personal stats data analysis

[Rothlover]

Yeah, I've done that against a few teams at CO and ACF nats. My success rate was about 50%, which a lot better solo than when you have teammates (though most of the time, this was with science-weak teams against a science-strong one.)

Can this stuff combined with the subject by subject and round by round breakdowns of ACF nats 03 begin to give some insight into how individuals and teams play/played subject strong individuals/teams?

[setht]

Matt: yup, people do use that strategy. actually, there are two different circumstances: one is the obvious one -- where you don't know the subject and your opponent does; the other is where you do know the subject but have been playing qb long enough to know that there's only a couple possible answers (if that)....such as Andrew's Yamamato example. another one would be any question that mentions the "Russian fleet" -- one should always immediately buzz with Tsushima Strait. of course, the Russian fleet was also at the Battle of Copenhagen -- probably by far the most important naval battle to not show up in qb as a tossup (at least, I've never heard one)....but, no one writes that tossup.

Jeff Hoppes wrote a fine Battle of Copenhagen tossup for BLaST 2004. It does not mention the Russian fleet.

In other news, I think people are hoping for too much from statistics. I think there are too many stupid little factors to allow a very meaningful statistical analysis of someone's performance at a tournament--how are you going to account for food coma after lunch (which hits some people more than others, doesn't always hit at the same time, etc.), people playing poorly due to lack of sleep, etc.? How are you going to account for simultaneous variations in packet and opposing team? If each team played the same round over and over again against every other team, you might be able to say something interesting; or if each team played just one other team on many, many rounds, you'd get useful data. But most teams play most other teams just once in the course of a decent-sized tournament, and most tournament sets are not very uniform in quality, difficulty, or distribution.

-Seth

[Rothlover]

Yeah, I certainly agree Seth, but I still would like to see what you can get by playing around with the numbers. Even taking into account the numerous intangibles and absurdly small sample, there should still prove to be something interesting in there, if it is taken with multiple grains of salt.

[Stained Diviner]

Statistics can still find trends. Because of all the variable you mention, the correlation may be weak, but some trends should still exist.

When numbers work out perfectly, statistics are a waste of time. It is only when things get a little messy that statistics tell you anything. Scientists are able to conduct medical research even though no two patients are exactly alike.

[samer]

First, the question. Can somebody explain to a layman what these numbers mean? Does a "slope" of -60 mean you had an absolutely terrible tournament? Should one expect a number of players to have a positive slope? Long arrays of equations mean almost nothing to us humanities people, or at least to this humanities person.

An interesting thread. I must admit that when I was a stats wonk, the thought of this sort of linear regression never occurred to me.

I can certainly say, though, that perception is an issue here: a statistic that implies "you suck" is not going to be very popular. A case in point here was PC/G (points created/game), which was one of the starting points for PATH. PC/G's major flaw, from a perceptual standpoint, was that it compared players to a hypothetical average player, so a significant number of players (in fact, the majority of players at some tournaments) had negative PC/G values.

[ezubaric]

so a significant number of players (in fact, the majority of players at some tournaments) had negative PC/G values.

But not the Lake Woebegone HS team!