Wednesday, October 10, 2007

Open Seeds

Experience has taught us that the single best indicator for success at nationals is the results of the previous year’s nationals. This isn’t a new idea; we’ve known this for as long as I’ve been playing. What’s interesting is that the previous year’s results have become and increasingly smaller factor in determining seeding. I think there are a few reasons for this: 1) the format gurus (ok, mostly just Tarr) have convinced us that seeds aren’t meant to be predictive 2) we now have access to a lot more regular season results 3) we have been influenced by the method for seeding college – where the amount of turnover makes the previous years results a lot less indicative of success.

Another factor which seems to be completely ignored is when teams bring less that full-strength squad to tournaments. From the standpoint of creating an algorithm, it’s not clear how you account for this. However, in ignoring the strength of the team that actually showed up to a tournament we are throwing away usual information.

I guess what I’m suggesting is that by not considering these two factors, our seeding is not as good as it could be—Chain has been in a pool with 2 semi-finalists the last 3 years. I will say the format at Club nationals is the best at minimizing the effects of initial seeding, but I think it’s naïve to claim that seeding doesn’t matter at nationals.

For these and other reasons, I’ve argued before that it might be time for the UPA to get rid of the prohibition on not seeding a team above a team it finished lower than at a series event.

That being said, here’s my shot at seeding:

The Top 3

Due to the above mentioned prohibition, there are 3 possible ways to seed the top 3. 1)Jam, Sockeye, Bravo. 2)Bravo, Jam, Sockeye, 3) Jam, Bravo, Sockeye. The third way makes the least sense to me. If you’re willing to say that Bravo’s season entitles them to a higher ranking than Sockeye, then certainly Bravo’s 3-1 record against Jam should entitle than to the 1 seed. That being said, I prefer the first seeding. I think you have to give credit to Jam for winning the hardest region.

Furious – the team has been in at least the semis for 8? years. Was strong at ECC before losing to Jam.

Sub Zero, If Goat wins NE regionals they are the obvious 5 seed, I think there loss pushes Zero up to 5. They’ve had a good season, only losing to teams above them with the exception of 1 loss to Goat.
6)Boston, 7)Goat, 8)Ring, 9)Condors. I honestly have no idea how to seed this group. Ring is 2-0 against Boston and 0-2 against Goat. Condors has no head-to-head games with this group. I’m gonna go with this seeding to avoid regional re-matches, but I think any shuffling within this group can be justified.

The three teams in this grouping are Rhino, Doublewide, and Truckstop. Rhino has a four point win over Doublewide, and Truckstop has a one point win over Rhino. I’m going to going to give credit to Rhino for making quarters last year and generally having a slightly better season than Truckstop, despite the 1 point Truckstop win. 10)Rhino, 11)Doublewide, 12)Truckstop.

It’s kind of strange for me to say that Chain should be the 13 seed. This is the sixth year in a row Chain has been to nationals, and this will be the lowest we’ve ever been seeded. This, despite the fact they we’re coming off our best year ever, and added several strong players.
14) Machine – featuring a bunch of guys who wish they still played for Chain, and some new guy from the west coast.
15) Pike – the comeback kids.
16) Van Buren Boys – we were all a little nervous about the poor spirit in the mixed division spilling over into open when these guys decided to make the switch. After regionals, it seems like our fears were justified.

That gives me the following (obviously correct) seeds:
1. Jam
2. Sockeye
3. Bravo
4. Furious
5. Sub Zero
6. Boston
7. Goat
8. Ring
9. Condors
10. Rhino
11. Doublewide
12. Truck Stop
13. Chain
14. Machine
15. Pike
16. Van Buren Boys