PSA: Large soccer underdogs = fake edges

This area is for discussion of proline plays. Picks, handicapping and edge plays all welcome.
Slips
Posts: 125
Joined: Sat Mar 07, 2015 10:30 am

Re: PSA: Large soccer underdogs = fake edges

Post by Slips » Sun Sep 03, 2017 1:05 pm

richhhh wrote: Right now plp is showing one edge for today .

Lots of books offering lines really but really only one that means anything is pinny. The rest are recreations books like bet365 or a handful of legitimate books like sbo which take more juice then pinny.

Then there is Matchbook showing us your above example.Problem is you can only bet 39$ at those odds. Not enough liquidity as you say.


If Pinnacle moves its line the whole market will move including Matchbook ( users will edge play Matchbook to pinny like we do with OLG) . Why is Matchbook so great? Lets say for example Matchbook is the boss is plp supposed to gauge if its liquidity is efficient? Also if Matchbook was boss then pinny would moves its lines towards Matchbooks. The Back and Lay concept is nice to reduce the noise of three way vig but I dont think Matchbook solves this known issue.
There is no boss book. Pinny dictating the whole market is simply not true, so the fair line should come from an average of multiple books. Any time the exchanges have a line with a narrow two-sided spread, even if it's just $39, I would trust that more than Pinny. At the very least the average of exchange+Pinny price is much more accurate than using Pinny alone. Betfair would be my first choice for exchange, as their volume is generally higher for soccer and their API is easier to access. All signs point to underdogs being overjuiced on Pinny--and thus overvalued on PLP--and incorporating the exchange price is the most effective way of fixing that.

MattyKGB
Posts: 217
Joined: Mon Apr 29, 2013 5:27 pm

Re: PSA: Large soccer underdogs = fake edges

Post by MattyKGB » Sun Sep 03, 2017 9:43 pm

Ok gang, buckle up...it's math time.

The purpose of this exercise is to determine the best way to convert a given set of money lines (with vig included) into implied win probabilities.

Terminology: "Vig" in this context means the sum of the odds-implied "raw" win probabilities minus 100%.

"Best way" will be judged by the largest log-loss (also known as loglikelihood). This is pretty standard for prediction contests.

Introducing the contenders:

1) Equally distribute the vig on a linear scale, a.k.a. the "Slips method":
V adjusted prob = V raw prob - 0.5*(V raw prob + H raw prob - 1).

2) Equally distribute the vig on a logarithmic scale, a.k.a. the "PLP method":
V adjusted prob = V raw prob / (V raw prob + H raw prob)

3) Equally distribute the vig on a logistic scale (choosing logistic because it's designed for values between 0 and 1)
C = sqrt(1 - 1/V raw prob - 1/H raw prob + 1/(V raw prob * H raw prob))
V adjusted prob = C * V raw prob / (C * V raw prob + 1 - V raw prob)

4) Equally distribute the vig on a probit scale (probit is also designed for values between 0 and 1)
V adjusted prob = Z(Y(V raw prob) - 0.5*(Y(V raw prob) + Y(H raw prob)))
where Z is the cumulative distribution function of the standard normal distribution and Y is the inverse cumulative distribution function of the standard normal distribution.

The original context of this discussion was soccer 3-way lines, but I don't have easy access to a soccer odds database and the math gets a little messier with 3-way lines so let's start simple: NFL, MLB and NHL.

NFL arguably has the most similarity to soccer in terms of market liquidity and fav/dog odds distribution, but it also has the fewest number of games per season so the smallest sample size.
NFL vig.jpg
NFL vig.jpg (87.82 KiB) Viewed 13664 times
Logarithmic takes the overall crown, but there is a lot of volatility from year to year.
The "year's chalkiness rank" is a ranking of each year's average raw win probability for teams that won their games. So 2009 had the highest proportion of wins by heavy favourites, etc.
Logarithmic tends to shine in the less "chalky" years while logistic wins in the more "chalky" years.

MLB:
MLB vig.jpg
MLB vig.jpg (46.26 KiB) Viewed 13664 times
Pretty much the opposite of NFL. Interesting...

NHL:
NHL vig.jpg
NHL vig.jpg (44.85 KiB) Viewed 13664 times
Same as MLB.

Thoughts?

MattyKGB
Posts: 217
Joined: Mon Apr 29, 2013 5:27 pm

Re: PSA: Large soccer underdogs = fake edges

Post by MattyKGB » Sun Sep 03, 2017 9:43 pm

Here's a worksheet so you can play with the 4 methods yourself.
Attachments
Vig examples.xlsx
(8.73 KiB) Downloaded 533 times

Slips
Posts: 125
Joined: Sat Mar 07, 2015 10:30 am

Re: PSA: Large soccer underdogs = fake edges

Post by Slips » Mon Sep 04, 2017 11:47 am

Fake edge on 87H (Bolivia) for tomorrow - should be -5% according to Betfair, not +2.2%.

I think you math nerds are missing the forest for the trees. We could go back and forth forever on the best abstract model, but we can all agree that PLP is seriously overestimating these soccer dog edges.

That said Matty, interesting results. A few suggestions if you're amenable.. could you show how the methods rank overall across all sports combined (would be nice to include NBA and CFB/CBB if possible)? And also could you include the "naive American-odds midpoint" and see how that compares? I have a hunch that might fare surprisingly well since it's a good basic model for how bookmakers adjust their lines in the real world.

