90 Degrees | Episode #33 Building a WINNING Model & If Parimutuel Betting Is The Future Of Sports Betting
Hey, on this week's episode of 90 degrees, we talk about finding value betting on extreme outcomes.
We teach you how to build a successful betting model and our paramutual markets,
the future of sports betting, that and more on today's episode of a 90 degrees podcast.
Welcome to the 90 degrees podcast. I'm your host, G. Stath George, and I'm excited because
today I've got a guest who I look up to the way he thinks about sports betting, the debate he
engages in on Twitter. Some might be, might be, it might feel like a royal rumble to somewhere
watching from the outside, but I'm glad to have him. His name is plus EV analytics,
actuary gambler, instructor of the art of sports betting analytics. Plus CV, thanks for joining me.
Thank you for having me. I want to ask you what was your first introduction to sports betting.
It doesn't even have to be the first time you made a bet. When you learned that sports betting
was actually a thing. That would have to go back to reading the sports section of the newspaper
and my childhood and they would publish the odds. I remember seeing minus 180 and PK and plus 140
and having no idea what that meant, but started this desire to learn more about this.
I've always been into sports. The best day of my childhood was games six of the 93 world series.
I'm from Toronto, just like you, George. My dad actually took me to the game. I was in the 500
level for the Joe Carter home run, which was awesome. When I've always been into sports,
always been into math. When I first started to learn probability in school, I was like, yeah,
this is something that really, really fascinates me. Love of sports, love of math, love of gambling,
this whole idea of you're throwing your money in and you know what's going to happen and you're
winning and you're losing. This is the exhilaration of it all. It comes with all the uncertainty of
not knowing whether you're going to win or lose. That's always been kind of a part of what I'm
all about. I never really thought a lot about sports betting until about 2010 or so. I did play
a lot of poker, a lot of online poker, a lot of live poker, not high stakes, but you know,
sit and go tournaments on poker stars, things like that during the height of the poker boom.
But eventually that kind of went away. The games dried up. Games got tougher and I didn't have,
you know, six hours a day to sit around playing online poker anymore. And that's when I started
thinking more seriously about sports betting. And I think I've told the story on a couple of
other pods. So apologies to your listeners who have heard it before, but my first sort of
inkling that this could be plus EV was when I'm playing these lottery, Ontario lottery, you know,
parlades for 10 bucks, 20 bucks a pop just for fun. And I had parlayed a hockey team to win the game
with one of the players on that team to outscore one of the players on the other team. And I was
into it enough to understand what a correlated parlay is and why that would be good. I didn't think
that would be enough to beat the 39 percent hole that was built into this parlay product. But when I
went to the store to hand the cashier my slip, I ran it through his machine and the bet got rejected.
The Ontario lottery corporation would not take my action on. I think it was a $10 parlay. And the
reason was that the system had already exceeded their maximum liability for that specific combination.
So driving home, thinking to myself about what could possibly have happened to cause that,
the only conclusion I could come up with was that enough smart people out there were playing the
same ticket that I had tried to play. And the only reason they would do that is if it is an
advantage situation. So got home, started working on my first sports betting analytics model that day.
And I had a good run of it for a few years. Beautiful. I want to catch up the listeners for those who
don't know. OG is like the Ontario lottery gaming corporation. They offered sports betting.
They offered certain events where you could, you had to bet all the point spreads, you know,
minimum two up to maximum. I think it was 10 of them or 12 of them. And they would give you less odds
that you would get in a sports book. But there was some built in advantages. The lines were
stale. Once they put up the lines, they only adjusted them once on the Friday. So you can imagine
come Sunday. How many lines were no longer applicable. The other thing was they had other options
where you can bet. It was called pro line where you can bet some events and outcomes. And a lot
of people found advantages through correlated outcomes. So you could still overcome a high built
in big. And plus he needs not the only one who had thought of this. There's a lot of people who
have been beating the old G market for a while until they started to adjust and close some of
those gaps. I wanted to talk about you said you figured out a lot of smart betters are doing
this and it kind of satiated you to go out and build some models and figure things out.
Outside of the money is that what rewards you is the gamesmanship of I can beat this
finding the correct strategy. Oh 100%. Now the money is great. I'm not going to lie about that. But
beyond that, it's an intellectual challenge. And it's one that I sometimes don't get in other
aspects of my life. So many people know I'm not a professional better. I have a day job
which I love as well. But I find that there is a more direct link between ability and success.
In gambling than in business. Despite all the randomness, anybody could have a good day or a bad
day or a good year or a bad year. But the great thing about success in gambling is it's not about
people's perception of you. Well, your boss likes you so you get a raise or you get a promotion or
whatever. You are in control of your own success or failure. It's not about looking good. It's about
being good. And yeah, you know, anybody could win or lose any given day, week, month or year.
But long term success is really about how smart you are, how much work you put in, the connections
you make, how much you can learn. You know, it's entirely about your own abilities.
Now, you say you're in your day job, you're in actuary by trade. Is there a skills overlap
between that in sports betting? Oh, 100% there is. And especially the type of work I do,
which is pricing for insurance, it's kind of analogous to being an odds maker.
People ask me, okay, you're in actuary, that must make you a good gambler. And that is somewhat true.
There are a lot of skills and techniques that I've learned throughout my education and my
experience that have helped me. But what I would say is even more true is that being a gambler
has made me a good actuary because I deal with concepts like risk, expected value, variants.
To me, they're not just concepts in a textbook or on a spreadsheet. There's something real
that I interact with every day outside of work and get to know intimately. And that really helps
me apply a lot of the concepts that have been kind of inspired by my gambling career in my real career.
