Gridiron AI: AI is Revolutionizing Fantasy Sports Predictions

Join us as we sit down with Andy Troiano, co-founder of Gridiron AI, to discuss how they are revolutionizing fantasy football and sports betting through artificial intelligence. Gridiron AI uses advanced machine learning models to generate unbiased weekly player projections and betting predictions. By eliminating human bias and emotions from the equation, Gridiron AI provides a purely data-driven outlook on upcoming NFL games and fantasy performances. Whether you're a casual fan looking to gain an edge in your league or an experienced sports bettor, Gridiron AI has the insights and tools to help you win. Tune in to learn more about Gridiron AI's founding story, their AI modeling techniques, and how they aim to compete with Vegas oddsmakers using the power of data.

GRIDIRON.AI
https://gridironai.com/
https://discord.gg/Dm2XXKcZZD
https://twitter.com/gridironai
https://twitter.com/andytroiano

FRY-AI.COM
https://www.fry-ai.com/subscribe
https://twitter.com/lazukars
https://twitter.com/thefryai

Creators & Guests

Host

Ryan Lazuka

The lighthearted Artificial intelligence Journalist. Building the easiest to read AI Email Newsletter Daily Twitter Threads about AI

What is Behind The Bots?

Join us as we delve into the fascinating world of Artificial Intelligence (AI) by interviewing the brightest minds and exploring cutting-edge projects. From innovative ideas to groundbreaking individuals, we're here to uncover the latest developments and thought-provoking discussions in the AI space.

baseball is um kind of where a lot of stuff really started the dad has always been there with the NFL it's there is a
0:20
lot of data but baseball's been at like the like the front end of sports data for a very long time you put in so much
0:27
time to think about a simple every keystroke you put in to code something and this just spits out everything you
0:33
need it really gives us the ability to give like a whole distribution of potential outcome yeah it's such a great year last year it seems like if you're
0:39
pulling just from the stats he would be the number one guy right so why is he not that I'm going down what is it that
0:44
the algorithm is doing to to make it a little different so my background is mainly data science
0:50
and analytics um primarily in like the small business lending space um so I started doing
0:56
um like machine learning and focusing on data like in 2014 I kind of saw like the the wave was coming
1:03
um and and kind of got in a position where I could you know I could I could position myself into uh into a job doing
1:09
it and I actually started with baseball analytics um because there's a ton of good resources to learn how to one code into
1:15
how to do data science um in baseball just kind of on the side and I was able to Parlay that into an
1:21
actual career um just building models like doing a bunch of different types of analysis
1:27
um you know so my passion is both football and baseball I know we're great iron AI but you know uh baseball is um
1:33
kind of where a lot of stuff really started because the data's always been there with the NFL it's there is a lot of data but baseball's been at like the
1:40
like the front end of sports data for for a very long time so are you a big sports fan do you have
1:46
a favorite team yeah I mean obviously like the eagles this is why I wore the hat and plus my hair is pretty long but yeah I'm big uh
1:52
Big Philly sports fan so it's been a fun couple years with Jalen hurts um you know the Phillies are hit or miss you
1:58
know year to year um hopefully they get hot and make another run but like big eagle sand yeah
2:05
both teams fell just short I know at least it was exciting well we missed we
2:12
missed the uh we missed the perfect storm of the Sixers got to the finals and lost it would have been like the
2:17
first time ever the city had every major sports team lose in the finals so I'm pretty sure like the um the MLS team the
2:23
union loss in the in the MLS cup two are in the final so it would have been like every single sport they kind of get
2:29
close to put them but didn't do it but the Sixers kind of faded out as as they always do but yeah that's kind of funny
2:36
um all right so uh that's pretty cool uh so one thing I wanted to know is you
2:41
know I I had a chance to look at Gridiron AI I thought it was really cool but can you just give us a run down what
2:49
is Gridiron AI in a simple way yeah so we
2:54
um use artificial intelligence and machine learning to really do two different things so one is Project
2:59
player stats and then also um we recently got into game outcomes so
3:05
betting so like betting models with like who's going to win and spread who's going to win money line and then
3:10
ultimately the game scores um and one of the things that we do is we don't we don't add personal bias to
3:15
it so what the model spits out is what you see on the site there's very few exceptions where we actually like touch
3:21
anything that come out of kind of the the back end and all the kind of all the different models that we have run on a
3:27
week-to-week basis um what what this allows you to do really is to get a like an unbiased
3:32
opinion on where you think something's going to be because when it comes to Sport and one of the reasons why we we
3:40
took this approach is um like Zach um one of the co-founders just randomly posted on LinkedIn like I'm just sick of
3:46
looking at people's opinions on projections I want to start a company that you know takes that that bias out
3:52
and I actually found him on I saw the post on LinkedIn randomly I was like hey I'm in I can I'm down to do all the data
3:58
science work um and then uh you know that's kind of how it started another guy named Liam
4:03
joined in and then we've kind of just been trying to tackle that problem um of trying to remove bias from from
4:09
sports betting and from um from from fantasy football very cool so that's kind of what
4:15
inspired you to create this project is is you're tired of hearing everybody's opinions oh this team's gonna win I'm
4:21
you know I'm betting whatever on this team or or whatever you just kind of wanted an object it purely statistical
4:27
analytic report of this stuff is that correct yeah for the most part like you
4:33
know the thing about fantasy football is get it's really hard to remove like your
4:39
personal opinion and feelings about players really in any type of fantasy sports that's that's that's true it's
4:44
like you know you draft a guy he has a bad ears it could just be a bad year but you're probably never gonna draft them
4:50
again even though the best outcome for your team might be draft that player again because it's just a one-off year
4:56
um because the other the other thing about football that makes it interesting and one of the reasons why like like we haven't really expanded out into like
5:02
into baseball or any of these other sports is like your samples are very small which is also some of the
5:08
challenges it's like you know the NFL season only has I think like 272 games
5:13
something like that which is a pretty small sample each team only has 17. if you look at baseball baseball's got
5:19
about like 4 000 games more or less so you do the math like in baseball it's
5:24
kind of more normalized over a full season and football it's not like that it's um you're because every game means
5:31
so much you're how week to week performance week performance for a player really impacts like your decision
5:38
making right right that makes sense so tell us like so the average person goes on to
5:44
Gridiron AI is it going to help them with predicting the outcomes of games is it going to
5:49
help them with fantasy players you know for their Fantasy Football Draft coming up or maybe it's happening right now
5:54
what is it going to help them do like the average sports fan yeah both and it really just depends on
6:01
what you're into if you're into sports betting yeah we'll have when the season starts um like the week to week
6:06
projections which we also track on NFL pick watch so if you're only into sports betting which is a lot of people that
6:11
only care about sports betting we'll give you our readout on kind of what we think is going to happen in these games
6:16
and then if you're like a DFS player or a weekly fantasy player we'll have the fantasy projections every week as well
6:23
