Sunday, July 26, 2009

2009 HU tourney wrap up

Well, that's another heads up tourney in the books. Wanted to get feedback on the format.

Short story is that there was no "game money" this year, blind caps changed, payouts moved from 50/25/12.5/12.5 to 40/30/15/15, and the tie breaker system is new. Also wanted to ask about the duration since the first portion went from 7:30 to 10:30 and brackets finished about 1:50 versus a projected 1:00 finish. For more details see long story below (yeah I know it is squarely in TLDR territory, if you want to just jot your thoughts without reading below I don't blame you).

* No "game money" this year. Last year, every win in the first stage of the tourney would be worth a couple bucks. That ended up depleting the payouts for the top finishers and didn't add up to a substantial amount of money even if you won a lot of games. Based on feedback from last year it was dropped this time around. Turns out that it took a long while for someone to be so far behind that they couldn't qualify even if they won out.

* Changed blind caps for the bracket rounds. It was unfortunate that Marsh didn't make a run into the brackets because he was the one who felt that the blinds got too oppressive at the end of last year's tourney and once the blinds capped you really just couldn't fold. It would have been nice to get his perspective of the changes. The difference was this year, even with the BB at 20% of the average chip stack I did feel that one could fold when the blinds capped as I saw happen on multiple occasions.

=== Disclaimer === serious tournament math geek discussion below ===
* Different preliminary matchups. Last year we had fewer people and just did a true round robin. Since that would have been too many games for this year the tourney used pools of players instead. In a true round robin the player records are about as fair as you can get it. Ties were broken by head to head matchups. This year since not everyone played everyone else I went with the ranking system described below.
- Win/loss percentage
- Strength of schedule (a win over a player who ended up 6-2 counts marginally more than a win over a player who ended up 4-4)
- Secondary strength of schedule (a win over a player who ended up 6-2 with tougher opponents weighs in more than a win over a 6-2 player with victories over weaker opponents)
- Record within your pool
- Strength of your pool (if players are tied after all above tie breakers then a player who had to play two players from a stronger pool would be seeded higher than one from a weaker pool)

A couple of things came up while working on the format. At first I had it set up where everyone played all nine players from outside of their pool but changed that to help prevent tie deadlocks since two players could happen to beat the same players since each pool member played the exact same players as everyone else in the pool. The tie breaker system also ignored head to head records which would have changed things. Jeh, Drew, and Joe all had the same record and Drew beat both Joe and Jeh and was seeded highest of the three. But Joe was seeded below Jeh despite defeating him. I did consider using head to head record but didn't for two reasons 1) in simulations it was too easy for a three way circular tie to happen where A beat B beat C beat A and head to head wouldn't help and 2) I could tell it was going to be a total pain to write a formula for the spreadsheet that did that. As it turns out the difference was only in seeding and did not affect someone being bubbled out because of the ranking.

* The tourney ran late. We actually started pretty darn close to schedule which was great. We had a situation where one player was playing longer matches on average than other players. Not saying that's bad or anything. Just pointing out that despite the flexibility afforded by the format for the prelims we still were gated by one player before starting the bracket round. I was tinkering around with a format like ELO chess rankings which would allow any player to play any amount of games against any other player but it would have to provide incentive for players who went 4-0 to continue to play games at risk of breaking their win percentage yet guard against players having nothing to lose and playing recklessly. In the end I couldn't get a system figured out in time but I think that system has potential since it would let us just set a time limit for the first stage and we would be able to stay closer to schedule.

That's it for the wrap. Let me know if you have any comments on any part of the tourney.