Creating a Sports Betting Model 101 - Intro to Adjusted Stats (Power Rating Systems!)

so what is the most important part of building a power ranking system well this video is going to tell 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 today we're gonna talk about the absolute basics of building a system based on power rankings power rankings are a way to analyze teams in comparison to one another it's a way to rank teams from start to finish from best to worst using equivalence so you can actually compare the two apples to apples instead of apples to oranges now developing a power ranking system that's accurate takes them back end work you can't just rank teams base off basic stats and expect it to be accurate especially when it comes to predicting and especially when it comes to predicting against the spread or trying to use that prediction to bet with so you need to make sure things are accurate so the most important concept that any good power ranking system is going to apply is using adjusted stats basically the question you're trying to answer with adjustment stats are if these two teams played the average team on a neutral court what were their statistics be because not all teams are created equal in the example I'm going to use is going to help illustrate that adjusted stats take into consideration for the most part the strength of the opponent played so the example I'm going to use today is college basketball I got last season stats loaded up here in Excel for every team all 300 I think it's up to 353 now but all 350 three of them have all their stats right here now college basketball is my favorite sport to analyze with probably because it's a sport I started with and second the sport lends itself to easy analysis it's not that complex you don't have that many stats or variables out there like you do in baseball or football so the stats we're gonna look at today are pretty simple I'm going to remove the stats we don't need so these are the stats that are going to be relevant to the example I'm going to use today when it comes to adjusted stats so we have points opponent's points field goal attempts free-throw attempts offensive rebounds and turnovers the reason I have these four stats in there is to calculate possessions so with those stats out of the way the question I'm going to answer today with these adjusted stats are which one was more impressive I have two games here that were played last year the first one is the ACC conference tournament where Florida State beat Virginia 69 to 59 the second is a game on February 19th where Buffalo beat Ohio 114 to 67 Buffalo scored 114 points now if you're using unadjusted stats Buffalo's offensive efficiency would be one one four slash eighty point seven times 101 forty-one Florida State's offensive efficiency would be sixty nine divided by fifty eight point two times 101 eighteen point five five so buffalo has a higher offensive efficiency here but the reason that is is because the stat is unadjusted that is what we call the raw stat the raw unadjusted static but here's the deal Virginia as everybody knows has one of the best defenses in the country in 2019 Virginia's defense was at least a top-five defense they might not have been as good as Texas Tech's defense but they're definitely up there so the question we're trying to answer is is scoring 69 points on Virginia's defense more impressive than scoring 114 points on Ohio University's defense that is what the adjusted stats attempt to answer and that's why they're the core building block of any power ranking system that's going to be accurate because you have to adjust your stats like this in order to be able to compare apples to apples otherwise you're gonna have teams like Buffalo no offense to them who played in a weaker conference racking up impressive point totals their offensive efficiency is going to look more impressive than a team like Florida State who had to play against a team like Virginia that's just the way it is so adjusted stats attempt to normalize that and equivalence that to see who truly is more impressive based on the opponent's you played so the first step of adjusting stats is to calculate national averages so we're going to calculate the national average here for all teams average equals average B to be 999 just or three five five just drag that over to do averages for the wrists scroll all the way back up and now we're going to insert a couple columns here Oh II and de offensive efficiency and defensive efficiency so for Abilene Christian that's going to be b3 / H 3 times 100 so their offensive efficiency raw raw was 111 point two nine nine and then fill that in for everybody else as you can see you can already point out the problems Abilene Christians offensive efficiency of 111 is higher than a team like Alabama so Abilene Christians offensive Fisher Steve 111 is higher than Alabama of 105 do you really think Abilene Christian had a better offensive Alabama I don't think they did even though Abilene Christian won the southland tournament and everything I don't think their offense insufficiency was better than Alabama's so that is why adjusted stats are important because it fixes that defensive efficiency is going to be measured the same way here it's going to be C 3 / H 3 times 100 alright again are you really going to buy are you really gonna buy the fact that Abilene Christian had a better defensive efficiency than a team like Auburn who made the final four I'm not buying it I bet a lot of that can be explained by the flack that the Southland conference last year is very bad again this is what adjusted stats are going to fix another answer adjusted stats will give us here in basketball is tempo the Florida State Virginia game only had 58 possessions because Virginia plays at the slowest pace in the country under Tony Bennett every year Buffalo under nato's was known to play at a very fast paced every game so it's no surprise they're gonna score more points that's why he measure points per possession instead of raw points scored to factor in the tempo and the pace of the game was played that's what drove me to college basketball analytics in the first place because I went to TCU and Air Force is in our conference and everyone would always talk about how great Air Force is scoring defense was on a points per game basis because they played a Princeton offense which the game down to a halt it was no surprise that their scoring defense was so great because they limited the number of possessions every game and that's why scores and airforce games are so low I'm not saying their defense is bad but their scoring defense being so low had a lot to do with their pace efficiency attempts to measure in actual defense and offenses capabilities on a per possession basis so whenever you see a basketball broadcast advertising points-per-game shake your head because it's truly on a points per possession efficiency metric that actually matters and actually will show a team's true capabilities all right so we need to do the averages for the offensive and defensive efficiency so the national average efficiency was 106 so I'm gonna bring this number up here I'm gonna put this up here league average 106.000 7-5 so here's how we adjust our stat we're gonna take this game right here and calculate Virginia's efficiency as well it is time to adjust the moment you have been waiting for so the first step is to find the team's season-long