MattyKGB
Posts: 217
Joined: Mon Apr 29, 2013 5:27 pm

Re: PSA: Large soccer underdogs = fake edges

Post by MattyKGB » Mon Sep 04, 2017 8:23 pm

Slips wrote: I think you math nerds are missing the forest for the trees. We could go back and forth forever on the best abstract model, but we can all agree that PLP is seriously overestimating these soccer dog edges.
Some of us get their kicks from finding problems, some of us get their kicks from solving them. I'm in the latter group.
Slips wrote: That said Matty, interesting results. A few suggestions if you're amenable.. could you show how the methods rank overall across all sports combined (would be nice to include NBA and CFB/CBB if possible)? And also could you include the "naive American-odds midpoint" and see how that compares? I have a hunch that might fare surprisingly well since it's a good basic model for how bookmakers adjust their lines in the real world.
Linear rank Logarithmic rank Logistic rank Probit rank American rank
MLB 4 5 2 3 1
NFL 3 1 4 2 5
NHL 3 5 1 2 4
Total 4 1 3 2 5

ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

Re: PSA: Large soccer underdogs = fake edges

Post by ProlinePlayer » Tue Sep 05, 2017 7:15 am

Slips wrote:
richhhh wrote: Right now plp is showing one edge for today .

Lots of books offering lines really but really only one that means anything is pinny. The rest are recreations books like bet365 or a handful of legitimate books like sbo which take more juice then pinny.

Then there is Matchbook showing us your above example.Problem is you can only bet 39$ at those odds. Not enough liquidity as you say.


If Pinnacle moves its line the whole market will move including Matchbook ( users will edge play Matchbook to pinny like we do with OLG) . Why is Matchbook so great? Lets say for example Matchbook is the boss is plp supposed to gauge if its liquidity is efficient? Also if Matchbook was boss then pinny would moves its lines towards Matchbooks. The Back and Lay concept is nice to reduce the noise of three way vig but I dont think Matchbook solves this known issue.
There is no boss book. Pinny dictating the whole market is simply not true, so the fair line should come from an average of multiple books. Any time the exchanges have a line with a narrow two-sided spread, even if it's just $39, I would trust that more than Pinny. At the very least the average of exchange+Pinny price is much more accurate than using Pinny alone. Betfair would be my first choice for exchange, as their volume is generally higher for soccer and their API is easier to access. All signs point to underdogs being overjuiced on Pinny--and thus overvalued on PLP--and incorporating the exchange price is the most effective way of fixing that.
I would disagree and have the same opinion as richhh. I do believe there is a 'boss' book and that book is Pinnacle. I also find that a line move by Pinnacle is all it takes to trigger a move across the entire market. Just simply watching Pinnacle moves and then betting with books that have not yet adjusted is considered in itself to be a winning method.

Getting a fair price by averaging several books has been suggested before. It is in my opinion a very poor method. Taking the strongest line and then averaging in the lines from weaker books is only going to weaken the best available number.
As I stated before, I am not really a big fan of Betfair. The low spread between the back and lay is very much an illusion. A +600 line at Betfair should really be considered as a line of +570. Once the line has been adjusted for commission the real vig is much higher than that charged by Pinnacle. And as a general rule I would accept the line from the book which has the lowest vig.

PLP

ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

Re: PSA: Large soccer underdogs = fake edges

Post by ProlinePlayer » Tue Sep 05, 2017 7:34 am

Slips wrote:Fake edge on 87H (Bolivia) for tomorrow - should be -5% according to Betfair, not +2.2%.

I think you math nerds are missing the forest for the trees. We could go back and forth forever on the best abstract model, but we can all agree that PLP is seriously overestimating these soccer dog edges.
You keep using the term 'fake edge' as if it were an accepted fact. But the truth is that although I'm open to considering other methods of calculating the edge I am not convinced that it is necessary. It seems most are of the same mind.

What you call a 'fake edge' is an edge which disagrees with the Betfair line. It does not mean the edge is wrong. It is simply a matter of which book you believe has the the stronger lines.

PLP

sharpedgepicks
Posts: 492
Joined: Tue Mar 11, 2014 11:18 am

Re: PSA: Large soccer underdogs = fake edges

Post by sharpedgepicks » Tue Sep 05, 2017 10:52 am

PLP, please keep posting odds the same way you've been doing as it seems to work for most customers. Also, changes cause confusion or adjustments by users.

If Slips is right, other 'sharps' or blind PLP followers will be betting fake edges and bad results and he will benefit - he always wanted PLP to post less edges/odds so I guess fake ones are good too...

My 2 cents: bettors love big faves, think they are locks and almost all the VIG is built into the fave. I know this goes directly against the longshot bias theory but I think that's wrong and dated. I've been profitably cashing super longshot soccer tixs for over a decade. Maybe I'm just really lucky and the benefiter of fake edges, who knows?!

sharpedgepicks
Posts: 492
Joined: Tue Mar 11, 2014 11:18 am

Re: PSA: Large soccer underdogs = fake edges

Post by sharpedgepicks » Tue Sep 05, 2017 10:56 am

IMG_2178.jpg
IMG_2178.jpg (800.31 KiB) Viewed 13498 times
Recent win... just b/c some here hate my shameless self promotion. I think of it as aspiring youth to overachieve...

Arch
Posts: 213
Joined: Wed May 01, 2013 10:23 pm

Re: PSA: Large soccer underdogs = fake edges

Post by Arch » Tue Sep 05, 2017 1:43 pm

sharpedgepicks wrote:
IMG_2178.jpg
Recent win... just b/c some here hate my shameless self promotion. I think of it as aspiring youth to overachieve...
Grats on those underdog wins mate :thumbup:

Post Reply