Updated: Oct 21, 2018
Im sure most coaches will answer that question with, MORE! For anyone that has been following college football over the past few years, you should be well aware that the advantage has swung in favor of the offense. The popularity of no huddle, spread, up tempo offenses along with rule changes making it tougher to defend the pass as well as hit the QB, has led to the surge in scoring offense in recent years. So far through seven weeks of the college football season, the 130 FBS teams are averaging 30.6 points per game. You don't need an advanced degree in statistics to come to the conclusion that scoring a lot of points helps you win games, however, here at Ace we like to dig a little deeper. This article isn’t just about telling you that Alabama is scoring 53.6 points per game , therefore they are 7-0. That would be too simple and not tell the whole story about actual Points Per Game totals and the main effect they have on wins. Fact is, Alabama is what we call an outlier and those results cannot expect to be replicated by the average or even the very good teams in college football.
The purpose of our analysis around points per game is two fold
1) To reject the Null Hypothesis ( Fancy Statistics Term) that the Win means (averages) are equal across the categories of Pts. Per game. In simple terms we are setting out to see if in fact points per game leads to more wins.
2) Once we confirm that Pts. Per game are a predictor of wins, I wanted to identify the number of points per game that positively impact the number of wins. As you will soon see, the difference of a couple points per game in either direction can be the difference between winning and losing, or in the case of a college football coach, the difference between sitting home watching bowl games from the couch or participating in a bowl from the sidelines.
For this analysis , since Wins are a continuous target, a linear regression (ANOVA) based approach is used. This approach uses several techniques to assess the linear relationship between the input of Pts. Per Game & the continuous target of Wins. As you can see from the graph below, there is a statistically significant , strong main effect of Pts.Per Game on Wins. The important term here is Significant. When a result is statistically significant, we mean that it is unlikely that the result occurs because of chance or random fluctuation. Since statistical significance was achieved, we can say that the more points per game you score the higher likelihood of wins. So we are able to reject the hypothesis that Wins are equal across all categories of Points Per Game. This is easily identifiable by the graph as you follow from left to right. The further to the right is a Higher average of Points Per Game and the higher up is the higher average number of Wins for that group. You can see the linear relationship between the two by the line getting higher as it travels further to the right.
So, now that we solved the mystery of the more points you score, the better chance you have at winning, lets take a look at how many points in fact lead to more wins.
After running our analysis, the computer model came up with 5 Levels of Points Per Game and the average number of wins associated with each of those levels. The levels of Points Per Game were broken down into the following categories with the levels of High or Low win totals having more influence. Each group contains of an average of 26 teams which in total equals the 130 FBS teams.
As you can see from the line graph above the High Group which is averaging more than 36.8 points per game is avergaing just about 5 wins each (out of an average of 6.3 games played) In fact the 26 teams that are averaging more than 36.8 points per game have a combined record of 128 wins and 38 losses through the first 7 weeks for a 77% win rate.
Compare that to the Low Group which consists of 27 teams who are averaging less than 24.30 points per game. The 27 teams in this group are averaging 1.8 wins each and have a combine record of 45 wins and 113 losses or a 28.4% win rate.
The middle two groups consist of 25 teams that average between 27.70 - 31.80 points per game and have an average of 3.6 wins per team and 26 teams that average between 24.3 - 27.70 points per game for an average of 2.8 wins per team. In college football every game counts and when your fighting for a bowl game, you can see how a few points in either direction can make a big difference as there is just about a full 1 win difference between the two middle groups alone, and a full 3 games difference between the high and low group.
Maybe that doesn't seem like much to you, but I promise you it is. As a former coach, I can tell you that the difference of winning 1 more game a year can mean a bowl game, a championship and a contract extension, while the difference of 3 less wins can get you fired! So these numbers matter.
For your Reference Here are the teams in the High Group & Low Groups through October 17th 2018. You may notice a few outliers that fall below the win average such as Georgia Tech & New Mexico with Only 3 wins the high group and SanDiego State with 5 Wins in the low group but for the most part, the wins in each group are pretty well distributed. As I will discuss later, while Points Per Game is a main driver of Wins, its not the only driver.
So what does this mean for you?
It’s important to keep in mind that even though there is a Significant Strong main effect of Pts per Game on wins, there are no guarantees. Just because your team scores more than 36.8 this weekend, doesn’t guarantee they will win. Just take a look at the list of teams in this category and although they are averaging more wins than any other group, the group as whole is has a win rate of 77%. Don’t get me wrong, a 77% win rate as a football coach gets you in the hall of fame and a 77% win rate picking games in Vegas makes you a millionaire over the long term, so I’ll take 77% all day but it’s still not a guarantee. What statistical and predictive analysis give you is evidence and information as to what is likely to happen and studying it is a way to at least give yourself the upper hand.
According to our analysis the Effect of Points Per Game on Wins has a Predictive strength of 50%. Predictive strength measures how well a predictor accurately predicts a target. You can use predictive strength to compare the predictors. Predictive strength is displayed as a percentage. A predictor with a predictive strength of 100% perfectly predicts a target. A predictor can be a single input or a combination of inputs. So even though PPG is significant and has moderate predictive strength , it does not perfectly predict Wins.
As a coach, you can measure your team in terms of which Points per Game Category you fall into (High, Medium, and Low) and based on that have an idea of where you need to be in order to improve. You can tell your team that 3-5 more points per game could mean and extra win and potential bowl eligibility.
It’s not about needing to score 45 points every week but maybe it’s the kicking game or improving your red zone scoring % to give you that extra few points a game needed to take you into that top group. Keep in mind that this data is for games through October 17th, most FBS teams have played between 5-7 games, some against lesser opponents (Cupcakes), so the scoring averages and categories are likely to change as the season progresses and competition gets stiffer. We will run the analysis again later in the season, so be sure to check back each week.
As you know and will learn over the coming weeks (if you follow my blog) is that in addition to scoring points, there is a lot more to winning football games than just PPG, just ask Toledo, Memphis & Georgia Tech. There are so many other factors that go into winning football games such as defense, turnovers, red zone offense & defense, running vs. passing, time of possession, penalties, special teams, 3rd Down Conversions etc. Over the coming weeks we will continue to share our analysis on what drives wins, so be sure to come back often to see what’s up next.
My hope is that you can learn from each post and take something away to help become better coach, a more informed fan, or improve your weekly picks. The more knowledge you have about what leads to wins, the better decisions you will make.
Want More Info?
I am always happy to answer questions, provide additional information or even help you analyze games. Feel free to contact me through the website with a request. I also offer my data collection excel file for a small fee which contains thousands of sortable data on all 130 FBS teams, and can create custom Pivot tables for any teams, opponents or games you want to analyze. I am available for self /opponent scouting, game analysis or general questions, so please reach out by email.