where you're able to kind of like after you're done with your draft in the sake of like a weekly League you can save
6:28
your team and we'll we'll try to tell you who on your team you should start on a given week
6:33
um we don't really have any right now integrate into league so we can't pull in like oh who's on the waiver wire and
6:39
who should you pick up that's kind of like in the future where we're trying to take it but right now it's mainly just
6:44
we'll give you our projections and you need you can use them as you know as you see fit very cool so so tell me like I'm
6:51
a sports fan I'm a big sports fan uh so I follow like all the games all sports football is my favorite I like college
6:57
football I'm at the University of Tennessee right now as we speak so I think go Vols but anyways
7:03
um so tell me like is this is this limited to pro football right now or is
7:09
it like uh can you do college sports or something like that or how does that work yeah it's like right now we're just
7:16
we're just um we're just the NFL um and you know I think that you know
7:21
there's a point where we get big enough where we can where we can start expanding out into other sports and logically like college football would be
7:27
a pretty easy um move from week to week there's more games there's more players it's really the same structure of data
7:34
um there's just a lot of there's a lot of startup costs um in the sense of like it takes a lot
7:40
of time to get to get models that one work well and two like the infrastructure behind them to make sure
7:45
that they all work every week and troubleshooting it so I think that there is there is like a point where we would
7:50
expand out to other sports but right now the main focus is just the NFL which is
7:56
also the biggest sport in the United States right now um so um yeah so that's where we're at at this
8:02
point yeah it's a great time because NFL season is here uh so what do you see as
8:08
uh like what do you see like I look at other
8:13
sports predictors right what is the AI component like what how does AI come
8:18
into play to help you with your projections here like how does some of that back end of tech work I guess yeah
8:23
so um so a lot a lot of what we do is um we model we have a a couple different
8:30
we have a couple different models that ultimately feed our final projections that go on the site and the way that we
8:37
try to leverage AI is to build um and model probability distributions
8:42
so we basically have have the AI give us 500 outcomes for a given game for a
8:49
given player for all of their stats and then we use that to kind of roll it up into what you see on the site and we
8:56
don't do that as much for the betting side but for the player side it really gives us the ability to give like a
9:01
whole distribution of potential outcomes which allow us to say like all right like this player has some upside they
9:07
have some downside and um kind of what are those outcomes because in the sense
9:12
of like you know for most most projections that you'll see like on ESPN CBS these other sites like I guess
9:19
ESPN's gotten into like giving you like a boom bust now but we've been doing that since we started they just got into
9:25
that because there is a range of outcomes in in all sports for every player and just looking at what somebody
9:31
is going to do in a given game for like for their average is not like it's gonna
9:37
it's not gonna be that it's going to be either more or less it's good to see what what is the upside and the downside
9:43
for a given player and and why that's important especially if you think about and sense of like like football if
9:50
you're let's say it's a Monday and you need like a Hail Mary you can go and look at our upside projections and be
9:56
like who on my team has the most who has like the highest ceiling and try and play that person over somebody who may
10:02
have a lower ceiling but higher floor because you just need you need to gamble right right that makes sense so how long
10:08
have you guys been around so we we started um late 2015 like probably Midway
10:15
through the season it's kind of when like um the three original co-founders like me this guy named Zach and Liam
10:20
kind of all just found the same LinkedIn post by Zach and we're just like all right let's do this and then we spent
10:25
basically the whole off season and most and like the first couple weeks of the 2016 season getting everything set up I
10:33
think we launched actually in like a Google doc originally it's like hey we don't really have a site yet but like
10:38
here's our projections Reddit go go crazy on them and it was a good learning
10:43
experience so right now is it a paid service yeah so right now we um during the season we charge five dollars a week
10:50
and we only run that through like early December and we just like give it away
10:56
you know in the preseason and for kind of the postseason because at that point like if you're a weekly player and you
11:01
don't bet like there's really no value on the site because you're out of the playoffs or whatever but maybe you want to do some DFS in the playoffs so we we
11:08
kind of like we turn we shut down the billing during the off season because we only really provide a service during the
11:13
NFL season and the meaningful part of the NFL season really ends in December for a lot of people now we if we expand
11:20
out into like DFS more we made change that or have a subscriptions like just DFS
11:25
um but for now we kind of keep it pretty pretty loose on like you know when we when we just quote unquote turn off
11:30
billing for the year um just because we don't we don't want to charge people we're not getting value from us like it's not we're not here to
11:36
take your money we're here to help you win and just billing you when you're not playing isn't isn't a good way to do that yeah no absolutely that's not bad
11:43
at all five bucks a week I mean people people spend seven dollars on coffee a day so that's no big deal uh so um
11:50
especially for sports fans you know uh so how many users do you have right now do you track that at all or do you do
11:56
any Data Tracking yeah so we're up to uh close to like 9 000 total users on the site
12:03
um that's not the same as our paying so we actually have two services so we do have like a free service um where you can just go make an account
12:09
and you just get like the top I think we've been talking about what the final
12:14
number is but I think you get the top 10 at each position each week so you can still get something for free and then if
12:19
you want to unlock everything it's when you we we have the five dollars a month the other thing you get with a
12:25
subscription is you get to customize your league settings so one of the reasons why we project out stats like
12:31
not just a final step but the counting stats too I think every site really does this as well but it allows you to put in
12:36
like your custom scoring because everybody most some people have crazy scoring rules and we we try to capture
12:42
that by you just you create your team you can go to settings and you can be like all right if you get 20 points a touchdown for some reason like you can
12:48
go and put all that in and it'll recalculate our like you know our life distributions to what your to what your
12:55
settings are in your league right so like if if your algorithms come up with a guy who let's say gets a lot of
13:02
receptions but maybe doesn't get a lot of yards but they're in a PPR League where you get points per reception then
13:08
he might you know be higher on the list than he would in some other or standard lead or something like that yeah exactly
13:14
right and we have like pre-configured settings so when you go make an account we have like PPR standard vandals and
13:21
DraftKings like if you just kind of keep it standard you don't even have to make a custom team you can just go right to
13:26
those presets and then it'll have them recalculated for those scoring
13:31
um for those different scoring settings just by default because I would say probably 80 of people probably just do
13:37
PPR or standard scoring and then you have the GFS players that you know either use fanduels or DraftKings how
13:44
accurate are your algorithms usually I mean I know like like do they are they a
13:49
lot different than you see on ESPN or something like that or uh how does that how does you data differentiate from
13:55
