Point spread difference, how do I calculate my edge???

[slope99]

I usually bet on NFL parlays, anywhere from 3 to 5 games.

My strategy has always been to pick games where my spread is furthest away from Vegas spread which I consider to be correct spread. I know this helps my edge but by how much?

Let's use a 3 team parlay as an example:

I pick MIAMI +3.5 when Vegas +2.5

I pick INDY +8.5 when Vegas +6.5

And finally, I pick PHILLY +5.5 when Vegas +4

(My 3 team parlay odds are x7, or 12.5% house advantage or "per game odds" of 1.91)

I am guessing I will beat house edge with these picks... But by how much?? How do I calculate? Is there a way?

Thank you so much. Gearing up for next NFL season lol.

-Alex

[mathdotcom]

Very simple answer:

If Miami +2.5 = true odds then Miami +2.5 has 50% chance of cashing
Miami +3.5 = odds of winning = 50% + odds Miami loses by 3

Suppose odds of Miami losing by 3 are 10%, then you expect to win that leg of your parlay 60% of the time. Do this again for the other bets. If you have a 60% chance of winning the first leg, 65% chance of winning 2nd leg, and 70% chance of winning 3rd leg then your odds of winning the parlay are 0.6*0.65*0.7 = 0.273. If your parlay is getting odds of at least ~+270 then you're good to go.

Not so simple answer:
Is Miami +2.5 exactly a 50% proposition? What if it's Miami +2.5 +109?
What if your bookie is giving you +3 and not +3.5?
Doing the above with a very accurate estimate of Miami losing by 3
Are all your bets independent? [In this example they are]

[FourLengthsClear]

Futher to Mathy's post above, this should help you.

http://www.sbrforum.com/betting-tools/half-point-calculator/

Note though that the data used to derive the 'push probability' figures is out of date and takes no account of the projected total.

[AlwaysDrawing]

If your spread is very far away from Vegas's, the problem is with your model.

[slope99]

Very simple answer:

If Miami +2.5 = true odds then Miami +2.5 has 50% chance of cashing
Miami +3.5 = odds of winning = 50% + odds Miami loses by 3

Suppose odds of Miami losing by 3 are 10%, then you expect to win that leg of your parlay 60% of the time. Do this again for the other bets. If you have a 60% chance of winning the first leg, 65% chance of winning 2nd leg, and 70% chance of winning 3rd leg then your odds of winning the parlay are 0.6*0.65*0.7 = 0.273. If your parlay is getting odds of at least ~+270 then you're good to go.

100% agree with your reasoning but...

Very simple answer:
Doing the above with a very accurate estimate of Miami losing by 3

So I'm guessing 10% odds of Miami losing by 3 not accurate? How can we determine odds of Miami losing by 3 with more accuracy?

Very simple answer:

Is Miami +2.5 exactly a 50% proposition? What if it's Miami +2.5 +109?

Good point. So we should include error of "true odds" in the equation.

Which brings me to another question. What spread do you consider closest to true odds? MGM-Mirage? Stations? An average of all Vegas casinos? Other?

Thanks

[slope99]

If your spread is very far away from Vegas's, the problem is with your model.

It's my bookie's spread. I frequently see 1 point difference, even 2 points for big spreads. Always favoring the underdog (making the favorites even bigger favorites).

[Duff85]

It's my bookie's spread. I frequently see 1 point difference, even 2 points for big spreads. Always favoring the underdog (making the favorites even bigger favorites).

Bet dogs, print moneyz, buy an island (or get stiffed by bookie) whichever comes first.

[slope99]

Looks like I have alot more to learn.

[LT Profits]

If you are getting an extra point on single-digit spreads and an extra two points on double-digit spreads, you will have an easy life (assuming he pays) because there is no need to handicap games.

[LT Profits]

I should add the closer to game time the better.

[jolmscheid]

Hey LT..I understand finding an edge for point spreads, but what about moneylines?

For instance if a game is listed at -140 and my projections show that it should be -185, how do I calculate my edge from this?

[uva3021]

185/285 - 140/240

[LT Profits]

Yeah MLs are easier because no ATS/ML conversion involved

[jolmscheid]

Thanks uva and LT...and is it the same calculation for moneyline underdogs too?

[subs]

Futher to Mathy's post above, this should help you.

http://www.sbrforum.com/betting-tool...nt-calculator/

Note though that the data used to derive the 'push probability' figures is out of date and takes no account of the projected total.

100% agree with your reasoning but...



So I'm guessing 10% odds of Miami losing by 3 not accurate? How can we determine odds of Miami losing by 3 with more accuracy?

Futher to Mathy's post above, this should help you.

http://www.sbrforum.com/betting-tool...nt-calculator/

Note though that the data used to derive the 'push probability' figures is out of date and takes no account of the projected total.

or get some good data and do it urself. not too hard

Good point. So we should include error of "true odds" in the equation.

Which brings me to another question. What spread do you consider closest to true odds? MGM-Mirage? Stations? An average of all Vegas casinos? Other?

Pinny 1st, u need the no-vig odds tho.

any1 who shades his lines so badly will be aware that they can be beaten v easily so GL with lasting more than a few weeks or getting paid. a little disguise may be a good idea... maybe add some Fav MLs (if not too bad) to ur parlays?

GL

[LT Profits]

Thanks uva and LT...and is it the same calculation for moneyline underdogs too?

If you are looking a +185 vs. +140: 100/240 - 100/285. Remember for favorites, the numerator is the odd and for underdogs the numerator is 100. The denominator is always odds + 100.

[jolmscheid]

Perfect...thank you for the clarification LT!