You know, one of the things that I apply to like a lot of aspects of life, not just gambling.
Whenever somebody presents me with a dilemma because I'm apparently I'm a very good advice guy,
someone to sound off on when you've got a tricky situation. I always break down with people,
what's the upside, what's the downside and walk through all the different
examples. And then if the upside is worth the potential downside and a lot of that,
it's like a simplistic way of saying, you know, expected value, which is something that you have
in your name. How do you apply math into your sports betting? Say you see a line that you believe
is mispriced. How are you checking that with math for those that don't know? So before we get into
that, what you said is actually very, very valuable about looking at your upside and looking at your
downside. In a way, it's actually superior to expected value, probabilistic kind of thinking,
because in a lot of this, it's hard to put exact probabilities on things that might happen in the
future. My favorite author, a guy named Nassim Taleb, has a book called Anti-Fragile. It talks
all about, okay, well, the best way to manage risk is not to try to predict the future. It's to put
yourself in a position where you have more upside from unknown surprises that then you have downside.
So what you're saying really reminded me of that and it's a great strategy.
You know, I just just to interject one second. You know, there was a famous instance a couple
months ago where Steve Fesick, and for those who don't know, is a professional sports battery,
also sells picks. He ended up losing a lot of money laying minus 6,600 with Purdue. And there
was a whole discussion about if it's plus EV, it's still a good bet. And I was talking with
Johnny from Betstamp about this on my podcast. And I said the one thing that I don't think we
discussed enough is the risk of ruin when we're discussing positive EV examples.
You're risking 66 units to win one essentially by laying that bet. And some people say, well,
maybe he wasn't trying to win a unit. And he didn't risk 66% of his bankroll to do so.
Then I say, then what's the point of trying to capture even a quarter of a unit and risking even
20% of your bankroll? It's just the risk of ruin. I don't care that that situation is plus EV.
There are certain situations I'm not willing to do like blow my whole bankroll on one plus EV play.
And I think that when we talk about weighing downside versus upside, the upside wasn't great enough
for me to risk all that downside in that situation. Yeah, in general, I agree with you. I think
in physics case, him having been around for as many decades as he's been around. I think his
percentage of his bankroll he was risking on that one play was probably insignificant in terms
of risk of ruin. So I think that's a good general principle that you never want to put yourself
out of substantial risk of ruin. Probably wasn't a huge consideration for this bet this day.
I know people like to say, I mean, never lay more than, you know, minus four under, never
lay more than a thousand minus a thousand, anything like that. I don't personally subscribe
to that philosophy. Leader, do I? It's just that the okay, if it wasn't, you know, a large piece
of his bankroll. So what was he trying to scoop up like breadcrumbs, a 10th of a unit? To me,
again, it doesn't, when you hit to a certain price that's large enough, even if you're not risking
your whole bankroll to win one unit, you're risking a percentage of your bankroll to basically collect
nothing, something insignificant to your bankroll. I get it's a pluscv situation, but there's a lot of
pluscv bets you can make and you don't have to make every single one because you got to,
because to me, you still have to factor that downside. And I don't believe in that hard line of
minus 400, I won't bet anything higher. But at minus 6600, I get to that point where it becomes
uncomfortable. Yeah, I didn't wake up this morning thinking I'd be here defending Steve
Fezick this this evening, but here kind of here we are. And I think that the one thing that we also
have to keep in mind is that the favorite long shot bias is real in a lot of different sports
and a lot of different markets. And so when something is priced at minus 1,000, minus 2,000,
minus 6600, it's at least possible that the favorite long shot bias is so extreme on this,
because a lot of people think like you they don't want to lay minus 6600 ever on anything. So
for the book, you know, that they're going to end up for the market, they're going to set their
pricing such that it's going to be quite unfavorable on the other side, which people will find more
attractive. If people like to lay a little to win a lot, you know, whether that's minus 10% EV
or minus 40% EV. So oftentimes you will find, I don't know about this specific example because I
don't I don't have a model that would give me a number on this game. But just in general,
you know, you you can very easily pick up the best of it on a lot of these things by laying a
big price. In general, I have no opinion on that specific bet and whether it was good or bad,
but it's at least possible that this thing had a 99% chance of winning or for whatever it is.
I think what what the kind of counter argument on that is there's so many unknowns in a sporting
event that even if you think who was Purdue, forget forget the team, you think they're you think
they're 99% to win, you know, you could very easily be wrong and you don't have to be wrong by
much to turn a plus EV bet into a minus EV bet when you're dealing with that level of juice. So
those are the arguments kind of on on on both sides that I would put forward on on that example.
All right. Before I interrupted you, we were going to talk about applying math to sports betting.
Yeah. And applying math to sports betting, you know, in a way, it's simple. You just you look at
all the possible outcomes. You assign a probability each probability times your your pay out equals
your expected value. And if it's positive that it's negative, don't, you know, I realize it's
probably an oversimplification, you know, compared to what we're where you're trying to go with this.
But I think that the real tough part is, well, how do you know what the probability is on any given
event? The cool thing about sports unlike pretty much any other gambling proposition out there is
you are not dealing with a controlled universe of known probabilities like you are at a black
jack table or a perhaps table or flipping coins or whatever. You are always estimating
probabilities. And that is a whole different world than operating in known probabilities because
now you have to deal with, oh, what if my estimate is wrong? You know, when I flip a coin, I know
the chance of getting heads is 50%. It would take, I know there are biased coins out there,
but it would take a lot, a lot of data to convince me that any given coin is not 50, 50.
