A 2020 research paper “Predicting the national football league potential of college quarterbacks” from the University of Colorado attributed ESPN QBR rating and quarterback rushing performance as the two main predictors of NFL success. Here is an excerpt of the paper.
… we explored the relationship between the performances of rookie quarterbacks in their first five years in the NFL, as measured by ESPN’s Total QBR, using data from their college careers and, in some cases, scout grades. We found that quarterbacks who recorded higher college QBR values performed better in the NFL than players with lower QBR values. Deconstructing the aspects of college quarterback play important for NFL success, players with better college rushing statistics performed better in the NFL than players with worse rushing statistics. The same was not true for players with better college passing statistics. That is, among quarterbacks selected for the NFL, college passing ability was not significantly correlated with NFL performance. These results were present both when the EPA components used for QBR calculations (EPA from passing and EPA from rushing) were used to measure college performance, and when traditional college metrics (passing yards per attempt and rushing yards per attempt) were applied. Combining results from the two stages suggest that college quarterbacks must be high-quality passers to make the NFL but, on average, quarterbacks also have to be good rushers to succeed in the NFL.
To look at these findings I collected the ESPN data on both for the last four seasons of college football. The data source can be found at https://www.espn.com/college-football/qbr/_/sort/schedAdjQBR/dir/desc. To draw a cutoff point on these I used the PASS metric sorted for the All-Time bests which goes back to 2004. From that data I selected Patrick Mahomes’ QBR of 82 which was the lowest of the top quarterbacks listed. I also selected Russel Wilson’s RUN metric of 16 for the same reason. Then I collected the quarterbacks since in college in 2004 that made those cutoffs. I selected quarterbacks that made the cutoffs for two college seasons in both qbr and run. That table is in the spreadsheet as ‘Since 2004 qbr run’. Both Jayden Daniels and Caleb Williams made the list. The NFL successes on the list are Jalen Hurts and Lamar Jackson. Most of the others would be considered busts. So not convinced on that small sample if the paper premise is true.
Anyways I collected NCAA stats on the quarterbacks on the list for comparison. That is spreadsheet qbr_run. Of the quarterbacks I see a similarity between Jalen Hurts and Jayden Daniels. Hurts really lucked out getting drafted by the eagles. He has a strong offensive line and top receivers. Others on the list were mot so lucky. Marcus Mariota had a poor offensive line, injuries, a system that did not fit with him and instability at coaching staff. Johnny Manziel at the time had work ethic issues and was not into doing the hard practice to be great. From what i’ve read on Justin Fields, he has issues with seeing the field, processing his reads fast enough, is too slow, and he holds onto the ball far too long. Whether these have always been there or are rookie issues that a good coaching staff can correct is debatable. Anyways there is a spreadhseet that compares the stats on these qbs.
From that I decided to extend the stat comparison to look at the top quarterback prospects in the draft. That is sheet ‘ All Prospects Weighted’. The table on the left on that sheet is the FBS rankings for the Career leaders as listed on ncaa.org. On the right table I list their statistic data and do a comparison highlighting the top three in each category with a color code. Giving points for each color code these are summed and weighted in each major grouping – passing, running and offense/scoring. I used statistics that are independent of quantity or else I created a per game associated statistic. The weights givin to each category can be adjusted by a weight column to the right of the table. Each category is summed for each prospect at the bottom.
The prospect table just reinforces the general held belief on the top three prospects. I slightly increased the passing weight as that is most important. Daniel Jones is included in the table using his ncaa statistics.
I like both tables used together. For instance Jordan Travis had some good sums for running and offense but nothing in the passing group. But looking at the left table you could see that he had good rankings on quantity independent statistics of Passing Yards per Attempt and Passing Efficiency. That is why I would be interested in him on day three if the Giants stay with Jones and do not pick a QB early.
QB-statistical-analysis spreadsheet download
The weighting was added to give more weight to the passing stats. without the weight the more number of offense/scoring stats was taking over when summed and decreasing the importance of the more critical passing stats. I equate any group’s (pass,rush,score) highest value to the weight and adjust the other group values acordingly. so as set now the passing stats are worth 40 points out of a 100 and the other two categories are worth 30 each. You can adjust that weight if you download the file.
