Creating a Sports Betting Model 101 - Intro to Linear Regression (The simplest model ever created!)

do you want to create a sports betting model but have no idea where to start or what to do well this video is gonna be for you welcome to sports betting truth where it is my goal to give you actual sports betting advice without the touting Schilling hype or false promises now the idea of this video today it's pretty simple this video is going to present the easiest absolute most simple model possible to make in Excel for sports betting the absolute simplest model you can make the idea of this video is to give you a general idea of what exactly sports betting modeling is using the most basic concepts out there just so you have an idea of what is possible a sports betting modeling and where exactly you would begin I think a lot of people are overwhelmed just by the word model they hear that word and they run for the hills the point of this video is to show exactly what it is but at the simplest level at the most basic level but keep in mind this video is at a very basic level it is literally a one-on-one level video if you have any experience at all with this you're probably gonna look at this video and be like that's that's basic because that's the point the point is to introduce the concept of sports modeling to people who have no experience with it at all now this is gonna be a macro free model everything in this model is going to be done with basic Excel functions and I'm going to point this out from the get-go this model is not going to win you money so don't watch this video expecting to use this model to profit it's too basic of a model it's too easy and too simple of a model for it to make you money now I will say this model would probably be better than doing absolutely nothing at all but it still wouldn't win you money in the long run but like I said that's not the point of this video the point of this video is to highlight the absolute basic most core level principles of sports betting modeling so with that being understood let's get underway so here's my excel file titled simple NFL model I'm using the NFL here just because as the least number of games in a season and I feel like that would be the easiest to do because it's just that have a lot of games compared like baseball I have last year's NFL schedule here on this tab titled schedule I have the O a team the away score the home team in the home square that's all I have right now so what I did is I went to Pro Football Reference comm it has a lot of good resources for all sports not just football and so what I did is I decided for this model to use one stat because that is the absolute simplest way you can do models using one stat so I figured a good all-around stat that we could use will be the stat right here called SRS it stands for simple rating system it's basically a very simple way to assign a power rating to every NFL team and sports reference does this for all sports not just football so what I did is I took every team's SRS rating and I put it in this tab right here next to their name so this is every team's SRS like the Patriots a 5.2 SRS the Dolphins - 8.8 obviously the higher the SRS the better and so I go back to my schedule tab and I have this two column right here margin and then I this column right here a SRS and this column right here HS RS so away SRS home SRS so what I'm going to do here for margin is pretty simple just a simple Excel equation away score - home score B 2 minus D 2 minus 6 for that game and then I'm just going to fill it all the way down double click right there and all the margins are calculated for a SRS what I'm going to do is a vlookup so equals vlookup B 2 or a to my about a 2 comma table array so I go over to an SRS tab and I highlight this table make sure you put the dollar signs in front of the letters and numbers here on the a1 to be 32 so when you fill it doesn't change the letters and numbers and then column index number so how many columns over it is two so the second columns two and then false and that will fill in the Atlanta Falcons SRS and then I just fill that all the way down and now every team's SRS every away team's SRS is now populated so now I'm going to do the same thing for home SRS you can just copy and paste this if you want to and just change b2 to c2 and now it has the home team's SRS in there and now you fill that in so now we got this filled out so we have the margin which is going to be our dependent variable the dependent variable here is what we're trying to measure according to our stats so in this case that's the final score margin of the game because we're trying to see how the final score margin is influenced by these stats right here these stats right here what are called our independent variables aka they're not going to be influenced by anything they're independent but we're hoping that the margin is influenced so now that we had this all filled out all the way for every game what we're going to do now is calculate our equation so this model is going to be based on basic linear regression principles so what you want to do is go to the data tab but most people aren't going to have the add and installed I don't on this new laptop so what I'm gonna do is add it myself and show you how to do it so you go down to file options add-ins and you're looking for analysis toolpak you want to click on that and click go and then click analysis toolpak and analysis toolpak VBA and then hit OK and now we have this tab right here data analysis