Skip to content

Kentucky Sports Radio

University of Kentucky Basketball, Football, and Recruiting news brought to you in the most ridiculous manner possible.

A New Way To Evaluate The FBS Passing Game

NCAA Football: Vanderbilt at Kentucky

If you’ve watched an NFL game or studio show on ESPN in recent years, you’ve undoubtedly heard of their advanced way of evaluating quarterbacks; QBR. Total Quarterback Rating is its full and proper name and evaluating quarterbacks in a more advanced manner is its game. According to various explanations scattered across the internet, QBR is calculated by combining numerous factors like Expected Points, Win Probability, and division of credit (i.e. Air Yards and Yards After Catch). Considering factors like these are a drastic improvement over the previous Passer Rating system which was constructed using numerous head scratching techniques. For one, Passer Rating doesn’t consider sacks in the equation and we all know the significant impact sacks can have on the passing game. Second, the values assigned to weight the rating are completely arbitrary, meaning there’s no real reason why they’re compiled that way. Lastly, the final rating doesn’t mean anything. It’s just a random number that has no relation to points, yards, or any other part of the game. After coming to the realization that this system needed repair, I set out to create a new way to rate the FBS passing game.

First, in order to do something like this, we have to have a variable to measure against like wins or points. By doing this we can get a more precise and logical value on how the passing game impacts a certain team. Since it’s difficult to quantify wins in college football for offensive players given the vast variety of different offenses, it’s easier to measure using points per game. Next we need independent variables in which to test. The variables I used were Pass Yards per Attempt, Completion Percentage, Interception Percentage, and Sack Yards per Attempt. Pass Yards per Attempt are difficult to quantify in college football because of teams who run specialized versions of the option. For example, Georgia Tech was 2nd in FBS last season in Pass Yards per Attempt despite passing the ball only 194 times and completing 53% of those passes. To correct this issue, I multiplied a team’s Pass Yard per Attempt figures by their Completion Percentage. By doing this, option teams who have high YPA numbers drop due to their low Completion Percentage figures. All other factors are kept the same. Once we do this, we can figure out how many points the passing game adds to a respective team.

You may have noticed that touchdowns are absent from my ratings. Unlike the NCAA passer rating that uses touchdowns in the formula, I don’t believe touchdowns should be used. Why? For a number of reasons, actually. Tempo isn’t considered, meaning teams like Louisiana Tech (with Dykes coaching) can inflate their touchdown figures with more possessions. Brian Burke of Advanced NFL Stats explains the issue further.

They (TD passes) are the result of many other factors beyond QB passing proficiency. For example, a QB could be on a team with a terrific defense that frequently produces turnovers deep in an opponent’s territory. Or he could benefit from a great running offense that sustains drives. Further, TD passes are the culmination of all the other attributes of the passer. Accuracy, avoiding interceptions, and avoiding sacks all lead to TD passes.

Once all things were considered and the simple regression was performed, I applied the ratings to last season’s FBS teams. The below table contains the top-10 teams from last year in the new model, their rating above average, and a comparison to their NCAA Passer Rating. The coefficient values are as follows; Pass YPA*Completion percentage (6.19), Sack YPA (-1.06), Int. % (-1.17). They’re all significant at the 5% level. 16.8 points added is average for FBS.

top passers

First, and most importantly, the column with the regression label estimates how many points per game the measured factors were responsible for. After crunching the data, we get rankings that are very similar to the NCAA system, but we now have a value which makes sense, rather than a value that’s completely arbitrary. We don’t get results that are out of the ordinary either, even when we disregard passing touchdowns. The entire set of rankings can be found here. Neal Brown’s Texas Tech offense ranked 20th nationally while Kentucky ranked 112th last season.

In my mind, the coolest thing about these ratings (if a regression rating system can be cool) is the fact you can evaluate individual players as well. This is especially useful when a team like Kentucky is experiencing quarterback turmoil. By plugging in a players’ individual numbers we can determine who would be the best fit as a passer. The below chart contains 2012’s data of Kentucky’s three quarterback options.


When looking at last year’s statistics we can determine that Maxwell Smith would be the best fit going forward for Kentucky’s passing game. He led the team in Pass Yards per Attempt*Completion Percentage and had a low interception percentage. He wasn’t particularly mobile as evidenced by his Sack Yard per Attempt numbers, but he was able to get rid of the ball when in trouble, as proven by his five sacks in three full games. However ratings like these leave out one very important factor in judging a quarterback; rushing ability. After all, we’ve seen the impact that rushing quarterbacks can have. Tim Tebow and Johnny Manziel have shown us that. But adding rushing ability into the equation was a slippery slope because quarterbacks typically don’t rush. Why would you want to penalize an effective passing quarterback with no rushing ability? Peyton Manning can’t run at all and he’s still arguably the NFL’s best. Since I couldn’t find a solution to this problem, I left rushing out.

If rushing is considered and his spring game performance continues into the fall than Jalen Whitlow will have a strong say in who gets to start. However, if Maxwell Smith picks up where he left off last season, Stoops and Brown should give him the nod as starter. It’s anyone’s guess as to who will start, but if efficient passing is the only thing considered (it’s not) and last year’s numbers hold somewhat constant, Maxwell Smith should start because he adds the most through the air (providing he’s 100%).

Article written by Jonathan Schuette

10 Comments for A New Way To Evaluate The FBS Passing Game

  1. Weekend Reader
    8:37 pm June 26, 2013 Permalink

    Thank God! I for one am sick of NBA draft rumor articles. Give more football. Stoops!!!

  2. whattttt
    8:53 pm June 26, 2013 Permalink

    What bridgewater was not in the top 10 LOL

  3. bigbluetrue
    8:57 pm June 26, 2013 Permalink

    I think Whitlow will be the starting QB when it is all said and done. Mobility in the new system is crucial–something Smith has little of.

  4. chemcatUK
    9:16 pm June 26, 2013 Permalink

    Did you MULTIPLY pass yards per attempt by completion percentage or DIVIDE it? The way it’s written in your table (PYPA/Comp %) indicates that you divided, which contradicts what you wrote in the piece. Obviously dividing by a lower percent would INCREASE the stat.
    I think this was just a simple oversight. I LOVE THE STATISTICAL ANALYSIS present here Keep it up!

  5. Jonathan Schuette
    9:29 pm June 26, 2013 Permalink

    #4. Good eye, it was multiplied but I put a / there by mistake. I just tried to show it was put together. I somehow forgot about the * symbol which just makes me feel brilliant. Anyway, it’s fixed now.

  6. I wonder
    9:29 pm June 26, 2013 Permalink

    If I sleep with my sister, does that make me a UK fan automatically?

  7. Stop Hammer Time
    9:31 pm June 26, 2013 Permalink

    Harrison twins did not make the grades. Hope Polson is ready.

  8. FDLJ
    9:43 pm June 26, 2013 Permalink

    Whenever DJ, the guy that calls me Dicky, texts me and tells me check out an article on KSR, I’m never disappointed! What an informative and great article JS.

  9. Weekend Reader
    10:53 pm June 26, 2013 Permalink

    #6 Negative on that. It would just make you a whole lot like your own father. Stay on the Louisville blogs more linebeards like your sister.

  10. Lamont
    12:57 am June 27, 2013 Permalink

    Great analysis with deep statistical algorithms
    I see a 2-10 season with plenty of hope for another 50,000 plus Spring Game attendance