I've been working on a model that has show some very positive results in the back testing I have done but due to lack of time I wasn't able to fully back test it. I'm not sure the best way to run it yet so I'll be tracking a few different scoring options.
I'm not going to get into how the model works so don't ask but it gives me a number usually between +25 and -25. There's 4 ways I'm going to track it until I find out which option is the best.
1st option is bet any 10 point difference between both teams.
Example 1 Bruins +15 vs Leafs +4 (bet Bruins)
Example 2 Bruins +2 vs Leafs -10 (bet Bruins)
2nd option is bet any 20 point difference between both teams.
Example 1 Bruins +25 vs Leafs +1 (bet Bruins)
Example 2 Bruins 3 vs Leafs -25 (bet Bruins)
3rd option is >10 team vs any <0 team or bet against any <-10 team vs any >0 team.
Example 1 Bruins +10 vs Leafs 0 (bet Bruins)
Example 2 Bruins +5 vs Leafs -15 (bet Bruins)
4th option is >10 team vs any <-10 team.
Example 1 Bruins +10 vs Leafs -10 (bet Bruins)
Example 2 Bruins +20 vs Leafs -25 (bet Bruins)
Option 4 should yield the highest win% but it will also have the fewest bets so it would probably be leaving some value out there. Option 2 seems solid and would make a lot of sense. Options 1,3 will have the most bets but not really sure how they're going to work out.
St Louis Blues -166
Columbus Blue Jackets -115
Tampa Bay Lightning -175
So far all picks have come from betting options 1 and 3 which no real surprise. If I was more aggressive in handicapping each team before the season it would probably be a little different but I decided on the passive approach.
Boston Bruins -174
New Jersey Devils -120
New York Islanders -172
Tampa Bay Lightning -172
Montreal Canadians -105
Chicago Blackhawks -279
Vancouver Canucks -175
San Jose Sharks -185
A lot of big favorites so far but that's really no surprise early in the season, the model will picks dogs just none yet.
Chicago the only 20 point dif game so far and they're also the only +10 -10 game as well.
The record that I will post is the 10 point dif record because that will cover all games and I'll update the other records from time to time.
The Canadians are the 1st underdog pick so far although I kind of think the Lightning will win. Boston and Ottawa were very close to being plays but they just missed and Anaheim just missed being a 20 point play.
MNF Pick -San Francisco 49ers -3 (not a system play)
San Jose Sharks -110
Anaheim Ducks -115
Columbus Blue Jackets -115
Nashville Predators -170
LA Kings -200
I've also started using the same model on totals so I'll post those picks as well. I have no idea how the totals will turn out, it seems like it would work in theory but I haven't back tested it. Yesterday was the first total play I didn't post it but it was Boston/Colorado under.
New York/New York o5.5 +110
San Jose/Washington u5.5 -130
Anaheim/Philly o5.5 -105
Buffalo/Carolina o5.5 +120
Montreal Canadians -120
Minnesota/New York u5 -105
Updated Records
Sides
Any 10 point diff (25-21 -$568)
+10 versus any - (16-11 +$45)
Any 20 diff (4-4 -$65)
+10 versus -10 (4-1 +$300)
Totals
Any 10 point total (13-14 -$286)
+10 versus any + (5-8 -$364)
Any 20 point total (1-4 -$383)
+10 versus +10 (1-2 -$155)
Notes
1. So far it's not looking all that impressive but the theory behind it is solid and the limited back testing I did confirmed it so I'm going to keep running this for a bit yet.
2. Underdogs are 1-7 in any 10 point diff category so I think that will turn around.
3. Starting certain teams with +10 or -10 probably a bad idea. Between -5 and +5 may work better but only if I have a very good pre-season read on a team.
4. I also changed how I grade pushed with totals. Shouldn't make a huge difference just made more sense.
5. For some reason I had the New York Rangers winning their first game and losing every game since not sure why and not sure if it effected any bets.
6. The any 10 dif for sides I wasn't sure about from the start I figured this option +10 versus any - (16-11 +$45 would be better and it looks like it is.
7. Wasn't sure about totals in general but I still think it may work.
Overall I'm not impressed so far but it's early in the season and it might just be a variance thing for now so I'll probably do another update at around the game 20 or 25 mark and I should have a better picture of whats going on.
No picks today but I think I figured out why my model was performing so much worse then what I back tested.
When I back tested all I did was use the number for a specific team and not combine it with who they were playing against. All I did was play +10 or better or fade -10 or worse but when I start running it for real I combined the numbers for both teams and even though that sounded better apparently it doesn't work better.
So now I'm just going to bet teams that are +10 and fade teams that are -10 and not worry about who they are playing. Or bet top 5 and fade the bottom 5. I'll probably track both because I think +10 -10 might be better for totals but top 5 bottom 5 better for sides.
Yesterday
Top 5 bottom 5 went 7-5 +$148
+10 and -10 went 10-4
Ok fellas I readjusted the model. Its the same concept I just made it more aggressive and eliminated combining home/away.
I also changed how I'm going to play it. I have it setup as rankings now and I'm going with top 3 and bottom 3 which would be close to 500 picks a year. Add in totals and that will be close to 1000 picks a year. Also if a top team is playing a bottom team it will be a 2 unit play.
Todays picks
Montreal Canadians +125
New York Islanders +115
Anaheim Ducks +102
Arizona Coyotes +105
Philadelphia and Tampa Bay over 6 -102
Chicago and Ottawa under 5.5 -114
NY Islanders and Colorado over 5.5 -110
Montreal and Vancouver over 5 -135