efficiencies so I'm just going to do a vlookup here all right so Florida State's season-long offensive efficiency for the year was 108 point 403 one for Virginia it was 118 for Ohio it was 99 for Buffalo is 115 again unadjusted raw season long efficiency and then we're gonna do the same thing for defense call them ten so Florida State's season-long unadjusted raw defensive efficiency it was 97 Virginia at 92 Ohio 104 Buffalo 96 alright so now we got that in so what we want to do now is calculate the difference from the league average for every team for both stats basically what we're measuring here is value above average or below average so Florida State we're gonna do t5 minus t2 so Florida State's offensive efficiency unadjusted was two point three nine points above the league average so they're above average offensively Ohio here is below average off and now we're gonna do the same for defense so Ohio still had a defensive efficiency those below the league average so they were above average defensively not by much but they were not a bad defense that Buffalo scored all these points on alright now that we have these deltas what we're going to do is add those deltas to each team's game efficiency so in this case it's going to be right here for Florida State q five one one hundred eighteen point five five six seven so equals Q 5 plus X 6 which is Virginia's defensive efficiency below average actually it's minus Q five minus X six so if that factored in Florida State's adjusted offensive efficiency for the game is higher than their raw because Virginia's defense is so good it made their number go up because we subtracted that from here which rose the number by thirteen point one one points basically what this number represents right here what this one three one six seven four seven number represents right here is theoretically how many points per hundred possessions Florida State was score on a neutral floor against an average opponent against the league average opponent with a league average defensive efficiency of 106 how many points per 100 possessions would Florida State score against that team well theoretically it would be a hundred thirty one point six seven four seven that's what the adjusted stat is measuring so we're gonna do the same for Virginia we're gonna do their game efficiency which was Q six 101 minus X 5 so Virginia's efficiency went up as well because Florida State above-average defensive team there's one up to 110 theoretically if Virginia played the league average team the average team if they're an average team hypothetically would have a defensive efficiency of 106.000 seven five we could theoretically expect Virginia to score a hundred ten point one six points against that team per 100 possessions that's what the stat justement is measuring alright let's do the same for Ohio Buffalo alright Ohio's game efficiency of 83 - Buffalo's seasons efficiency below average of minus nine point one one so Ohio's efficiency rose in 92 and now Buffalo this is the stat we want to measure equals game efficiency of 141 - Ohio's a fitness e below average of minus one point seven six so 143 so we just answered the question I asked earlier so Buffalo scoring 114 on Ohio is more impressive according to this adjustment than Florida State scoring 69 on Virginia according to this adjustment according to the stage 1 adjustment so theoretically Buffalo if they played the league average team with the league average defense would score 143 points in a hundred possessions so that is how you adjust stats that is the basic level of adjusting stats now if you want to build a good power ranking you would do that for every stat and you wouldn't just do it once this is what I call a level one adjustment you are adjusting raw stats into a first level of adjustments but what you do is you do that for every game for the season right you would take a team's season-long stats like over here you would take these stats for the season and do it for every team for every game they played and then you would get an adjusted overall stat and then you would do it again because the first wave or the first compliation was taking the raw on adjusted stat and turning into a first wave stat right but then you would do it again because now you're adjusting based on the first wave of stats into a second wave of adjusted stats and then you'd do it over and over again until the numbers stabilized so any power ranking that uses adjusted stats isn't just gonna leave it at that that's just the first wave you do it over and over again to normalize the stats because the first wave of adjustments are still going to be pretty raw because once you adjust it the first time you're gonna want to adjust again based on those adjustments so the second time around you're adjusting it's adjusting based on more realistic stats more normalized stats that's the concept behind it but any power ranking system worth a damn is going to use this concept for statistics that is what adjusting stats accomplishes you can do this for any sport as long as you can calculate national averages you're good you just need to do it on a game by game basis though if you can't do it on a game by game basis at the very least you're gonna have to calculate the season average statistics the combined average season statistics of all the teams that the team played and there's a lot of ways you can go about this right there's a lot of ways you can adjust stats you could weight more recent games more heavily than games further in the past right give more weight to recent games that's another way you can adjust stats there's no limits here you're only limited by your imagination when it comes to adjusting stats now most stats are adjusted that way but there are some stats that aren't quite adjusted this way generally stats where a team's tendencies come into play might be adjusted different I could do a whole video I could do a whole video on how to adjust tempo in basketball because tempo and basketball is a stat that is not adjusted this way because tempo and basketball is dictated by how the team wants to play whether it be their style or in the case of a team that falls behind early who plays at a slow pace usually will have to speed things up to get back in the game tempo is adjusted in a different way but that's another video for another time I hope you found this video useful it's just the first step in making a good power ranking system that you can use to make predictions with it's just the first step this video is intentionally pretty low level just to illustrate the very basics of Stata justments but like I said you can advance this concept very far and make it as complex as you want it to be I plan on making more videos like this going forward to help illustrate basic sports-betting concepts as well as more complex ones so if you like this video hit that subscribe button go ahead and hit the like button and stay tuned for more videos until next time this is sports betting truth signing off

Channel: Sports Truth with William Leiss

Views: 26,251

Rating: 4.9099097 out of 5

Keywords: Adjusted Stats, Power Rankings, Power Ratings, Sports Gambling, Sports Betting, Sports Wagering, Sports Handicapping, Sports Modeling, Sports Betting Model, Sports Gambling Model, Las Vegas, Sports Betting Picks, Winning Sports Betting Picks, College Basketball

Id: aEuYDWEAteY

Length: 15min 17sec (917 seconds)

Published: Sun Jul 28 2019