NHL Picks based on Computer Predictions

**killawookie** · 02-19-13, 02:32 PM

Here, I'll save you some money.. it won't work. The most lines are set based off probabilites so you can't go 55% like spreads and break even.

Come on you should know better than to think it's that easy than to follow some "computer" prediction.

**epicdevoid** · 02-19-13, 02:54 PM

Right, I get that. But, the computer predictions are based on probabilities as well. Let me show this another way...

Away Money Line	Home Money Line	Projected ML	ML Pick
Winnipeg +120	Buffalo -140	Winnipeg +179, Buffalo -179	Buffalo
Montreal +135	NY Rangers -155	Montreal +125, NY Rangers -125	Montreal
NY Islanders +100	Ottawa -120	NY Islanders +115, Ottawa -115	No Pick
Toronto +135	Tampa Bay -155	Toronto +107, Tampa Bay -107	Toronto
San Jose +105	St Louis -125	San Jose +120, St Louis -120	No Pick
Detroit -105	Nashville -115	Detroit +132, Nashville -132	Nashville
Vancouver +100	Chicago -120	Vancouver +161, Chicago -161	Chicago
Los Angeles -125	Edmonton +105	Los Angeles +100, Edmonton -100	Edmonton

The win probabilities project to predicted money lines. As an example, Buffalo is computer projected as at an Money line of -179. Vegas is giving you -140, so that would be a good bet. Conversely, Winnipeg is projected at +179, and thus, a Vegas line of +120 just doesn't cut it. So, you would want to bet whenever the computer predictions give you a better ML prediction than Vegas.

Not saying that the strategy will work or not. Trust me, I'm skeptical too, but there could be a little edge.

**scottgodson1985** · 02-19-13, 04:23 PM

good work! i like the chart and agree with almost everything the cpu says

**epicdevoid** · 02-20-13, 09:29 AM

All based on risking $100 for underdogs, and $100 to win for favorites..
Results from Tuesday: 3-3, down $5

ML Pick	Result	$
Buffalo	L	-140
Montreal	W	135
No Pick
Toronto	L	-100
No Pick
Nashville	W	100
Chicago	W	100
Edmonton	L	-100

**PuckIt** · 02-20-13, 09:46 AM

I have a hard time thinking that Philly should be at +284 tonight against the Pens.

**epicdevoid** · 02-20-13, 01:14 PM

I don't know. Is giving Pitt a 74.7% chance of victory too high?

I'll give this another go for today's games (2/20):

Away Money Line	Home Money Line	Projected ML	ML Pick
Philadelphia +155	Pittsburgh -175	Philadelphia +295, Pittsburgh -295	Pittsburgh
St Louis -120	Colorado +100	St Louis -117, Colorado +117	No Pick
Los Angeles -105	Calgary -115	Los Angeles +118, Calgary -118	Calgary

**EVPlus** · 02-20-13, 05:04 PM

Originally posted by epicdevoid

Right, I get that. But, the computer predictions are based on probabilities as well. Let me show this another way...

Away Money Line	Home Money Line	Projected ML	ML Pick
Winnipeg +120	Buffalo -140	Winnipeg +179, Buffalo -179	Buffalo
Montreal +135	NY Rangers -155	Montreal +125, NY Rangers -125	Montreal
NY Islanders +100	Ottawa -120	NY Islanders +115, Ottawa -115	No Pick
Toronto +135	Tampa Bay -155	Toronto +107, Tampa Bay -107	Toronto
San Jose +105	St Louis -125	San Jose +120, St Louis -120	No Pick
Detroit -105	Nashville -115	Detroit +132, Nashville -132	Nashville
Vancouver +100	Chicago -120	Vancouver +161, Chicago -161	Chicago
Los Angeles -125	Edmonton +105	Los Angeles +100, Edmonton -100	Edmonton

The win probabilities project to predicted money lines. As an example, Buffalo is computer projected as at an Money line of -179. Vegas is giving you -140, so that would be a good bet. Conversely, Winnipeg is projected at +179, and thus, a Vegas line of +120 just doesn't cut it. So, you would want to bet whenever the computer predictions give you a better ML prediction than Vegas.

Not saying that the strategy will work or not. Trust me, I'm skeptical too, but there could be a little edge.

Laying -140 is rarely a good bet - ESPECIALLY in the NHL with the high level of parity.

This or any computer program is only as good as those who designed it.