what they use say on like ESPN fantasy or something like that yeah so that's that's when it gets like that's why it's
14:02
hard to talk to it's like I don't know how any of those sites build their projections or where they get their data from but we typically
14:08
um we're pretty good I think that like I don't have it off the top of my head but we're within like four or five points
14:13
for every single player every single week and it's really just like it's high variance like and that's the challenge
14:19
it's like and and the the data sources that we can track publicly we post on
14:25
Twitter every week of like all right like here's our average error and then here's like I think there's like four or
14:30
five sites that like actually just put their data out there for people to use which is great and we we can kind of
14:36
spot check what we're doing against kind of the the easily available data just to make sure we're doing well like nfl.com
14:42
publishes those where we beat them every week um but it really it varies a lot like
14:47
you know if you there's a random wide receiver that scores 30 points like nobody's going to have them at that like
14:52
you know wide receiver three but in general um um we I mean I've been using it
14:58
non-stop since since I've since we've started it and I've done pretty well in fantasy um where you see a lot of variants I
15:04
think that like part of part of the value that we also provide is like we're a unique way like our projections aren't
15:11
really in line with everybody else and the way that we use that is we um and Nick from our site he does a
15:18
bunch of like podcasts and he'll usually post a bunch of stuff what we try and do is see what data is available from like
15:24
fantasy pros and all these other sites and be and try to figure out where we are different from them and where we're
15:30
massively different because if you think about it like like the top five or six players everybody's gonna have in the top five
15:36
or six it's really after you get past that is where you you really try to find Value like a good example is like Justin
15:43
Jefferson it's like we don't have him as wide receiver one which like everybody's like why it's like look man like I don't
15:49
know like I can't I can't really tell you exactly why but Justin Jefferson had an amazing year last year he hit 1800
15:55
receiving yards like we just don't think he's gonna our models just don't think he's gonna have that again and and some
16:01
of that is because like that's not a common occurrence so if you're expecting him to do that again yeah he might do it
16:06
and if he does hey great on him like he's an amazing player but like the the models don't lie like we don't we think
16:13
it's gonna come back down to earth and the other thing about Justin Jefferson last year is he didn't really score a lot of touchdowns like most of the point
16:20
difference between like where he we have him now and where he was last year just
16:25
come from a regression back towards a like a normal Elite wide receiver
16:30
um projection because I mean like we have Adams and there's other guys up there too and the difference between
16:36
them is like three or four points like it's it's not that much but you know we you know we just don't have Jefferson
16:42
eye and people are just asking us why I'm like look man should we just we just trust the system like and we're not
16:48
saying don't draft Jefferson but like you know if you're if you think about it how you translate that to the draft it's
16:53
like if if you're if you're sitting there and you know you can get somebody else well I mean Jefferson's probably a
17:00
bad example because it's going to go first overall or close to it but when you get to these guys like later on
17:05
after the first round that's really where you can find the values because we you can get these guys later in the
17:10
draft that we have higher and we've hit on them in the we we hit on them every year like there's always three or four
17:16
guys like last year we were high on Josh Jacobs like he had a really great fantasy or nobody else had him as high
17:21
as we did and it's just like it's what the model says like we just go with it you know and if it works like it's it's
17:27
part of the part of the system is why we don't that's why we don't over analyze our projections we spot check them to
17:33
make sure that like injuries or things like that don't negatively impact what we're trying what we're spitting out
17:38
because that is one of the challenges that we have and we may get to that in a minute but it is like injuries and
17:44
things like that are hard for an AI to really grasp for for us to model out so we have to do some like spot checking
17:51
just to make sure that nothing gets wacky there but that's like that's the extent and there's not even like looking at Players it's like look if um if Alvin
17:58
Kamara is out all year and we think he's gonna play in 16 games like are you suspended for three games we can't we
18:05
can't really tell the we can't really feed that into the algorithm in a way where it learns that effectively because
18:10
there's just not a lot of examples of it right like you're not going to have
18:16
there's not thousands of players every year that have suspensions that are going to play unless games so you can't
18:22
you can't teach it that right so you'd say that's kind of like like one of the limits that you're kind of working on or
18:28
trying to to find ways around like injuries do like obviously players go
18:34
off that's just fantasy football or that's just football in general yeah you know a good team that's supposed to win
18:39
maybe has a couple turnovers and they end up losing the game and you're like oh you know I mean stuff happens that's
18:46
that's sport so you can't control that but in general it seems like it does really well uh do you see any other
18:52
risks or limitations besides like the you know the the injuries or maybe like
18:58
the the Wild Card players the receiver that gets one reception all year and then he goes for 30 points I mean that's
19:04
pretty much it and like those are like those are that's a problem for a different day like how do you actually
19:09
how do you how do you actually how can you come up with a model or any type of machine
19:15
learning algorithm or AI that that will be like and this is really the key to being super profitable at DFS is finding
19:22
these guys that like have a higher than normal chance of catching a bomb like a
19:28
60-yard touchdown because a lot of these receivers that have like abnormal weeks they just call like a 60 yard pass so
19:35
how do you actually effectively identify those and project them out that's a different problem like if we got to a
19:40
point where we had a model that did that well like that wouldn't be part of like the normal projections there'd be like
19:46
another part of the site that would be like Boomer bus probabilities or like not really like upside probabilities and
19:54
like why we think that um and they'd be like hey these are guys who want to take a risk and like DFS or
20:00
something else like these are the players who take a risk on because they're either do or the model just thing said hey they may catch a 6cr pass
20:06
for a touchdown Which is gonna swing you one way or the other in like a lot of different type of um like type of like
20:13
competitions so you said you're working on that right now you're working on a way to get those sort of summaries of
20:18
why they think that yeah I mean it's not it's not like a problem we think we're gonna solve but it is like I mean and
20:24
that's part of like with uh with the rise of like large language models and like like newer techniques like
20:30
that's where that's where I think some of this stuff is gonna go it's just gonna be hard
20:36
um and a lot is it's like we're not like a research company like we're not like Google or we don't have like a massive
20:41
amount of resources just to throw it a problem and say look just figure it out or don't but it's fine um so live it is
20:47
just like what we try to do is Tinker like experiment with things when we can and then wait until we feel good about
20:54
to then go and like release it so like if it's like so the reason why we got into like betting is we just were doing
21:00
it ourselves not even trying to put it on the site just because we had all the data and we're like hey we all we all
21:05
