Quote:
Originally Posted by mcteague
The argument I was making, seemingly wrong from what I hear, is this.
If in 1000 coin flips streaks of 6 or more will commonly occur 10 times ( randomly picked number) And we assume those streaks are evenly distributed. If we refuse to bet against heads once a streak of 5 has occurred, we can gain the bets when tails does go to 6.
Assuming equal probability of streaks, we should gain 5 units per 1000.
For some mysterious reason, mysterious to me anyway, this is not true
|
It's wrong. There are people who explain stats more concisely than me so I won't try to confuse you but it's wrong.
Was thinking about a baseball bet for the system players that would go like this;
I pick 25 hitters for myself and 50 hitters for you. All must be listed as healthy going into the season and projected to get a full share of plate appearances.
The bet is on batting average and we count the full season for my 25 guys.
For your 50 guys you can use any system that can be generically prescibed to your whole group before the season, does not involve examing the stats of their opponents, and should result in at least (roughly) half a season's worth of plate appearances for each player (the reason you would get 50 to my 25).
Examples of a system would be only couting the at bats on a game day following a game day in which your player did not get a hit, only counting at bats on a game day following a game day in which your player reached base or drove in a run, only counting at bats on Mon/Wed/Fri/Sun, etc.
Do you like your chances? I mean I have to use the whole season you can take advantage of the distribution of streaks.