[a4u2fear]

I recently developed an NHL system, that has been a discussion in handicapper think tank and players talk in a couple of threads.

I created a power ranking for each team on a moving avg. Comparing the difference between each team in a matchup every night, I found that if the difference was > X or less < Y, bet the team in favor of the difference, as that produced the highest amount of units won. That was obviously after much testing of the numbers in the ranges to find the best fit.

Well, I had a few struggles this season with the system. It was likely due to a bunch of factors - poor SO record this year, maybe just a down year as well.

[a4u2fear]

So I went back with a new thought. Look at the whole range of differences between each matchup, and find out what the win pct was for each difference:

(Home minus Away) <= -1, Home Win Pct = 36%
-1<(H-A)<=-.75, H = 40%
-.75<(H-A)<=-.5, H = 42%
-.5<(H-A)<=-.25, H = 47%
-.25<(H-A)<0, H=48%
(H-A)=0, H=50%

0<(H-A)<=.25, H=57%
.25<(H-A)<=.5, H=60%
.5<(H-A)<=.75, H=64%
.75<(H-A)<=1, H=71%
(H-A)>1, H=76%

This is very good. My power ranking is very encouraging. When it's very likely to bet the away team, they win, when it's very likely to bet the home team, they win, and when they're equal, well, its a tie.

So job is to then compare each nights ML (no vig) to the difference of my power ranking.

[a4u2fear]

Upon comparing the no vig breakeven moneyline to the difference power ranking, the profitable areas were:

(H-A)=0, out of 673 games, we played the away team 438 times for +35 units, the home team for 147 games and lost -16 units, and 88 times the ML was out of range (or so close it was not played)

Anywhere (H-A)<0, the away team was profitable, the home team was not.

Anywhere (H-A)>.25, the home team was profitable, the away team was not.

My guess is like most, the home team is, at least in hockey, more likely favored than the away team, and bettors normally like to bet on the home team; which is why the range of profitable bets for the away team is larger than those of the home team.

[a4u2fear]

Can anyone provide any thoughts on my results?

Has anyone come across anything similar? Should I make my ranges smaller/larger?

[a4u2fear]

What I like most about this, is I will be playing a lot more dogs, as the prior system was into mostly favorites.

I have not had time to test this over and over, but tonights plays would have been

PIT -130 (64% win rate)
SJ -106 (60% win rate)
CHI was a no play (57% win rate)

[nash13]

Hi how will the rankings adjust to the performance over the year? In my research I found out that games near the end of the regular season are most out of line in hockey.

[bihon]

Do you get the same yield by deviding the testing perid into several parts or even by shuffling the games?
Say you had 1000 games, do you get roughly the same results for each 250 games?

[a4u2fear]

both of your questions are similar. I use a moving avg so essentially the rankings shouldn't be that far off in beginning vs end. I will test results in increments as it is a good idea and post results later

[ToyotaLC]

I was with you for a long time a4. I quietly bowed out a month and half ago, but still respected the work you were doing. My only observation is that you, we, all of us following your thread, are having a totally different discussion IF the shootouts and OT went our way. I think your analysis is solid, but the OT/SO results are uncharacteristically skewed in the negative. Can you imagine the kudos that you'd be getting if you went 17-4 in SO? Considering most of the bets were on faves, those swings (as you have said yourself) would be huge. Just my thoughts.

[a4u2fear]

I was with you for a long time a4. I quietly bowed out a month and half ago, but still respected the work you were doing. My only observation is that you, we, all of us following your thread, are having a totally different discussion IF the shootouts and OT went our way. I think your analysis is solid, but the OT/SO results are uncharacteristically skewed in the negative. Can you imagine the kudos that you'd be getting if you went 17-4 in SO? Considering most of the bets were on faves, those swings (as you have said yourself) would be huge. Just my thoughts.

Thanks Bro. No surprise out of the 2 games I picked last night in this thread, we lost 1 in OT.

I will be back next year, and I will not disappoint. Look out!

[a4u2fear]

Tonights plays would be:
CAR +110
OTT +120
TB -112
WPG +160
COL +110

Dog delight!

[a4u2fear]

due to further review, Carolina is not a play, misread.

[b_rad_1983]

a4, you do great work. Only some of us know how much system knowledge you have.

what if for this season, you take out all your away plays. How much of an impact will that have?
I had to do that for my betting this year. Problem I found is that theres alot of high juice, rarely get any dog plays.

[IrishLuck777]

by no means can i understand everything you do to get your system to work, but a few random ideas i have might help some. Include juice when factoring in the win % because a small winner at ev will actually be a loser with the juice. Also, maybe considering playing games that are above a given expected roi and not everything with a small edge or using a 1-2 unit scale to get those better games to stick out.

I know you also talked in the begining of only playing puck lines then you switched to money lines. maybe back test everything using puck lines, money lines, and regulation time lines and see if things vary.

I hope things work out and you can bang out some money next season

[Miz]

what is your question exactly?

All the obvious things about backtesting apply. It's like nobody trolls the old posts. You can learn a lot by reading what people were doing from about 2005-2007.

Good luck on your system/model. If you consistently beat the closing line, you are probably on to something. If not, don't toss it in the trash just yet, but be skeptical of it.

[a4u2fear]

Thanks for all suggestions, they are good. I've been going through all thes me tests, good news it's looking very good, it successfully picked buffalo last night and dallas night before, both +200 or more dogs.

Miz- the results above are taken all way back to 2007. The real question is what is the perfect way to select games? Irish gave me another idea by checking estimated ROI and not just profitability

[Slanina]

Could you resend that email with NHL lines. I must've deleted it accidently. Found out when I went to start testing more years. Covers changed and now only goes back 5 years and it's hard to get MLs.

[Rudeboy]