The top 8 teams will be....(2004)

[Joe Johnson]

This is a good question. It got me to thinking. As a result I did a quickie simulation of a tournament to answer the following question:

What effect does random Alliance Partners and Opponents have on the record of an individual team?

The simulation assumes that the number of teams is large with respect to the number of seeding matches (i.e. I don't have to worry about the affects of teams playing against eachother).

It also neglects ties.

I basically broke teams up into quartiles in terms of capabilities. I drew random teams out of a hat, determined which alliance won based on which quartiles the teams on the alliances were from. From this, I was able to estimate what percentage of matches teams from each quartile would win*.

ANYWAY, I get these results:
1st quartile teams win about 23% of their matches
2nd quartile teams win about 41% of their matches
3rd quartile teams win about 59% of their matches
4th quartile teams win about 77% of their matches

Again, my simulation is very rough, does not take into account a lot of stuff and is intended mostly to account for the effects of random alliance partners.

The implications of the above data is that with 80 team per division in Atlanta, and assuming 8 seeding rounds, there should be between 2 and 3 undefeated teams and between 7 and 8 teams with only 1 loss.

So... I make the following predictions for Atlanta:

1. having no losses is a lock on a drafting position
2. having one loss will put you on the bubble
3. having 2 losses will more or less seal your fate as a draftee not a drafter.

That's my story and I am sticking to it.

Joe J.

* This is how my model determined which alliance pair would win the match:

The model gives each team in a match a point total associated with the quartile they are in (0-25%tile teams get 1 pt, 25-50%tile teams get 2 pts, 51-75%tile teams get 3 pts, 76-99%tile teams get 4 pts). An alliances score is the sum of points in their alliance (a 52%tile team and 76%tile team would "score" 7 points). In the event of a tie score, an alliance wins half of the time and loses half of the time.

This is not a perfect win/loss scheme, but if you figure out what teams win which matches, it roughly seems to make sense to me.