But when you're dealing with the probability of team A beating team B, you're you're not in the
world where the laws of physics naturally dictate probabilities. You have to figure it out
on your own. That's what makes sports betting so much fun. So how confident are you in the numbers
that you put out? Do you ever question your math? Are you a hardcore? This is the math and how do
you apply a subjective lens being someone who looks at sports betting through a numbers perspective
and a math based perspective? I don't question the math because math is math, but what you can and
should always question is anytime you're building a mathematical model, model contains data and
assumptions and assumptions are always the tricky part and whatever one should do when they're
building a model is step back and think about what assumptions are either explicit or implicit
in the model they're building and sometimes the assumptions are something as simple as
the game today will sort of be similar to the game that was played last year or two years ago
or three years ago and sometimes that's not even true. The way the game itself is played changes
over time. You look at basketball and the number of three-point attempts goes up pretty much
every year, year over year. So any model that you're building based on historical data implicitly
assumes that the game is going to get played in 2023 like it was in whatever time period that you're
using and a lot of times you just don't even think about that as an assumption you're making. You're
okay well I'm taking my data and I'm throwing it in a model because that's what people do and I
think that is where people can get tripped up not that the math is the math is not wrong it's the
application of it it's the okay well I'm going to build my model based on data from 2018 to 2022
and use it in 2023 that only works if what happened between 2018 and 22 is
um related in a very predictable way to what's going to happen the distribution of possible
outcomes in 2023. Hey before we get back to the show I want to talk about our sponsor pinnacle
pinnacle is the world's sharpest sportsbook and available to betters and Ontario find out what
professional betters have known for decades pinnacle is where the best betters play must be 19 plus
an Ontario please play responsibly not available in the US now back to the show let's put a pinnacle
because I want to get back to it and I really want to show people a breakdown on how to build a model
you you said something about we were when we were discussing Purdue and how it's generally known that
you know the general recreational better loves a longer shot and loves the big odds and because of
it oftentimes you find yourself with great prices like this super balls the big example of will
there be a safety yes no and you know laying big juice is actually very plus EV but the thing is
you've done well betting extreme outcome measures in futures markets with big prices tell us a
little on how you're finding an edge doing that sure so I think the best way to illustrate that
would be if we start off an example say you're flipping a coin so you're going to simulate the
outcome of an average team over a 17 game NFL season by flipping a coin 17 times and counting
how many heads you get and that's the number of win dues you're sort of singing so let's price
two different kinds of bets using this coin flip one is a standard line so eight and a half wins
probability of less than eight and a half wins flipping coins is going to be exactly 50%
and then let's look at what our alt market might be so what's the probability of
under five and a half heads flipping 17 coins this is you know your standard textbook
binomial distribution problem they teach you in high school probability and you can calculate
the probability of about seven percent of under five and a half heads fair odds for that
alt market would be about plus 1294 on the under five and a half heads in 17 coin flips
okay cool now let's switch it up a little bit the Denver Broncos the market win total on them
as we record this in late June is eight and a half so the market expects them to be a league
average team I have no reason to disagree with the market on that so let's go with that Broncos
win total is eight and a half the only difference between the Broncos and the coin is we know the
coin is 50-50 we think the Broncos are 50-50 but we could be wrong so let's suppose that the Broncos
might be an eight and a half win team they might be a six and a half win team they might be a 10
and a half win team with equal likelihood of all three of those events and we're going to price
those same two markets again so your standard market under eight and a half is still 50% so we
have not lost that consistency with the market number by switching from we know the Broncos are
an eight and a half win team to they might be six and a half eight and a half or 10 and a half either
way the true number on under eight and a half is even 50% so we haven't lost that anchoring the
market number now let's price our alt under five and a half with this extra uncertainty added
that we don't know whether the Broncos are six and a half eight and a half or 10 and a half win team
and if you do that your probability under five and a half is now 13% not seven like it was with
the coin flip fair odds are now plus 662 instead of plus 1294 so what's the difference between
the Broncos and the coin it's that we've introduced something called it's called parameter uncertainty
in some places it's called epistemic uncertainty in some places I like to call it the hidden generator
so there is this hidden thing called how good are the Broncos and we have an estimate for what
that is but we don't know what that is for sure and that adds extra variance to this distribution
especially I should get toward the tails or the more extreme outcomes it doesn't do much adding
this extra dose of uncertainty doesn't do much when you're working in the middle of distribution
but it does a lot when you're working in the outside so what I've been able to do over the years
is do pretty well betting plus money so this is the opposite of the other example with favorite
long shot bias this is this is books not properly pricing in this extra layer of uncertainty in
the distribution and it has led to some extremely plus EV spots on whether it's alternate lines
where the total is eight and a half but you're taking under six and a half or over 10 and a half
or whether it's even an exact number of wins so I did pretty well I think it was on the Rams
last year where their market number was like eight and a half or nine or something and so I think
I bet them to win exactly two exactly three exactly four I forget how many they actually won but
one of those actually hit at a I think it was 200 to one or 300 to one or something so it was a it
was a good it was a good outcome because the odds for each of those exact win totals were priced
more in line with the coin flip so the chance the chances of two heads and how many games of the
Rams wins two three five and five okay so the chances of five heads out of 17 flips are not the same
as the chance of eight and a half win team winning five games there is so much more likelihood
that the eight and a half team will win five games because hey the Rams might be worse than we
thought the Rams might be better than we thought where the coin flip you know exactly what you're
dealing with got it now this can apply to a lot of different teams is this where a subjective
layer comes in and you say okay here's the eight teams that I believe their upside or downside is
not being captured properly like I call them rigid teams versus teams with wide range of outcomes