throw some money we all sprinkle money on NFL games like we have the data we have the time and then once we were like
21:11
hey this actually works and like these projections are good we put it in the site because that's we think that people
21:17
can get value from that um it'd be the same type thing for this if we got to a point where like we could try to identify
21:23
um we could identify with some level of accuracy like high high point share players or boom players then we would we
21:32
would probably keep it ourselves for a while just to make sure that it works and then we would we would bring it into the app can you tell us a little bit
21:39
more about the tech like what kind of algorithms are being used where are you pulling like what do you I know we
21:44
hinted at it earlier but where are you pulling the data from are you pulling it from like the the history of their stats
21:49
are you pulling it like like how are you pulling those projections like for example Justin Jefferson you mentioned
21:55
him earlier he had such a great year last year it seems like if you're pulling just from the stats he would be
22:00
the number one guy right so why is he not the number one guy what is it that the algorithm is doing to to make it a
22:06
little different yeah so a lot of a lot of the models and like there are we have
22:11
we have a bunch of different models that run each week um for like our yearly projections we try to use
22:18
um multiple years uh for you know for players that have it multiple years of
22:24
um of their of their past performance to then try to come up with what their performance is right now and then we try
22:30
to layer in information about the teams that they're on because I think that is an important aspect that you have to
22:36
capture is like what's their o-line going to be like this year what's their quarterback like going to be this year because there is like at least
22:42
anecdotally like if you have a good wide receiver with a really bad quarterback there is
22:47
there is a lower like they're going to score less points because you can't get the ball of the players they're not going to score it they're not going to
22:53
be able to score so a lot of the models are are are using multiple years of data
23:00
averaged out like a bunch of different ways to slice and dice that we have like over like five or 600 features that we
23:06
feed into the model for it to then go and try and come up with what these outcomes are now I can't tell you if it uses all of them but we try and give it
23:12
as much information as possible for it to come up with the with the with the most accurate projection uh projections
23:19
and then what you're able to do like week to week it's a little bit different because you can we there are models that
23:25
use like sequences so instead of you having to feed it a bunch of data like like last game like last game stats or
23:31
last years in the year two years ago three years ago what you can do is just feed it that raw information in a
23:37
sequence so it learns like okay it knows that you know this array of data the last one was four games ago then three
23:44
and then two then one I mean you can we can use that to then help it extract what it thinks it's going to do in the
23:49
next couple games and then we can use that to then ultimately come up with the with the predict with the predictions
23:56
um yeah so that's a little bit I know that was like a lot of detail I could actually get into like some more uh on
24:01
that if you if you guys want to I don't know how technical your audience is or what they want to hear but um yeah it's an interesting problem to
24:07
solve because there's a million different ways to do it and they usually the answer is do them all and average them together which is what we try to do
24:13
as much as we can um I did have a follow-up so you mentioned like features you're giving
24:18
them features what do you mean by by that like what's an example of a feature features would be the stats it'd be like their counting stats it'd be like there
24:24
are rushing yards receiving yards ariars share of area it's like basically anything you can think of that the that
24:30
is available from from like the NFL like data wise we try to use and then what's
24:36
and then the the other thing that we try to do too is you have to like for all all of this you have to apply like
24:42
domain knowledge so there are there are really good articles on like Pro Football Focus and like from these
24:48
fantasy experts they do actually have pretty good analysis of like all right well we look for these things and like
24:53
this seems to lead to good like good outcomes we can take that ancillary
24:59
information build build those features from that and then use that as another
25:04
data point with like in our models which I think is really like not really the secret sauce but that is a in general
25:10
for machine learning that is an effective way to increase performance in your models awesome so are you like are
25:16
you using uh I'm Ryan by the way I um co-founded fry AI with Hunter
25:23
um but I won't get into too much details so since we're in the middle of the podcast here but um what uh like are you using an alarm
25:29
in the background forgive me if I ask any questions Hunter already has like what llm are you using to to analyze the
25:35
data and then like what's if you go over your Tech stack a little bit we do our audiences anyone from the you know they
25:41
just heard about AI yesterday to someone that has a lot of tech background um so we have a lot a lot of different uh uh
25:48
people on our demographic um but if you just go over those those questions that'd be great yeah so we we
25:53
haven't we haven't got into using like large language models we use large language models to start with it would
26:00
be to generate like player profiles and summaries from our data more so like
26:06
less so for the projections themselves um we we played around a little bit with
26:11
trying to get like a summary from game logs out of like large language models and it's not it's not as accurate as
26:19
just getting the raw stats from like the NFL like the NFL fastr package like these these pre-built open source tools
26:25
that people have that already do that are better than L like large language models right now for this
26:33
um I think that you'll see large language models be used a lot more to to do like the summaries and like and
26:40
answer questions about your like about the data so instead of you having to go say hey who should I pick up they'll
26:46
like and you have to go like well you save you having to figure out who to pick up on their own you can just ask them like hey who's the best player to
26:52
pick up and have it output it but I think from like a modeling perspective we're not there
26:57
um we're not we're not there in a way that makes it as any value right now gotcha so what's the difference between
27:04
like your site because you're you you're marketed as AI like what is the AI
27:09
component behind it right now like compared to the other thousands of sites out there
27:15
that's what I asked too oh sorry all right no it's no it's fine no
27:23
no come up with their projections but we use a lot of deep learning models um we use deep learning really to come
27:28
up with our projections and then we we don't just model stats we actually model
27:34
probability distributions so what we do is we we use like tensorflow probability to come up with
27:40
you know with I think it's like 500 to a thousand distributions for each player each week
27:47
and then we aggregate those uh an average them out to come up with what our ultimate projections are and then
27:54
what we end up doing is we we don't just use one model so like one of the other things that that we did I'm not gonna
28:00
get into detail because it is a little bit of Secret Sauce but we do we do model out a bunch of other things that
28:05
we then feed into the models that you see week to week and that's really when we started seeing the most performance
28:10
gains wasn't just like taking the data we have and just kind of throwing it into you know an auto ml package and
28:16
have it spit out projections what we end up doing is we have multi we have a bunch of different models that run that
28:22
then ultimately feed that final deep learning algorithm that comes up with what you see on the site like the
28:27
projections you see on the site and we've we've got a lot of lift from that so you're kind of like playing possibilities like if I'm understanding
