I was watching my Edmonton Oilers struggle through another thumping last night in the NHL, and started to wonder what the chance is that they are simply way underperforming. Ignoring overtime losses they are 16-32 this season.

Assuming they are truly a 0.500 team, this would be like having 16 heads come up in 48 flips of a fair coin.

I can use a cumulative binomial calculation to determine the chance of getting 16 or fewer heads in this circumstance. This would answer the question "What is the chance of a 0.500 team having a 16-32 record or worse in the first 48 games" and is around 1.5%

My question is whether there is a better way to formulate this question to evaluate the team. Such as "what is the probability that a team with a given record is actually a 0.500 team or better" and how that would best be evaluated.

Thanks!

I'm not sure why you'd believe that any given team is exactly 50% to win a match. Some teams are better than other teams.

You could do this, but you would need a prior distribution of win rates. For example, if all teams are truly coinflips against each other than the chance of a 16-32 team being a 0.500 team or better is 100%.

Probably the best way to do it is to assume a uniform prior on [0,1] then calculating the posterior distribution as in

http://en.wikipedia.org/wiki/Posterior_probability

section 'calculation'.

Thank you that's a very helpful perspective.

Well I'm from Edmonton and the Oilers this year certainly aren't a .500 team. They have some young talent, but this team is 2 years away from the playoffs. If we can dump Horcoff, and maybe find some young talent for Penner at playoff time, this team can make the playoffs in 2012 and go for the cup in 2013 once they start to gel.

For a true 50/50 coinflip in sports, it would depend on 2 teams that have played 5 or 6 times with 1 goal games each way etc.