so you click on data analysis and then you want to scroll down to regression so you click on regression and then you have this right here so our input wise is going to be our margin our y-value our dependent variable so just click that control shift down whoops control shift down and there's that and then click on the X range box scroll back up and select our deep or select our independent variables our X values control shift down so we have that filled in so you want to click on labels just so you know what the values are you want to have new worksheet you want to have residuals and line fit plots and then click OK and now it calculated our regression a lot to unpack here but trust me it's not that complicated so we have our line fit plots first of all what we're looking for right here is this line fit plot right here because this is what we're gonna use so basically it plotted all the margins and the predicted margins based on this simple model which is the orange line so basically what this number right here our square this is an important number so the R square here is basically how many of our dependent variables the margin were predicted by the independent variables the SRS in this case about 30 1.7 percent of the margins could be explained by the SRS which 31 percent is average basically the higher the R square is the more predicative your models going to be so you want to shoot for a high number here but it's not the end of the world if it's low I'll just say that besides we're only using one stat here so obviously it's not going to predict everything so 31.7% for one stat is actually pretty good so that's the most important stat appear under regression statistics what you're looking for right here for regression f you want to make sure this number is low which it is as long as it's under 0.05 you're good and then there's some important numbers down here the intercept is a very important one right here the intercept is basically where the line starts in this case minus 2.2 4 but also in sports gambling modeling the intercept is almost always the home advantage in this case the away team had a minus 2.2 for point disadvantage for every game according to this model which is pretty accurate it's pretty accurate with what we know about home advantage constants the simplest way to explain coefficient is that for every change in our independent variables how did the dependent variable change so basically for every 0.98 unit increase in the OAS RS the margin changed that's the simplest way to put it it's basically the slope of the line it's basically the angle of this slope right here and then the other important value here is our p value basically just like significant f if it's under 0.05 you're probably good the lower the better here but especially under 0.05 when you're using a lot of different variables instead of just one there's going to be a lot of numbers here but the lower numbers are going to be the ones that are more predicative so that's basically what we're looking for here so in this case this is one giant equation you want to copy and paste these three things and bring them back over to our schedule tab just put them right here it doesn't matter where you put them so the intercept is the home advantage so this is actually all one giant equation right so let's pick a random matchup let's go with the Cowboys and the Lions okay so the Cowboys SRS this past season was one point one so we're gonna put that right here the Lyons SRS this past season was minus three we're gonna put that right here and let's pretend that the Cowboys are playing at the Lions so our equation is basically going to be very simple it's going to be the intercept plus the ASRs times the Cowboys SRS and the HS are s times the lines srs so I'm going to put this equation right here equals intercept P 10 plus parentheses Cowboys SRS 1.1 or P 14 times the ASRs coefficient in this case right here 0.98 oh nine seven close parentheses plus the lions SRS in this case minus 3 times the HSR s coefficient in this case minus point minus one point oh two two seven one three close parenthesis that's our equation right so what this number represents one point nine oh six seven two seven nine that number represents the predicted margin of this matchup and since we're basing everything off away teams basically the Cowboys would be favored by one point nine oh six seven two seven nine points in this hypothetical matchup so let's say the odds boards are showing the Cowboys at minus three but your model right here says that the Cowboys should only be favored by one point nine oh six seven to nine so basically according to your model the Lions have an edge of one point one points there is one point one point of value on the lions at plus three so according to your model if you're gonna bet according to your model you would bet the Lions because of that edge assuming your model is right that is basically the simplest way to do a model right there now like I said that models Nakada winning money it's too simple but you get the idea that is what is known as a linear regression model there's other ways to model but this is the simplest example I could give now what if you wanted to add more stats than just SRS well I had that prepared right here on the stats tab I got these columns ready to go I picked four random stats to go along with SRS just I just randomly picked four stats I picked offensive yards per play offensive penalty yards turnovers forests and defensive rushing yards per play allowed