28:34
this correctly like you're we'll keep to our Justin Jefferson example because you know we've been talking about him but
28:41
let's say like Justin Jefferson for example has you know like has like since like some
28:48
probability of having like 200 yards in a game then maybe he has like 80 yards maybe he has some probability of having
28:54
10 yards maybe he has some probability of having zero zero points and you run that through like 500 times
29:00
then you take that aggregate you get a high you get a low and then that's what that's what it posts something from that
29:06
is that is that accurate yeah yeah more or less and like what we haven't done yet and this is something that that we
29:12
we very well could do is we could even we could eventually get to a point where you can just put in like we can give you
29:18
like those range of outcomes and say like he has a 30 chance of scoring over 50 because out of all of all the
29:24
distributions that that like the the that we got from the output like you know if we had a hundred like 20 of them
29:31
he had over 20 yards which I think goes into like player props and stuff like that but we don't need to go down down
29:37
not rabbit hole because we haven't done that on the site yet um because that is really not available
29:42
um but there is yeah there is a lot of a lot of cool stuff you can do with with the tools that are available it seems
29:48
like so you have like some things in the oven it seems like some directions you could take this you know like having
29:53
like maybe like a chat bot or something like that so like you could ask it like
29:58
personalized advice like should I start Josh Jacobs this week or should I start Nick Chubb or something I mean he'd
30:04
probably start both of them but let's say Yes um you know it would be able to say well
30:10
Josh Jacobs is playing like give it some sort of like response based on that or like there
30:16
seems like there's some different things that you can do like you said player props like you know let's say Justin
30:22
Jefferson's catch line is like six and a half like over under you know you could do like well he's got a 70 chance to go
30:31
over something like that um so do you see like these things like is this a direction you guys want to go
30:37
or do you have any other things like you're you're working on for the future for like the next steps of your development yeah so that's a good
30:44
question I mean like we're not we're not trying to be everything to everybody so like there are a lot of really good
30:50
sites that have really great content so we're not trying to really do a ton of content like I mean we we we do have
30:56
like we do try and do like weekly Recaps and like Nick is really good at just
31:02
jumping on and like he live streams his draft like we're gonna do stuff like that because we enjoy doing it but we're not trying to be like huge content guys
31:10
um I think that like if we I think that like next up is really just trying to get Integrations into different leagues
31:16
so people can automatically pull in their settings like I think very like tactical things
31:22
um and then like a lot of it really just comes down to like a meta level discussion like what is fantasy football like do you want like do you really want
31:28
some like like a large language model to tell you exactly who to play or do you want to like use the data and make a
31:34
call yourself um I think that there's really like people that want like both sides of
31:40
those um but I think that for now it's really just trying to trying to improve usability on the site because we are still small and relatively new so like
31:47
we only make incremental improvements every year but we're actually we're getting to a point where like we can actually start making more changes
31:53
quicker at that point I think like we don't really have a road map yet like we haven't even got to that like once we
31:59
get into the season everything stops and we don't really make any changes because you don't like there's too much stuff
32:04
going on in football you know every week There's Something New there's no real downtime
32:09
um but I think that you know I don't have a good answer for you outside of like there there's Endless Possibilities yeah I think I think
32:16
that's a good answer like you have things you could do but you're trying to get that Foundation solidified and
32:21
really do that well right now I think that's that's a good thing to do it's a good thing to start because you don't want to run a million miles a minute and
32:27
then you get back and you're like we're trying we're spreading ourselves too thin or something like that yeah yeah and like in part and part of the reason
32:34
I we've I think had the success in like why we've been around and have like we we were granting out like no users for a
32:41
while it's just like between all of us that are a part of it like we we all love it and we all we all give
32:47
a lot of attention to it but we just really focus on the things that we all feel is important and there's really not a lot of pressure like we don't we don't
32:53
we don't have like investors that are pressuring like growth or anything it's really just like a small group of us that started it that just are kind of
33:00
like treating this thing that's like our baby and we can move with the speed that we want which is probably not like what
33:06
other people would do in our situation but it's kind of just the reality of like where we are where we're all at in
33:11
our lives like we don't all of us do this as like a nights and weekends so it's not like we have we all have
33:16
full-time paying jobs that you know require a lot of time and attention as well and then this is really just like
33:22
right after that what do we do we we dive into kind of where we can try to add the most value to our user base
33:30
um and a lot of it is like you know oh yeah that's really it I mean there's there's um
33:35
there's a lot of sites that do a lot of things and we're just we just can't be that that's funny that you say uh like
33:41
some people want to use it and some people don't right like some people they don't want to hear the statistics and I
33:48
think about like you know whenever people have like fantasy football teams and like somebody's girlfriends in there
33:53
like somebody's wife or like their mom like something like that right and they're always the ones that like do
33:59
super well or like with March this like they just like no people who don't watch
34:05
football at all or or basketball at all they do so well and it's like they just
34:10
pick it based on the mascot or the color or like whatever it's kind of funny how that works and that's kind of the beauty
34:16
of it is like it is very cool like to me like to see all the analytics and stuff like that but then there's also this
34:22
component where it's like some people are like eh I I don't want to do that I'm better off just winning you know so
34:28
it is kind of funny that way it's an awesome tool also for like you know jumping in on your family like you want
34:33
you you said it before like you it depends some people just want the data I want to make the choices of themselves
34:39
because they want to have fun with fantasy football but like our family league I want to use
34:44
it to crush my family you know like so uh but is that sort of like how you got
34:49
started as well like um did you get started in this just because you wanted to do it for yourself and for your
34:55
fantasy football league or for betting yourself and that's sort of what how this guy created yeah so like that I
35:01
think there and if I think I think the whole reason why like so it started with like a
35:07
LinkedIn post from a guy named Zach um I'm pretty sure he is LinkedIn post was like I want to beat all my friends
35:13
and family in fantasy football who wants to who wants to get in on it I was like I'm in um but yeah I mean the betting stuff
35:19
very much though like all the betting stuff we were just doing to try and because we all we all I mean Sports betting's huge and we all we all like
35:25
making pizza bets on games so that we very much just started doing that ourselves because you know who to pick
35:32
week to week it's so subjective like I I don't know everybody has their own opinion and my brothers love to give me
35:37