we all know who the Kansas City Chiefs are and it's going to be very hard to capture extreme
outcomes with them whereas a team like Baltimore with a new offensive coordinator
you know Cleveland with an unknown quarterback Pittsburgh with an unknown quarterback a lot of
these teams are they have a lot wider range of outcomes so when you're gathering which teams to
look at and see the prices on how do you gather them so you're absolutely right that the the
width of that distribution the the magnitude of that parameter uncertainty is going to differ
from team to team haven't yet come up with a way to quantify that accurately I've tried a
couple things and maybe I'll try again this year one thing I did find last year this is not about
the width of distribution more about the position of it that the wind total markets seem to be
off in systematic ways when you have things like for example a second year quarterback the market
tends to underestimate the the leap that a team would make when they have a second year quarterback
at the helm and I found going back I think I went back about six or seven years there was a
significant kind of miscalibration where teams with second year quarterback would hit the over much
more than the under to a to a significant degree something else around if you have a a either a
new head coach and a returning offensive coordinator or a returning head coach in a new offensive
coordinator you know that's something the market overvalued it was a negative impact that the
market didn't really pick up stuff like that and it's you can find the article from back from last
year on on my site plus CB analytics dot com but you're you're right and that's something that I
I think about intuitively but I haven't been able to build into the model yet the the idea that the
the magnitude of the uncertainty is is is different between you know between different teams and
one of the ways that that manifests is in the how big is the drop off between your starting quarterback
and your backup quarterback is one unexpected thing that could happen is your your quarterback could
get injured and that could that could screw up your your whole season for a team and then my
friend Andy of the deep dive podcast coined the term clipboard Delta which I loved as the the
magnitude of the drop off between your starter and your backup so obviously for KC would be huge
for a team with a more average or below average starter and maybe a good backup that drop off might
not be I'd not be so big so that if the quarterback gets injured you really haven't changed the
distribution much where obviously if my homes or Josh Allen gets injured you know you've just taken
I don't know four or five wins away from from the full season expectation of that team
yeah it's it's also the two points to that is I remember when Lamar Jackson went down and a lot of
people they they only accounted like three point gap between him and Tyler Huntley because
Huntley showed a little bit of flashes in a small sample size and to me like there's certain quarterbacks
that they are the entire offense it's not just what he does when he's running the ball himself
or throwing it's all the running backs on that team average five and a half yards per carry because
the defense has to account for him Lamar Jackson himself so it's it's to what level do they raise
everybody around them up and the other part about you talked about second year quarterbacks taking
leaps like some of the most recent memories is the mahogas doesn't start as rookie year second year
takes them to the AFC championship game and Lamar with the MVP season 14 and two and
Jalen hurts last year or Joe burrow the year they went to the Super Bowl that those are some of the
cases and people don't realize is that second year even third year quarterback if they
redshirted a year sometimes those are the best biggest edges where if you take stabs at like
alt lines or futures markets you can find yourself with tremendous value I want to talk about
some stuff in the news if you don't mind if we can move to current events and I always
worry about current events because once they happen if someone were to listen to podcast five
months from now they it might not be you know relevant but I think there's some underlying stuff
that we can discuss there was the story about the NBA draft and champs tyrania taking a little bit of
fire this past week and a half he's of course the legendary NBA insider he works for the athletic
and stadium but he also gets paid now by fandall to co-host the show he changed the odds single
handedly on the NBA draft second overall pick market just by saying hey I'm hearing
there's a lot of momentum for this player to go to this team now people are questioning if it was
nefarious or not I don't believe it's nefarious but at the very least it opens yourself to some
criticism right yeah I think this one's been been discussed a lot on Twitter and the consensus
that I agree with is that there's just no there's no possible way that he would have done this
with any kind of you know ill intent I think his biggest mistake was actually taking money from
fandall in the first place because you know how could you not foresee something like this happening
where you you make a call and maybe you're wrong and people bet it you move the market and people
are gonna say well he works for fandall he's you know he's trying to screw us but if you look at a
guy like shams who has built an entire career off of breaking news and you ask yourself how much
would you have to pay this guy to put out something that is deliberately wrong and torpedo his own
credibility for the rest of his career like I don't know 20 million 30 million you have to pay
this guy to do that and there's just not anywhere close to that much money you know in the in the
market so and I think fandall is doing you know planning well without having to resort to tactics
like this uh to squeeze a little bit of extra money so you know it it's obviously ridiculous but
it's in a way his fault for setting himself up putting himself in a position um you know as a
trusted news source to to be affiliated with a sportsbook because even though there's no merit to
this there's perception and that that means a lot too do you think he's aware of the chatter that
happened uh following that and do you think he'll be gun shy in the future about you know
predictive stuff that he doesn't have concrete goods on I don't think so I think if if he was
it would be a lot better for his career to just sever his relationship with fandall and just go
back to to breaking the news because I think that's you know he's he's either number one or number
two depending on how you feel about woj you know in the world at what he does so if I was in the talk
two in the world that what I did you know I would make sure I didn't do anything to to jeopardize
that and so taking you know taking a paycheck which I can't imagine how large it could possibly be
from fandall you know that that's something that I would give up before I change the way I operate
all right what about this because I think this one drove me up the wall um the Toronto mayor race
had been happening for weeks and sportsbooks were offering lines and at several points Chris guy
was the second favorite with odds as short as plus 500 and I got sent this updated you know betting odds
by several different people and I just kept saying because I bet politics ago this is like a
nothing story he's not even gonna finish top six he's not even gonna get 2% of the vote
and I'm trying to explain to them because when when they hear me talking about preaching about
market efficiency all the time and then they see Chris guy rockets to number two I try to explain
