaroos [4]
    Melbourne [14]

    There were no games where Carlton had an interstate home ground advantage.

    Of these, only the Melbourne game could really be said to be one where Carlton had a real shake.

  3. Yobbo says:

    There’s a lot being said about deliberately losing by playing badly. That’s one thing. But there’s another way to lose, and that’t to not put your best team out on the ground, instead playing untested rookies and telling em “go get ’em tiger!”.

    You are still going to lose.

    Carlton have already admitted this second ploy:

    Carlton chief executive Greg Swann has dismissed the claims, saying the club was merely giving game time to its younger players when it lost its last 11 matches last year.

    “giving game time to your younger players” just because they might like a run is something that doesn’t happen when you are trying to win. Even if veterans are slightly injured and need a break, they play when you are trying to win games, like in finals.

  4. haiku says:

    You could seek to avoid having such strong incentives for finishing last by adding a slightly random element to the draft order.

    After the season has ended, put the numbers 12, 13, 14 in a hat and draw one out. Say it comes out as number 13. This is where you split up the bottom eight.

    You then have a second draw, with numbers 13, 14, 15, 16. The order that these are drawn out in is the order of draft picks. So finishing last might only get you pick number four.

    The reason you have the first draw is to make sure you’re not switching the incentives for finishing last into similar incentives to finish 13th or the like.

  5. Pingback: Club Troppo » Missing Link Daily

  6. CBA is right to mention Judd. Carlton didn’t tank to get Kreuzer, they tanked to get Judd who, from about round 15, they knew was coming to Victoria. They would not have got the best player in the competition had they not had the extra collateral of pick two and the priority pick instead of just pick three.

  7. Rob Macdonald says:

    Interesting thoughts everyone.
    (btw, I’m the Macdonald in ‘Borland, Chicu & Macdonald’;).

    Some things to consider for future research in this field are:

    (a) the few studies of tanking to date have concentrated upon match results (ie outputs of the sporting production process).

    (b) No-one to date has looked at inputs to the production process (ie team selection decisions, in-game coaching decisions).

    (c) It’s possible to look at team selection decisions, but how do you evaluate player quality? Champion Data Pty Ltd have their proprietary player rankings, but I’m yet to see anything that is properly validated via the academic ‘conventional means’ of peer-reviewed publication.

    (d) The question then is to design a mechanism to eliminate the incentive to tank. I like Haiku’s idea of a “double lottery” – haven’t heard that one before!

    (e) To Mark Picton: as an exercise in methematical economics, it would be an interesting exercise to formally model the incentive to tank under different conditions. As Tony the Teacher points out, that model would have to include the ‘trade-value’ of a draft as well as the ‘use-value’ (where the draft pick is used rather than traded).

    In many ways, season 2007 was just about a ‘perfect storm’. Both Carlton and Melbourne had an incentive to lose in Round 22, a No. 1 Draft pick was up for grabs for Carlton and a No. 2 (I think) pick up for grabs for Melbourne. Judd was obviously swayed by what carlton could put onthe table, but as a kid he barracked for Melbourne!

  8. Andrew Leigh says:

    Rob, do you have a ‘for citation’ version of your paper on the web somewhere? I could only find a ‘not for citation’ version online, so didn’t link to it.

Comments are closed.