theirs and it's usually counted what the model says and it's like I don't know like I know you guys loves betting but
35:42
I'm gonna go with what the model says because that's not like it's not biased it's like I don't like betting on the
35:48
Dolphins like plus 20 but like if it says to do it I'm in but your brother is telling you it's your brother's telling
35:54
you it's a sure win you got about the house on it so you got to follow him yeah I know right yeah
35:59
um so yeah I very much just like we we started this just trying to trying to get better rankings and and we were
36:05
getting on ESPN and CBS which are notoriously bad according to Reddit yeah they're pretty bad
36:10
I get a call from my I get a call from my uncle every week and he's he's kind
36:15
of funny he's a big sports fan but he calls me every week you know like ESPN has this has this guy called the guru or
36:22
whatever it's like a guru it's what they call it and he uh like they give their projections each
36:28
week of how well I'm gonna give a little Summary of why they think the person's gonna do such thing and he always calls
36:33
me every week like oh the freaking Guru again like screwed me over you know like
36:38
that like they're always terrible they're always way off he's like how do they still have jobs blah blah blah you
36:45
know and so it's just it's just so funny like I I agree that the algorithms are not that or the uh whatever they use is
36:52
not that good so I think this really is an area where there could be a lot of improvement a lot of people could
36:57
benefit like I know like I play a lot of fantasy football and I read that stuff like a lot like
37:03
I like the analytics so I think I think it's really cool a lot of people do yeah and like I love like I do love to go on
37:09
ESPN and all these sites and read their content because I do I do like hearing what people I like to take the people on my team and
37:16
make sure that these people think that they're good like I know I dropped by team based on the model but like you do seek like validation that your choices
37:23
are the right ones I mean also like if you know we have nothing to do reading an article about fantasy football it's
37:28
like not not the worst not the worst thing to spend 10 15 minutes on um that's really the other thing about
37:34
this too that that's kind of that is challenging is like a lot of people use this stuff to validate their choices and
37:41
like their drafts and like you know if you're and one of the things that we do see is like hey I drive like you have
37:48
this guy really low everybody else has him really high why because this person has them on their team and they're like
37:54
upset because they take it very personally that like we don't we're not giving them like a an echo chamber that
38:02
that their team is really good at this player is going to be really good this week because when we first started we posted a lot of our projections on
38:08
Reddit and Reddit is brutal um like those like in the fantasy football sub race people would go in and they would
38:15
like be real mad at our projections like I was like we were personally attacking them I'm like guys like this isn't like
38:21
this is my opinion I'm not telling like this is just what the model says like it's not me it's the model like stop attacking me because you don't agree
38:27
with what what might what the model says like it's not not personally attacking your decision to draft this player or
38:34
start this player this week um it's it's interesting like do you ever get upset when they say the Eagles
38:40
are going to lose to get does it get your heart a little bit or you're like dude luckily luckily it didn't have the
38:45
Eagles losing many games this year it actually had them winning the Super Bowl too so I was pretty pumped about that oh okay it hasn't won in the Super Bowl
38:51
well I had him winning it had him winning last year when they lost um but yeah we haven't I haven't
38:58
I don't have I don't even have I don't we haven't even got to the point where we do like preseason rankings like or
39:03
like pre-season um like preseason like win losses for teams
39:09
um that's something that's like on my list of stuff to start looking into I think it's challenging because you don't
39:14
know like there's so many factors that come into play when you're looking at the whole team performance like if Taylor hurts goes down like the eagle
39:21
like whatever you said the Eagles is going to win you're losing that like that's not happening there's no way they're winning like 12 games or 10
39:28
games whatever it is with the backup quarterback it's hard to account for that right because I mean like if if Jalen hurts is in let's say the Eagles
39:34
are over 10 and a half wins Jalen hurts is out it's like okay we're going down to like six and a half wins or something
39:39
like that yeah I know right yeah I don't want the ire of my own my home team my home team fans I want some on the model
39:45
saying like the eagles the Eagles are going to win like eight games then post that and then have like my brother like
39:51
my dad call and be like yo why didn't the eagles and he wins like I don't want to deal with that like at least not yet
39:56
so I want to ask you like a part from your project this is just a fun question but uh what do you think about like the
40:03
direction of you're you're in the AI space so what do you think about the direction of AI more generally where do
40:08
you think it's going that's a good question I mean I think that I think that like I think that it's going in a direction of just like helping out like
40:15
helping paths get done easier um yeah like I'm a bit torn like I'm I'm
40:21
very much like a wait and see kind of guy where like you know I I'm pretty good about spotting Trends and things I
40:27
mean like I got into Data at the right time but like I don't really know where any of this stuff's actually gonna go
40:32
because I mean I just I just think on my like self-driving cars like that was hype like mid-2000s like a
40:39
2010-11 and this still has the material as I remember just reading articles being like oh we're really close to having fully autonomous vehicles and
40:46
like I think the conversations shifted there to just Highway only driving because it's more predictable so like I
40:52
think like generally like it like the AI space is complicated it's like the future is
40:59
really hard to fully to fully understand like I I see I say and I say that like I
41:04
think it it'll be good tools to help people in their like their jobs and like their passion projects because like I
41:11
use chat GPT a lot at work to help with like coding and like and it makes it does make a huge
41:17
difference like when you want to experiment with something new anytime you're reducing like the barrier of learning or the barrier of getting
41:24
started you're going to see a lot of benefits so I think that like AI is going to help reduce the barriers to a
41:30
lot of things and that's gonna that's gonna do a lot of good for a lot of people um because like you know it's hard to get started it's really hard for people
41:36
at least for me personally to get started on something new because I know the startup cost of having to like
41:42
especially when it comes like coding or like new models it's like getting there like it takes a lot of like trial and
41:48
error to get something started but when you have like AI if you have like chat GPT or these other like bar where they
41:55
can get you started and help you work through errors while learning I think that it makes everything like way more
42:01
approachable like I have examples in my job right now where like things I thought was gonna were gonna take me like three months or four months to do
42:07
myself I got done in like three weeks because I was able to use like chat GPT that like because I knew there were
42:12
going to be things I was going to run into but instead of having to like search stack Overflow or Google or just
42:18
like go back to like all the books I have on my desk I can just go ask like chat GPT or to help me with very
42:26
very tactical things and it makes it makes you iterate a lot quicker so I
42:31
would say that like that's if I think like at least in the short term I think that like any anybody who's
42:36
using this to create like enablers for people to do things I think that that's I think that's where it's going to go
42:41
