This is going to sound crazy, but...

Ok, so I have been going through 100s of ways to bet NHl games, estimate scores, etc.

my latest, the results:
Pick every game on the PL after 20 games per team using a strategy to estimate the goal total per game compared with the associated PL, wagering 1u per game, results:
07-08, +112u
08-09, +69u
09-10, +16u
10-11, +120u
11-12, +126u

ive double checked the PL odds, the formulas, the calculations and there are 0 mistakes. I've forgotten about it (on purpose) a few times to revisit it and see if again I missed something or made a mistake, so far, none.

any thoughts? Am I crazy? I'm dumbfounded it's this good. Of course since I can pick the PL, I can also successfully pick the ML

Overfitting?

I'm not sure what you mean.

Originally Posted by flsaders85

Overfitting?

If it worked for 2012-13, 2013-14 without any tweaks, he could be on to something.

Originally Posted by a4u2fear

I'm not sure what you mean.

Test your system out of sample and do not apply to same years you used to develop system.

we're rich!

they all fail eventually

Without knowing the details, no way to say for certain.

"If it seems too good to be true...." really does apply with gambling though. Inevitably, past data will have patterns that appear to be meaningful, but are just natural variance.

Originally Posted by a4u2fear

Ok, so I have been going through 100s of ways to bet NHl games, estimate scores, etc.

my latest, the results:
Pick every game on the PL after 20 games per team using a strategy to estimate the goal total per game compared with the associated PL, wagering 1u per game, results:
07-08, +112u
08-09, +69u
09-10, +16u
10-11, +120u
11-12, +126u

ive double checked the PL odds, the formulas, the calculations and there are 0 mistakes. I've forgotten about it (on purpose) a few times to revisit it and see if again I missed something or made a mistake, so far, none.

any thoughts? Am I crazy? I'm dumbfounded it's this good. Of course since I can pick the PL, I can also successfully pick the ML

Your system picked every game, or just the games that fit a criteria? If you tested against all games, we are talking about 6k games, correct?

The system was not developed using a set of games.

the system picked all games after each team played 20 and did not distinguish between any. The difference in estimates for each team compared to the PL is what swayed my decision to pick one side or the other but always picked for each game

went through the games to find out if I could've gotten here by luck, so I found the avg puck line for my dog puck line bets and my fav puck line bets.

Avg Fav PL = +177
Avg Dog PL = -216

Results, using a binomial calculator, there was a 5% chance I could've just gotten lucky on the Fav PL results, but just a .3% chance I got lucky betting the Dog PL.

Combined, that is the percentage after betting almost 6000 games.

are you willing to share anything on what how you estimate each team compared to the PL?

I'm willing to share the plays, starting next season.

great, thx, waiting for october... did you try to test it in lower leagues or some european

Originally Posted by basket33

great, thx, waiting for october... did you try to test it in lower leagues or some european

NHL only. It also does not have any plays until 300 total NHL games have been played. Which means it will begin around mid-November.