to them you know this is the danger in reading into too much into markets that aren't liquid
and the betting limits are so small I sit to them literally I could take 500 dollars right now
if they'll take my if they'll book my bet and I'll change the whole market right now that's how
irrelevant this kind of information is did you come across this Chris guy short odds become the mayor
did you have a chuckle about it what do you think about that I actually never even heard of Chris
guy until until you told me about him are you not on the internet during the last three years
I know who is he I never I don't follow politics that closely especially not Toronto mayoral
politics he's not even a politician he's the big you know muscular guy with the big neck tattoo
that was screaming about Mac masks and anti-vax and yeah like he dipped into all the conspiracies was
on the either the Alex Jones tour he was running for mayor in Toronto and I guess he had a dedicated
fan bet fan base who kept putting on hundred dollar bets firing at bet three six five and they
made in the second short favorite a guy who finished with one percent of the vote okay so hearing
that from you makes the story make a lot more sense now because now it reminds me of back in the
the the Trump last Trump election that Trump was taking money like two weeks after the election
was over because people were so sure that the election was gonna get overturned even though there
was no possible way it was going to when it was just people sort of betting their beliefs
with with no kind of rational thought as to value so knowing what I know now about who this guy is
that would be my guess as to as to what happened here you're right about the market being small
not a lot of liquidity and I think market efficiency well I definitely believe in it on the whole
I think it can get overused especially in these smaller illiquid markets where all it takes
a couple of bets to move a number and you talked about it a little bit with with Jeff Davis
last week that when you're when you're dealing so-called information markets or he called them
no variance markets which is an awesome term I love that whenever you're taking action you really
have to give it a lot of respect no matter what it is or where it's coming from so I don't blame
the books for moving the way they did but yeah it's information markets is just a whole a whole
different world he talked about it with Drew as well a couple of weeks ago he's he's much more
into that stuff than than I am but just the the idea of betting on something that is not even a
known outcome but is not subject to the same randomness in the outcome that a sporting event
might be yeah it just changes the market dynamics so much you flagged the tweet to me from Jeff
Feinberg and it said I get good numbers because I have well had so many books I'm not a sharp
I just bet at the book the best number very normal customer behavior if I'm buying a dishwasher
and get a good price at one store well they never let me shop there again what part of that do
you want to chew the fat above yeah I am I have a huge fascination with sports book economics
tweeted a lot a tweet a lot about I take a lot of heat about it I think it actually has a lot
in common with insurance industry economics and so going back to Feinberg's tweet you know he
was upset he's saying well I'm not a sharp guy I don't have a model I'm just betting on the
golfers I like the only thing I'm doing that the books may not like is I'm price shopping for the
best one and how can that possibly be enough to get me kicked out of these books and it absolutely
can be enough to get you kicked out of these books because again going back to your excellent
interview with with with Jeff Davis last week it is really really hard to be a book maker you've
got to put up lines on every game every sport every market every day and when you have 2025 30 sports
books operating independently in the market putting up their own lines you know you you and you
look at you can even see it on bedstep the the spread between the best line in the market on one
side and the best line in the market on the other side well all of a sudden you're not holding
4.5 percent anymore like you are with your basic minus 110 you know depending on the day the
game the market you might be holding 1.5 percent 1.5 percent sit that a cold even zero or negative
sometimes and and so when that happens the market can be beaten with no knowledge of anything
other than just looking at all the numbers and picking off the best one for whatever you want
to bet like if you're if you're odds shopping you could bet it doesn't you can bet your favorite
teams you can bet only Monday night football overs you can bet whatever you like but as long as
you are executing the bet you want to make by having 20 sports books open and taking the one
that gives you the best number you are never going to be paying that 4.5 percent big you're always
going to be paying less and in some case in some cases much less and I I think I'm a little bit
unique in this take but I see that as an existential risk for the sports betting industry because right
now we as users of of odds screens whether it's bet stamp unabated spank odds don't best what they're
they all do pretty much the same thing we are in the minority as users of those tools within the
betting world and I ask myself well what happens if people who use odds screens are no longer
as small a minority as they are right now we don't even have to assume that people are getting any
sharper at analysis or modeling or stats or anything like that all it takes in my opinion is for
the the percentage of the betting population that is shopping for the best line and using as many
sportsbooks as they can if that goes from 5% to let's say 50% I think that would be devastating
for the margins in the industry as a whole now of course some companies would fare better in that
scenario than others the sharp books would would survive longer than the soft books because they
would at least be moving numbers appropriately and they would be kind of on the wrong side of it
less often than the soft books would be but nobody is going to be on the wrong side of it never
like that that's going to happen and if that happens more and more often you know I think the whole
the whole model is in jeopardy these books like when I when I listen to your interview with Jeff
Davis it reinforced in my mind how difficult a job that is like his job is 100 times more difficult
than mine pricing insurance because you know you can't you're not allowed to call from a house
it's already on fire and so I want to buy an insurance policy you're not allowed to do that but you
can look at the the news reel oh Patrick my homes was was ruled questionable four seconds ago
they're taking more than four seconds to move this line I'm going to jump on or even in game
somebody leaves a game with an injury and it takes the sportsbooks 25 seconds to update their
lines in that 25 seconds they can pile up all kinds of liabilities says that the the the
um the information the the sort of parameters of the risk at least an insurance are relatively
static over time but in sports betting the parameters of the risk can change minute to minute
second to second and you have to be on top of it every second of every minute of every day
and it was just exhausting even even listening to it and like how could anyone possibly do this job
and do it well because it just seems so overwhelmingly hard now you you're talking about the ecosystem
of sports betting getting smarter and therefore it becoming harder and harder to book make
is there an evolution that you foresee happening is there a solution uh something's going to have