um because I It's just tough I mean like chat Bots are great like I don't know it's a lot of hype there's a lot of hype
42:47
around like general intelligence um and maybe it's real like I'm not I
42:52
wouldn't call me like a Visionary but um yeah it's it's a it's a it's a great it's a crazy
42:59
time to be alive I will say that um it's a great time to be alive to be honest with you it's you know if you
43:04
told me if you told me you know when we started Gridiron or even told me like 10 years ago that you
43:10
know there would be tools like Chachi PT Bard or whatever whatever um githubs
43:15
like co-coding is where you could like take something like coding which is like use of a super high barrier to entry and
43:21
reduce that and help people be more effective I would have called you crazy it's crazy like I do a front-end
43:27
development work and like the first time I saw a chat EBT or any of these other alums doing coding it was like
43:33
mind-blowing you know like that word's overused so much but it really it really was because you put in so much time to
43:39
think about a simple every keystroke you put in to code something and this just spits out everything you need I mean
43:45
it's crazy I think it's one day it's gonna just why do you even need someone to code a website maybe to do fine-tuned
43:51
details for things but to do a general website or general stuff it can do it a lot of that stuff now it's it's insane
43:58
um you know if you want to just pump out of what describe a website you want to create you know I think chat GPT can pretty much do it for you a very basic
44:05
one right now but that's going to change quickly so it's it's nuts yeah which is great I mean people have to people have
44:10
to just learn how to use it and embrace it which I think like that's also part of it too is like you know the adoption
44:16
for guys like you and I Ryan is pretty high like I'm all about just like taking what's new and start using it and getting my hands there but there are
44:22
people that are just like they're going to be slower and like businesses are also going to be slower like I think the
44:27
Horizon for a lot of this stuff to be widely adopted um based on like my experience it's going to be slower than most just
44:34
because like not everybody moves at this at like the speed of light like one of the advantages and why I like Gridiron
44:40
is like there's only five of us five of us right now it's like we have to make a decision to do something we can just do
44:45
it like there's no like there's nothing there's no nothing complex about it there's no like corporate structure you
44:51
need to go through and ask for pull requests and all this stuff right yeah
44:56
after pull request to get business buy-in like find the value not be distracted by a million other things
45:02
that people think are a priority that may or may not actually be a priority like or they may actually drive value I
45:08
think that you know they're it's a lot easier to adopt things when you're not when you're on smaller groups so I think
45:13
that like Freelancers and stuff like that I think there's going to be a lot of opportunity for them for people to make money by using these tools and
45:21
going on like upwork or one or five or these other sites and and just selling services that are more or less just
45:27
helping just putting stuff into chat GPT and putting out and getting those outcomes into things you can sell yeah I
45:33
saw someone was uh doing like they said they made like a million dollars on upwork or Fiverr just by doing basic
45:39
work like because nobody else wants to do it but they just put it do it in Mass you know so it's nuts
45:45
um but yes uh what one thing I wanted to ask you hopefully Hunter didn't ask you this yet but for your like
45:53
for sports betting in particular I know you guys offer a bunch of different things do you have a win rate or like
45:58
what is that one right I thought someone advertising on your site before what that was but if there is one how long
46:05
how much data did you get to get there like how how much time did it take for you to get that to that number yeah so
46:10
like I don't have the top of my head so like you can go on like so we do put all of our picks and uh NFL pick watch
46:18
um so we were um like our accuracy I think for like spread is around 50
46:25
to 53 which is pretty good and then moneyline were in like the mid 60s and
46:30
it varies from year to year because the small samples um but we've been I think we've been doing it we started during the Kobe
46:37
football season like we started actually like publishing them in NFL pick watch
46:42
during covid so like that was what 2020 21 22 so this is the fourth year that
46:49
we've we've done it on the site um so we have a pretty good track for here and you can go on NFL pick watch and mine we have a link to it on our
46:55
site like and you can see like our trends like where we where we they're pretty good data too of like What teams
47:01
do you do the best with what teams you do the worst with um I will say that like I will say that one thing that the
47:09
sports betting stuff is really has taught me is the the way Vegas sets their lines is very good like when
47:16
you're when you're building models every model that we build when when it's trying to predict who's gonna win
47:22
against the spread it's very hard to get much left above like 53 54 accuracy just
47:29
because the lines are so well optimized it's honestly it's honestly crazy and you said you're right you were at 52 for
47:35
the for the spread it varies week to week and that's a that's the
47:43
challenge because we have weeks where it's like 16 we get like two games right and then we have other weeks we get 14
47:48
right so it's really it's really hard I mean like when I when I use our our
47:55
projections to bet the spread I I take that and I inject like
48:01
my personal opinion on it because one like I'm not a serious like I'm not I'm not a serious person when it comes to
48:06
sports but like I bet like five bucks so I'm betting this friend been like five bucks yeah you're not playing 50 Grand
48:11
at a time or something yeah you had to be a little bit different um we have had a lot of lot of success
48:18
on moneyline because you don't have as much it's harder the money Line's harder to make money on because you're not
48:25
basically you know you could be betting 10 you could be betting 100 to win 60. so you have to get that right more times
48:30
than not but we have had a lot of a lot of success with some betting strategies using the money line because it is so
48:36
accurate and this year we actually have a lot more data than we had last year and we've been adding in like a lot more
48:42
advanced like a lot more calculated inputs that that I think and I should
48:48
knock on wood that like I think should add lift because they're like things that make sense it's like pressure rate on the quarterback like like you know um
48:55
like first and second down um efficiency like those are things that we didn't have access to in the past
49:01
that we do now we have their calculated Fields but like like they actually do matter like how well you do in person on
49:08
like the passing Downs does actually like it is a it is a good indicator of like how good your team is if you look
49:14
at it over the course of a whole game gotcha so like the Vegas is I mean it's it's
49:20
scary how good they are predicting uh making the spread it's insane and I'm I'm sure they're going to start using AI
49:25
themselves as well so they'll even get better but it sounds like it's going to be like a race right like you guys are
49:31
it's I mean Vegas is probably doing their thing in the background nobody knows how they're doing the spreads but if they're implementing Ai and you guys
49:38
are doing it as well it's almost gonna be like a battle between you guys and and in Vegas to see who's going to stay
49:43
ahead of the curve yeah maybe like the thing about like the thing about spread is like Vegas actually doesn't care what
49:50
the final number is they just want the even money on both sides so they make they make it in the middle so like the so the reason like the
49:57
reason why these lines are so optimized is like let's say that let's say that you know let's say that they're let's
50:03
say the Eagles against um what's what's up what's a bad team I'm gonna offend Somebody by saying this
50:08
let's say um the commanders because they're in my division I'm a Browns fan so they're usually the bad team but I