Vegas has these same stats and are more efficient at utilizing them.

Any model that does not factor in situational analysis is a money burner (if not in losses than in juice). For example, I don't care what the numbers say, if team A played a hard-fought bloody OT game at sea level and then goes to Colorado the next day to play team B for an early game, I am not putting anything down on A.

I actually use the aforementioned in the NHL parity to give me an edge. I usually take dogs or faves at -1 (buy half a point on puckline) to give me +100 or better odds. Sure, I use stats but I also look at other factors such as injuries to key players.

**PuckIt** · 02-20-13, 11:32 PM

Originally posted by PuckIt

I have a hard time thinking that Philly should be at +284 tonight against the Pens.

Originally posted by epicdevoid

I don't know. Is giving Pitt a 74.7% chance of victory too high?

I'll give this another go for today's games (2/20):

Away Money Line	Home Money Line	Projected ML	ML Pick
Philadelphia +155	Pittsburgh -175	Philadelphia +295, Pittsburgh -295	Pittsburgh
St Louis -120	Colorado +100	St Louis -117, Colorado +117	No Pick
Los Angeles -105	Calgary -115	Los Angeles +118, Calgary -118	Calgary

Computer's just can't factor a rivalry...

**epicdevoid** · 02-21-13, 03:05 PM

Originally posted by EVPlus

Laying -140 is rarely a good bet - ESPECIALLY in the NHL with the high level of parity.

This or any computer program is only as good as those who designed it.

Vegas has these same stats and are more efficient at utilizing them.

Any model that does not factor in situational analysis is a money burner (if not in losses than in juice). For example, I don't care what the numbers say, if team A played a hard-fought bloody OT game at sea level and then goes to Colorado the next day to play team B for an early game, I am not putting anything down on A.

I actually use the aforementioned in the NHL parity to give me an edge. I usually take dogs or faves at -1 (buy half a point on puckline) to give me +100 or better odds. Sure, I use stats but I also look at other factors such as injuries to key players.

This is great feedback.

For yesterday: 0-2, lost $290

ML Pick	Result	$
Pittsburgh	L	-175
No Pick
Calgary	L	-115

We'll try this out today with 10 games, gives a better sample size...

Away ML	Home ML	Away Money Line	Home Money Line	Projected ML	ML Pick
120	-140	Florida +120	Philadelphia -140	Florida +153, Philadelphia -153	Philadelphia
-115	-105	New Jersey -115	Washington -105	New Jersey -187, Washington +187	New Jersey
125	-145	Winnipeg +125	Carolina -145	Winnipeg +274, Carolina -274	Carolina
100	-120	Buffalo +100	Toronto -120	Buffalo +196, Toronto -196	Toronto
130	-150	NY Islanders +130	Montreal -150	NY Islanders +148, Montreal -148	No Pick
-130	110	NY Rangers -130	Ottawa +110	NY Rangers -103, Ottawa +103	Ottawa
-125	105	Boston -125	Tampa Bay +105	Boston +112, Tampa Bay -112	Tampa Bay
155	-175	Columbus +155	Detroit -175	Columbus +203, Detroit -203	Detroit
-145	125	Vancouver -145	Dallas +125	Vancouver -129, Dallas +129	No Pick
100	-120	Minnesota +100	Edmonton -120	Minnesota +123, Edmonton -123	Edmonton

**face** · 02-21-13, 03:15 PM

philly played hard last night so value might actually be on the panthers, is there any way your model takes that into account? some of the philly guys are older, and i know that after i run 7 miles, the next day i do not feel like running. also bryzgalov supposed to be in net again after very exhausting night last night. we'll see, i could be full of it, but i think panthers win.

**epicdevoid** · 02-21-13, 06:37 PM

Originally posted by face

philly played hard last night so value might actually be on the panthers, is there any way your model takes that into account? some of the philly guys are older, and i know that after i run 7 miles, the next day i do not feel like running. also bryzgalov supposed to be in net again after very exhausting night last night. we'll see, i could be full of it, but i think panthers win.

No, the only thing taken into account is the win/loss, though the scores of previous games, margin of victory, home/away, date of previous games are used to weight the observations. So a 4-1 victory away would be weighted more heavily than a 4-1 victory at home. This is all then thrown into a logistic regression model which is a little too much to discuss on this board.

I like the running analogy. I've never seen any studies or evidence that the next day hangover holds any water, but it is for sure an interesting thought.