to happen right industries just over time there is and there's a couple of different ways
this could go and I I can't tell the future I don't know what's going to happen I think that the
most obvious consequence would be if you look at the the the sports betting environment five years
from now 10 years from now I don't think you're going to see the same number of different odds on
the screen that you see right now because that this adverse selection problem gets worse and worse
the more dispersion there is in a number of different providers number of different odds in the
marketplace because the best out of 20 odds is generally going to be better than the best out of three
or the best out of five so that could look like consolidation where books merge or buy each other
up such that there's just fewer different numbers in the market you know it could look like
um an odd service that multiple books would subscribe to so that yeah you have 20 different books
but they're all pulling you know the same one or two different providers for for for numbers
and so you don't have the same um selection problem because everyone's posting the same number they're
just different marketing skins on it um one one of the takes that I've been
um roasted the most for on Twitter is my my uh faith in the paramutual model as a as a
potential solution to this because you know just like in poker when you're betting against the other
players the house doesn't care how sharp you are how square you are what your bank crawl is
what your risk tolerance is they just let you duke it out amongst each other and they take their cut
off the top so if you imagine what a paramutual sports book might look like
number one you don't have the same you don't have to manage risk because you don't have any risk
so this whole problem of how I'm going to stay on top of every piece of information every second
of the day disappears for you as the provider with that the expense of maintaining a trading team
disappears and and it becomes um a lot simpler from an operator perspective to run the game now
it's not perfect there are downsides with paramutual model that would have to be sort of dealt with
as uh I don't know if you play horses at all but a a common complaint of horse bettors is when I put
my money down this thing was eight to one and by the time the horses were we're out of the gate
it was four to one and if I had known it was four to one I wouldn't have placed the bet because
my fair price is six to one yeah that that can happen in a in a paramutual model you're not
fixed odds you don't you don't even know what odds you're getting um so explain explain to me now
a point spread is minus eight and sharps love plus eight and they line up one plus eight and
all the squares end up on minus eight and how would a book adjust the paramutual would they
adjust the line because the the money gap could be different I understand the theoretical
concept of how it could work with the money line but how would it work within a spread
so there's a bunch of different ways to do this and I'll just give you one so you make a spread
you know say it's eight doesn't even have to be kind of right or wrong it just is what it is so say
it say the number is eight and people in the market can take minus eight or they can take plus eight
and you don't care you're just collecting money on both sides and at the end of the game you'll
have an outcome either if the plus eight will win or the minus eight will win and you will just
divide up the pot amongst whoever picked the winning side so um let's say the line the true line
move to five and you're still sitting at eight you're going to get more money on plus eight
than you will on minus eight you'll still get some on minus eight because the people will see
that okay the pool is is overweighted to the plus eight so that the implied odds it might be
you know minus one fifty on the on the plus eight and plus one forty five on the on the minus eight
so the the the market efficiency can still hold in a paramutual system I think what you'd have to
have is really good information you'd have to have real time updating of you know how much money
is on each side of the pool and you could even have like separate pools where you know the common
criticism of this is well no one's going to bet until three seconds before the game because
why would you put your money in with with less information well you could have a separate pool
running every hour from you know 24 hours before the game to to game time and each pool will take
in money on both sides it will close and then the next pool will open and then each pool will get
sort of paid out depending on the money that was in that pool when it closed so again I'm
spitball and there's a million different ways that that you could do this um the the main point is
I think there is more long-term sustainability for the industry in a paramutual model than there
is in a fixed odds model because the um the ability of bettors to line shop to get sharper to
to whatever is likely going to outpace the ability of books it's playing offense versus playing
defense it's always easier to play offense than to play defense because to play defense you have to
anticipate where the offense is sort of going to come from where if you're the offense you have
what's called optionality you can you can bet you want to bet the uh the Broncos you can bet them
at any one of 20 different books or you cannot bet them at all that's your choice but they have to
offer lines on every game every market every day every sport all right you're a math guy give me
a percentage of likeness that we ever see a paramutual market being the future
and then I also want one for sports exchanges maybe being the the replacement as well
so paramutual pools exist like even even Ontario Proline um has run for for
I'm talking about wide-scale uh paramutual markets yeah what I'm wide-scale I mean couldn't happen
I I would someday love to build one um you know how how much of the market how much market share
will I get I have no idea if it ever happens um you know will will it ever become a hundred percent
of um the market unlikely um but I think it has potential to grow when other um other models might
struggle when when you look at the draft kings and the fandals of the world I mean they're losing
money right now because they're blowing so much on marketing but there's going to have to come a time
where they either start to be profitable or their shareholders are going to dump them and that's
going to be a real moment of truth for the sustainability of the industry when you know you can't
just coast on marketing spend anymore the industry is mature enough that you have to show your
shareholders that you can profit and you know and then how well is that going to go for them
nobody knows hey just want to remind you the easiest way to improve as a sports better
is use multiple sports books and always get the best odds we recommend using an odds comparison
tool like bet stamp best stamp compares odds across every sports book for games futures and player
props save time and money by checking bet stamp before you bet download the app today if you are
looking to sign up for a new sports book account please check out the offers available at betstemp.app
forward slash circles off or hit the link in the description if you sign up through this page
it really helps support the show now back to the rest of the show I want to talk about because I
promised I want to talk about let's build a model and we don't got to go into full scale details but
I want for anybody who is an aspiring modeler who doesn't even know where to begin to have a better