50:14
like the Browns or the Browns are like I love the Browns um So like um so let's
50:19
say that it's equal goals at commanders and let's say that somehow their lines not optimized or like I said that we think it's a high probability and
50:26
everybody's hammering the Eagles like minus two which would be a great line against the Browns like at home like minus two like that's that's crazy like
50:33
the second biggest sees too much money on the Eagles they'll kick that lion up because they want people to start
50:39
betting the Browns like that line could move from two to six in one week if everybody's hammering the Eagles because
50:45
they need money on the other side are you saying it's more like Vegas isn't really doing that much it's more what the people the people are predicting the
50:52
games yeah yeah the the people are driving the lines more so than sites like ours like we can still like there
50:59
may there is potentially a way for us to find an edge of like lines that aren't correctly optimized and by doing that we
51:04
would have to come up with our own line and be like and basically try and predict what the final line would be for a game compared to the opening line and
51:11
be like all right we think the line's gonna move here so this is good value to buy right now but Vegas is actually
51:18
gonna always be ahead of the curve because they there's systems are already designed to make sure that they don't get killed on one bet and there's enough
51:26
people and like you know you can move lines like there there are like you know let's say that we had a million people
51:31
that'd be great so a million people use our site we said bet the Eagles and they all bet the Eagles that we could move
51:38
Vegas's line the thing about sports betting is I think it's getting to a point in the US where it's big enough where like
51:44
there's enough normal people that aren't like there's a bunch of people like my brothers or like my all my friends that
51:51
just like they enjoy just betting a couple bucks or 20 bucks on a game and they're gonna go to their podcast and
51:56
they're just gonna go and like sprinkle some money here or there that that it'll out it'll outweigh anything that like
52:04
sites like ours it may have an edge would do to remove that edge I don't
52:09
know if that makes sense or not it's almost like if you sample more people than Vegas does you could beat that you
52:14
could get more accurate odds if that's possible is that right I think so yeah yeah like
52:20
yeah if you if you could like you know I think that like going back to like where like some of this AI Tech could go like
52:27
if there was a way for you to like actually aggregate like sentiment about teams and like betting like where the
52:34
money's flowing um I think that you could probably come you could probably come up with a way to
52:40
say look we think all the money's flowing here so you should either get it on the other side now or be like wait so
52:46
I think that there is some there is some stuff there but it is you need to sample a huge amount of people and see where
52:51
their money is going and there may be like Services where like because I know that ESPN on like all their betting shows will like talk about
52:58
um we'll talk about like where where the money is because like you know Caesars will say that like oh there's 80 of the
53:04
money is on the Eagles like we don't have access to that data that'd be nice to know um but um but you know there there are
53:11
like Vegas hasn't like people do have access to that information that's just not available to us one thing we uh we
53:17
enter interviewed this one site they're called sherpa.ai it's like a it's a social media site where no they don't
53:23
allow humans so you can create bots on there and these Bots you can describe anything you want about them be like you
53:29
could say create a bot that loves fantasy football and all he does is BET on Sports all day and this bot will just
53:34
all day post on the social media site about who he's gonna bet on who he likes
53:39
and all I mean it's pretty incredible um but one of their use cases is um I thought it might fit with you guys is
53:45
you could create millions of bots and you do market research right so you could ask the Bots who they're going to
53:51
bet on and these Bots pull data from the outside world so instead of like at doing a market research group to call
53:58
people on the phone or however Vegas does it well Vegas does it because they get the betting in but um if you want to
54:04
sample a million people on who they're gonna bet you could do it via VIA their site um I don't think they're that big yet
54:10
but that's they they're pretty close to being able to do something like that that sounds really interesting yeah
54:15
maybe I'll try and connect with them because like the other use case for that too is like for daily fantasy so like you're thinking if you guys play DFS
54:21
even if you don't like a lot of the um a lot of the way you find an edge there is to find play it's like the start players
54:28
that other people aren't so if you can get like from Twitter or whatever like kind of sentiment on like who people are
54:34
like really high on you want to you want to fade some of those guys because you need a unique lineup like when you think
54:40
about like the million the million maker um like the people who make a Millions dollars in these tournaments it's
54:46
because they have like the amount of players that they have they have like a couple guys that everybody has but they have four or five that have like a two
54:53
percent ownership rate so you can only really get that by like trying to project out ownership which is really
54:58
hard right now but if you had a way to actually pull in like if you know all the sites are kind of hyping up like
55:04
this these groups of people and you see people saying hey this is what my lab looks like if you can get away from them
55:09
and fade those guys it actually gives you an advantage in these in these competitions to make sure you're trying
55:15
to get um low like low own like low owned players yeah it's almost like it's
55:21
better it's better just to take that risk because everybody's betting on the same people so it is like if you're
55:27
trying to win if you're trying to win like a like a like a tournament like not just a head-to-head game like you absolutely have to cigarettes and like
55:33
that's why we haven't really solved DFS yet it's very hard like to find an optimal lineup that wins consistently
55:40
especially in tournaments because you have to have enough variation and we haven't got to a point where like we do
55:46
a great job at that just because it it's a different it's a it's a it's a
55:52
uniquely different challenge for something that's so similar to Fantasy Football gotcha
55:57
but you'll get there one day yeah hopefully if you want to um promote anything now's the time to do
56:03
it like your site or anything else Twitter handles whatever you want you can just find us at gridironai
56:09
.com um Hunter already talked kind of about like kind of the services that we offer
56:15
um we're also in Discord um spread iron AI one word um it's open we have a lot of channels we we actually just moved from slack to
56:21
Discord like we're done being Boomers and slack we're in the Discord chain like you know we just yeah so we're in
56:27
there so we have a lot of good we have a lot of good people in there if you have questions um or or you just want to come say hi or
56:33
yell at me because you don't like what the model says that somebody you drafted um how they're going to perform like I'm
56:39
here for you guys um yeah so that's um thanks for the time Ryan I really appreciate it oh yeah I
56:44
could talk to you for another hour so yeah easy but good luck and everybody uh check out their site
56:50
um put put some money down maybe and uh who knows you'll maybe be a big winner and you'll be a millionaire don't take
56:56
it out on Andy if you're not if you're living homeless by the side of the road it's not Andy's fault um but uh yeah if you want to check out
57:03
our newsletter it's fryhyphen ai.com forward slash subscribe um you can check us out on Twitter as
57:08
well it's the Friday at the fry AI um that's it thanks so much Andy uh we'll be publishing this soon and uh can't
57:15
wait to talk to you again sometime good thanks Ryan [Music]
57:24
thank you