**face** · 02-21-13, 07:26 PM

3-0 panthers. goaltending is exhausting, so is defending 7 or 8 powerplays against stacked penguins. plus skating up and down the ice all night. you had to see it, it was quite a game last night. anyway, i banged panthers for +130 x 2

i am not much of a numbers guys so won't troll your thread, but i was an athlete for a long time, and legs need to recover, energy, calories, banging around, fatigue, etc. no studies though so inconclusive for you, just my two cents.

**juice050** · 02-21-13, 08:42 PM

say what you want about computers not being any good but ive been with accuscore for over 2 years and i must admit they finish in the + each and every year. sure a computer cant predict tiredness or rivalrys but in a nut shell it can be used as a very good source to help you make a better decision on what to play on that particular day. just my 2 cents though...

**epicdevoid** · 02-22-13, 12:52 PM

Thursday's results: 3-5, down $370
Overall: 6-10, down $665

Geez, maybe the way to play this is to go against the computers. ha. I'll pick this up again on Saturday.

ML Pick	Result	$
Philadelphia	L	-140
New Jersey	W	100
Carolina	L	-145
Toronto	W	100
No Pick
Ottawa	W	110
Tampa Bay	L	-100
Detroit	L	-175
No Pick
Edmonton	L	-120

**epicdevoid** · 02-24-13, 11:28 AM

Ok, the predictions have been updated from last week's games. Let's see if this thing does as good (or bad) as it did last week:

For Sunday:

Away Money Line	Home Money Line	Projected ML	ML Pick
Boston -165	Florida +145	Boston -158, Florida +158	No Pick
Winnipeg +145	New Jersey -165	Winnipeg +228, New Jersey -228	New Jersey
Vancouver -135	Detroit +115	Vancouver -113, Detroit +113	Detroit
Carolina +100	NY Islanders -120	Carolina +112, NY Islanders -112	No Pick
Columbus +240	Chicago -280	Columbus +402, Chicago -402	Chicago
Tampa Bay +160	Pittsburgh -185	Tampa Bay +130, Pittsburgh -130	Tampa Bay
Phoenix -110	Calgary -110	Phoenix -104, Calgary +104	No Pick
Colorado +155	Anaheim -175	Colorado +261, Anaheim -261	Anaheim

**epicdevoid** · 02-25-13, 11:03 AM

Sunday's results: 3-2, up $50
Overall: 9-12, down $615

ML Pick	Result	$
No Pick
New Jersey	L	-165
Detroit	W	115
No Pick
Chicago	W	100
Tampa Bay	L	-100
No Pick
Anaheim	W	100

Actually a positive day for once...kind of a shocker.

Here is Monday's picks:

Away Money Line	Home Money Line	Projected ML	ML Pick
Toronto +125	Philadelphia -145	Toronto -107, Philadelphia +107	Toronto
Montreal -130	Ottawa +110	Montreal +104, Ottawa -104	Ottawa
Dallas +140	Nashville -160	Dallas +121, Nashville -121	Dallas
Edmonton +160	Chicago -185	Edmonton +456, Chicago -456	Chicago
Anaheim +130	Los Angeles -150	Anaheim -147, Los Angeles +147	Anaheim

**epicdevoid** · 02-26-13, 09:53 AM

2 winning days in a row...look out...ha

Mondays results: 3-2, up $135
Overall: 12-14, down $480

ML Pick	Result	$
Toronto	W	125
Ottawa	W	110
Dallas	L	-100
Chicago	W	100
Anaheim	L	-100

For Tuesday:

Away Money Line	Home Money Line	Projected ML	ML Pick
Carolina +110	Washington -130	Carolina -100, Washington +100	Carolina
Dallas -130	Columbus +110	Dallas -124, Columbus +124	No Pick
Winnipeg +155	NY Rangers -175	Winnipeg +194, NY Rangers -194	NY Rangers
Boston -150	NY Islanders +130	Boston -105, NY Islanders +105	NY Islanders
Buffalo +130	Tampa Bay -150	Buffalo +225, Tampa Bay -225	Tampa Bay
Pittsburgh -160	Florida +140	Pittsburgh -177, Florida +177	Pittsburgh
Calgary +130	Minnesota -150	Calgary +119, Minnesota -119	Calgary
Phoenix +155	Vancouver -175	Phoenix +179, Vancouver -179	Vancouver
Colorado +150	San Jose -170	Colorado +162, San Jose -162	No Pick