sense after listening to this podcast my first question my first question is does it start with a
theory you theorize that a could mean more outcomes of b is that where you start with your modeling
process it could be for me the answer tends to be yes just because I don't have the time or the
energy to say hey I'm going to originate a WNBA model from scratch and I could do it if I had
unlimited time and data and energy but a lot of the stuff I do personally tends to concentrate
on derivatives especially correlated parles I love correlated parles just because they are
they're kind of a smaller box because you can say well the market number on the side is this and
I have no reason to disbelieve it the market number on the total is this and I have no reason to
disbelieve it so taking those two things as fact what kind of consequences can I derive from that
so it's it's less onerous than having to say well I know nothing about this game and I have to
price aside or a total from scratch I mean again that's doable and there are people who originate
you know major sports out there and they're very good at it I'm just not one of them all right we
start with a theory where do you how do you scrape the necessary data to start building this model
yeah I wouldn't say scraping is a great skill of mine there are people who are just
wizards at scraping data from all kinds of places I tend to just find a publicly available source
like I think it's called sports book reviews online has a great historical database for all the
major sports of lines spreads totals and and the outcome of the game so that that's kind of my
my go-to easiest possible spot to to get data but there are there are others up there I mean
feel like if I wanted to build a WNBA second half total model I would just start by googling
you know where do I get WNBA second half data and hopefully something would pop up but I just
kind of see see what's out there and if not I have a great network of friends Twitter followers
other betters and you know just see try and source that stuff through through a network all right
you've got your theory you've got your data how are you what program are you using to build your
modeling um either excel or are so if the number of rows is of a manageable size let's say a
hundred thousand rows or fewer in my data set I will tend to go to excel I know I'm I get a lot
of hate from the data scientists snobs of the world by you know they say you know excel is not a
real data science tool I vehemently disagree with that there is a ton you can do in excel that is
as good or better than what you can do with our Python or anything else what I love most about
excel is when you have a complex algorithm that has a set of steps in it you can actually see column
by column step one step two step three rather than them being different blocks of code so that if
something looks weird or there's an error somewhere it's very easy for me to see where it is rather
than having to you know comment out different sections of code run only this section and look at my
data set what's wrong deepbubbing is much much easier in in excel if the number of rows is too
large to to function normally in excel I will use R I just started learning R a couple of years
ago during covid and I have found it useful especially for larger data sets so yeah excel are
all right you've got your data you you have your theory you've put it into a program
it spits out a number is that already back tested or are you back do you have to manually back test
well you don't have to do anything but it's a good idea to back test any model that that you're
building preferably using something called a holdout sample where you'll take let's say 30% of
your data set randomly select it put it to decide don't use it in building your model and use it
for testing at the end that way you protect yourself against what's called overfitting where you
just have a model that predicts the past really well but doesn't predict the future well at all so
when you're back testing what you want to see in general for a betting model is a strong correlation
between your predicted your modeled EV and your back test ROI so on bets where I would have given
you know between 10 and 15% expected value if I had made those bets what would my return have been
um you know if it's between 10 and 15% okay thumbs up if it's 5% then it's like all right well
my model is good I'm picking winners I'm winning but it's not perfectly calibrated so I can either
take my numbers and regress them to market which just shrinks all your edges so that my
10% edge becomes 5% and there you go it's coral it's it's it's uh it's calibrated well
or I can look deeper into my model and see okay well what what factor or what combination of factors
let me do think I had a 10% edge on this set of games when I really only add a 5% edge
and kind of fix the the root of the problem either way works now we you talked about earlier you alluded
to like whenever there's a paradigm shift in sports and the the game fundamentally changes
when you're going back and pulling old data you almost have to make sure that your point of
reference is relevant to how the game is played now right whereas it's not skewed by a game that
was completely different maybe even as as as little as five years ago and as much as 10 20 30 years
yeah and and even even if the game itself hasn't changed and you're building a model that uses
the market number as one of your inputs which is generally a good idea to do then your implicit
assumption is the market is as sharp now as it was five years ago and it usually isn't the market
sharpens over time that would be one thing that would be due to believe you have an edge it's
greater than what you actually do so so you're right you know you anything could change and yeah
you say you have to make sure that nothing changes between your your data set and and the
the set of games you're going to be betting into okay well that's great but what if something did
change what do you what do you do about it where you can just kind of throw your model out and move
on to do something else with your life or you can try and think of what kind of adjustment what
kind of change and method how how how you might reframe the problem to minimize the impact of
that difficulty because if you're having a hard time building a model because teams are shooting
more threes than they have before chances are other people are going to have the same difficulty
building their models so it may actually be an opportunity that if you can find a creative way
to overcome that obstacle you might be able to to carve out some space for yourself in the market
where where you're doing this kind of better than other people are how often is with a model are you
taking like a subjective approach to tweaking and refining it and like how often are you doing that
yeah there's there's no magic answer for that it depends on on your model it depends on your
subject matter like you you ask yourself is is there any reason why the model I built three years ago
might not be valid today and if the answer is yes it's a good idea to go back and at least run
a couple of checks if not rebuild the entire model but if you're modeling something like you know
a correlation between two things where okay either of the things might change over time but the
correlation should be pretty static then you can say okay this is this is not something that I have
to go back and and rebuild every six months or every year because I'm going to get pretty much
the same answer as as I always got so it's it's very domain dependent very subject matter dependent
again it goes back to asking yourself what are the assumptions both explicit and implicit in my