random stats so what we're going to do right here is run the same concept that we did right here and fill these stats in we're gonna fill these stats in just like we did right here so let me do that like I said vlookups over to the stats tab three false whoops oh I did V look it that's why if you look up alright and this right here is the away offensive yards for play for the Falcons we're gonna filled it all the way down and then we're gonna do the same thing across the board I'm not going to show that to you alright so we got all these stats filled in so what we're gonna do now is the same thing we did with the simple model we're gonna regress based on these five statistics so go to data data analysis regression our Y range is gonna stay the same we're still trying to measure the margin for a dependent variable and our independent variables are gonna be all these here we go let's see if this is gonna be any good and all right our R square barely improved so it looks like these stats I picked probably weren't that predicative compared to everything else but again their basic stats they're not advanced stats and they're unadjusted stats that's another debate for another time so what we're gonna look for here is our p-values so basically this value is very low so that one looks good this value is low so that looks good but all the rest of them are over 0.05 so basically none of these variables I picked are predicative compared to these two so that explains why the r-squared didn't really change that much but the equation is the same you would do the same exact thing like you did for the simple model it's the same exact concept nothing changes only you have to add more coefficient calculations so if we did a hypothetical matchup of the Dolphins and the Browns basically we would take every team's statistics so the Dolphins right here they're right here and then there s RS was minus 8.8 and then the Browns knew they are and then the Browns are right over here here's their stats and their SRS was minus 0.3 so you'd do the same thing the intercept in this case is not going to match the home advantage because 11.9 for obviously doesn't make any sense it doesn't always match the home advantage but if your model is OnPoint it usually will so it's going to be intercept in this case are six plus the Dolphins srs times the srs coefficient right here plus the Dolphins yards per play times the coefficient which is right here plus the Dolphins penalty yards times the penalty yard coefficient plus the Dolphins turnovers force times the turnovers force coefficient plus the Dolphins defenses rushing yards per play allowed times that coefficient right here and then you do the same thing for the Browns the rushing yards per play allowed times that coefficient and there we go there's our equation so according to this the Browns at home against the Dolphins should be favored by twelve point eight four which is a very high spread you rarely see spreads that high in the NFL so that also should introduce some skepticism about this approach on this quick model I threw together but hey it could be right you never know if your model spits out a line like that maybe it's right that's why you test these things over time and do a bunch of trials and testing in the long run to see if it's actually a worthwhile model or not that might be a good number it might not be you don't know until you test it out over time a bunch of trials to see if it's worthwhile or not that's what you're trying to do there's only one way to find out if a model is any good or not and that's the testing there's a bunch of different ways you can test a model that's another video for another time but that's how you would gate whether this number is any good or not but anyway that is the absolute simplest way to approach a sports betting model again these models will not win you money but I hope it helped illustrate the concepts of models and I hope at least now you have a direction on where to go going forward again this is the very ground level can get so much more complex and advanced than this but it's up to you to see how much he ultimately want to advance yourself but at least now you have the roots the foundation to build upon when it comes with modeling this is only one type of model again this is a linear regression style model it's not the only way to do it there are other modeling methods out there you have power ranking models you have Monte Carlo models but this is a linear regression model probably the simplest way to do it I hope you found this video educational and helpful if you like this video go ahead and hit that like button and subscribe I will be bringing more videos like this to you to help educate you and teach you about sports gambling and approaches you can take for long term sports gambling profitability until next time this is sports betting truths signing off

Channel: Sports Truth with William Leiss

Views: 94,191

Rating: 4.8957653 out of 5

Keywords: Sports Betting, Sports Gambling, Sports Wagering, Sports Handicapping, Betting, Gambling, Wagering, Modeling, Mathematical Model, Mathematical Modeling, Analytical Model, Analytical Modeling, Statistical Model, Statistical Modeling, Sports Betting Model, Sports Gambling Model, Las Vegas, Betting Model, Gambling Model, Sports Betting Picks, Sports Gambling Picks, Sports Picks, Sports Analysis, Las Vegas Sports Betting, Winning Sports Picks, Linear Regression Model

Id: t8ViRP2zZ3A

Length: 17min 19sec (1039 seconds)

Published: Sun Jul 28 2019