[jolmscheid]

So just to be sure, is this the correct way of finding the edge for favs and dogs:

For favs: (Projected line / Projected line +100) - (Current Line / Current Line +100)


For Dogs: (100 / Current Line +100) - (100 / Projected Line +100)

[jolmscheid]

Bump..

[LT Profits]

Looks good. Now what do you do when your line is -105 and real line is +110?

[FourLengthsClear]

So just to be sure, is this the correct way of finding the edge for favs and dogs:

For favs: (Projected line / Projected line +100) - (Current Line / Current Line +100)


For Dogs: (100 / Current Line +100) - (100 / Projected Line +100)

No. it is the probability of your "projected line" divided by the implied probability of the "current line".

So in your example above

(185/285) / (140/240) = 1.1128
Edge is therefore 11.28%

The excel formula for converting US odds to probabilities was included in the excel sheet I uploaded for you yesterday.

=IF(B5<100,SUM(-B5/(-B5+100)),SUM(100/(B5+100)))
Where the odds are in cell B5 obviously.

[jolmscheid]

Thanks 4LC..yes the spreadsheet gives the implied probability, but I still need to figure my edge from my projected line correct?

[FourLengthsClear]

Thanks 4LC..yes the spreadsheet gives the implied probability, but I still need to figure my edge from my projected line correct?

Not really. If it helps you just insert an additional column D to manually enter the US odds of your projected line. Then copy the formulae in column B to the "Win Probability" column (would then be column E).

[jolmscheid]

Thanks 4LC...that works...or I suppose I could use the odds converter for my projected line to figure my edge...

For example, I could enter in -130 for the US odds, and if my projection for the game is -145, I can enter that into the odds converter to see what the % is...then I would know my edge...correct?

[FourLengthsClear]

Thanks 4LC...that works...or I suppose I could use the odds converter for my projected line to figure my edge...

For example, I could enter in -130 for the US odds, and if my projection for the game is -145, I can enter that into the odds converter to see what the % is...then I would know my edge...correct?

Sure but the Odds Converter just does exactly the same calculation.

[jolmscheid]

Yup you are right...I got it now 4LC...your spreadsheet is awesome...thank you...

[slope99]

Got it!!! Thx guys. I was missing Half Point (or more) Calculator and Pinny no vig lines.

I have to ask though how accurate is Half Point Calculator? OK Push Prob column is NOT up to date (I never bet even lines anyways). But what about the rest?

[slope99]

In regards to the Half Point Calculator:

Subs mentionned getting some good data and doing it myself. I have also read somewhere in the forums that hasn't been updated since 2009.

I'd love to do this. It would also help me understand how it works. Anybody know how the H.P.C. works?

Thx again.

[subs]

hi Slope,

the easiest way is just to look at pinny's alt lines, however i think that the sell points side is cheaper than the buy side WR to the vig.


i'm pretty sure doing it urself will be less accurate, however if u want to value pts outside pinny's alt range then doing it urself is the way.

the aim is u r looking for how many times a game with a certain line and/or total has finished with all available MOV.
so maybe there were 300 games w line (use a user specified range is better?) and total X and Y.

u need results that look like this:
MOV number of results
30 1
29 0
28 0
27 2
26 0
25 0
24 2
23 4

(all made up)


all the way down the possible MOVs

then divide results by number of games in range for push %


it can be done with a bunch of IF and countIFS statements.


GL

[Romanov]

No. it is the probability of your "projected line" divided by the implied probability of the "current line". So in your example above (185/285) / (140/240) = 1.1128 Edge is therefore 11.28% The excel formula for converting US odds to probabilities was included in the excel sheet I uploaded for you yesterday. =IF(B5<100,SUM(-B5/(-B5+100)),SUM(100/(B5+100))) Where the odds are in cell B5 obviously.

where is this spreadsheet? i've been looking in the thinktank and cannot find it

[FourLengthsClear]

where is this spreadsheet? i've been looking in the thinktank and cannot find it

It was a Kelly related spreadsheet. Post #17 in this thread.

http://forum.sbrforum.com/handicapper-think-tank/1525353-kelly-calculator-more-than-15-events.html

[slope99]

Great stuff subs! Exactly what I wanted to know.

If available I will use no vig Pinny Alt lines instead of Half Point Calculator or my own data. Yea you are right probably won't get more accurate than no vig Pinny Alt lines.

And I see now how Half Point Calculator works and how I could make my own.

[slope99]

hi Slope,

the easiest way is just to look at pinny's alt lines, however i think that the sell points side is cheaper than the buy side WR to the vig.GL

Man that might mess things up if I use pinny alt lines, trying to find no vig line.

[lifemagnetic]

Go with your gut and hope for the best.

[slope99]

i'm pretty sure doing it urself will be less accurate, however if u want to value pts outside pinny's alt range then doing it urself is the way.

the aim is u r looking for how many times a game with a certain line and/or total has finished with all available MOV.
so maybe there were 300 games w line (use a user specified range is better?) and total X and Y.

u need results that look like this:
MOV number of results
30 1
29 0
28 0
27 2
26 0
25 0
24 2
23 4

(all made up)


all the way down the possible MOVs

then divide results by number of games in range for push %


it can be done with a bunch of IF and countIFS statements.


GL

Sorry but what is MOV? Difference between final score and the spread? If so that would make sense. For example 300 games spread -3.5, 20 games where favorite won by more than -9.5. So probability that favorite covers -9.5 spread when listed as -3.5 equals 20/300 = 6.7% (what you call push % I believe)

Where can I get all this data???

The ideal, I would think, would be to get it directly from Pinnacle (since closest to true odds) updated with new data as the season goes along.

And how far back do you go? 5 years? 10 years? The problem I see with going too far back is the game is different now than 10 years ago (can't touch the QB, fewer kickoff returns, etc.). Then again if you don't you have smaller database to calculate your probabilities from.