model and how likely is it that they have either become wrong over time or changed over time
and that's the kind of thing that should make you take a second look I mean I'll obviously if
your results start to deteriorate as well that that could be just a random run of bad luck or it
could mean that your your model is starting to deteriorate and that's a a whole thing in and
of itself and then something that I really enjoy as well is what I call small data where okay your
sample size is too small to be unequivocally unequivocally relied upon but it's too big to be
completely ignored and there's a whole world in between ignoring a data set and relying
completely on it and that world is is is kind of where a lot of the stuff I do lifts okay I love
that I think I think I think I have a better understanding of how to build a model and I hope the
listeners do as well I want to move into a fun aspect of the podcast everybody loves a little
word association that's where I give you a name and you tell me with one one word or one phrase
what you think of that oh I'm intent this was my intention because you know what I don't know
what it is but you are one of the few people in gambling Twitter space that are like a magnet
for drama it's unbelievable I'm trying not to be but it happens sometimes I don't shy away from it
when it happens but I don't see your body on purpose you're punchy you're you're you're you're
willing to take the gloves off and fight I've got a couple curve balls and loaded up in the deck
here boy all right let's let's let's get to it let's go for all right Billy Walters Billy Walters
I am looking forward to meeting him at Betpash this year I was super excited to find out he would
be he would be coming actually don't know all that much about him other than kind of what what
he's been in the news for and you know he's obviously one of the greatest if not the greatest sports
betters of all time I look forward to shaking his hand taking a selfie with him whatever whatever
they let us do at Betpash in August spanky oh spanky up spanky he's great he and I
disagree on a couple of things but it's always friendly with him when I when I first met him in
person at Betpash last year he actually lifted me up off the floor give me a hug I have nothing but
respect and love for spanky you know he he is he's outspoken like I am sometimes but he is someone
I like him because he has a very different perspective on things than I do he and I have very
different skillsets very different strengths very different weaknesses but you know in a lot of
ways it that's kind of the best friend or associate or partner to have is somebody who sort of
sees things in a different way and is good at what you're bad at so I I'm a huge fan of spanky
Simon Hunter Simon Hunter you're gonna have to you're not to remind me who he is I'm not sure if
you know but he has a really well dominating syndicate that he's a part of um if you if you're
I don't have to pass on that was this one because I don't even know who he is okay that's
listen if you don't know who Chris guy is and Simon Hunter I'm sorry Captain Jack
Captain Jack is awesome um again met him for the first time in person at at that bash um had dinner
with him and just the stories that he has again he's been around for for a long time he's done it
in the sports betting world he's done it in the casino AP world um he has just he is just such a
wealth of not only knowledge uh and skills but great stories you know he's he's the kind of guy
I would definitely recommend having a beer with uh if if you can how about this guy he's seems
to be a beacon of um some controversy orange sharp again like I know who he is but I'm not all
that familiar with the controversies surrounding him you know he's some kind of NFL guy right
doesn't he work for does he work for teams or consult with the teams or something
he sells pigs he does media he he he works for NFL teams yeah I don't I don't have a much of an
opinion of him good batter or other was all right Jeff Benson Jeff Benson is one guy that I
didn't get a chance to meet uh last year I hope I get to meet him this year and again he and I
disagree on a couple of things with regard to the economic to the sports book industry and
you know how to how to run a sports book but he is someone who cares about what he does
and if everyone in the world had the same passion for their job that Jeff Benson has for his
the world would be a much better place Steve Fezick Steve Fezick blocked me on Twitter um because
I made a comment about uh RJ Bell and and pregame which I believe he's still involved with
and so from everything I know about Fezick he is a fairly sharp guy knowledgeable guy again been around
forever I don't know why he's made the decision not only to sell pigs but to associate himself
with the uh goofballs over at pregame.com um who have been just exposed every which way you know
articles and Twitter and everything um I think he takes a lot of of heat I think he deserves some
of it um but if if I had the ability he has I would not have made the career choices that he does
but to each his own I don't I don't hate him or anything he just selling pigs as a whole is it
just uh um it's a it's a it's a tough way to make a living and it's just not something that
that I would recommend getting involved with or alabab will gears um he's another one that I
I'm not so in tune with the controversy of you know his backstory and it was true was it false and
you know I again I'm neutral on him I think he's a he's a sharp guy he's done a lot of great things
in his life um he's running a soccer team I think in Spain now which is cool like you know go
go do your thing make your money follow your dreams um you know I don't have a I don't have a
a strong positive or negative opinion on him last one Joey Kenish I thought you were going to say
digs no I don't want I don't want to be the wrath of digs for you right now let's let's avoid that
for a while he he I don't think he's clued into a new 90 degrees host in play I don't I don't want
to get put into the floor um Joey Kenish he's a funny guy he's not a tanner like I'm not sure
how like I think he probably makes money at this I'm not sure exactly how you know he he tries to
be funny sometimes it sometimes he succeeds sometimes he doesn't um he's had some interactions
with some friends of mine that have been less than great I'm not going to get into that I'll leave it
at that um but at the end of the day you know he's just he's a guy trying to build a following by
being entertaining on on the internet and that that is what it is all right I I don't think I got
you into too much trouble but if this episode drops on Monday and Tuesday I see you in a bunch
of Twitter fights I'm gonna apologize to you for getting you under under the heat right here
that won't necessarily be your fault or anything to do with this if that happens it's just just
just just a just a regular Monday for me awesome uh plus EV analytics I want to thank you for doing
this I know I've I've reached out to you and I wanted to do this interview for a while and I do
appreciate your time tonight that's been great thanks a lot Georgia County hey that's it for me
another edition of 90 degrees is in the books I want to thank my guest plus EV analytics
actuary gambler instructor of the art of sports betting analytics I want to thank the sponsors
of this podcast pinnacle and bet stamp and my producer Jason Cooper thanks for listening do me
a favor before you go like the content subscribe share and comment we'll be back next week with
another guest on the 90 degrees podcast where we give an inside looking to the sports betting
industry that's it for me hope you enjoyed until next time