Identifying top rushing attacks year over year in college fantasy is where you can gain an edge against both ADP and your respective leaguemates. Let’s see if we can find some of those edges using year-over-year statistical production and offensive trending.

Neutral Game Script Rush Rate and Fantasy Production

The first assumption I made (wrongly) was that there was a strong correlation between the neutral game script rush rate (defined as the percentage early down of rush plays compared to all plays when win probably between 0.2 and 0.8) and RB1 production. What’s important about using neutral game script rates is to show offensive and play-calling intent. When the game is close how does the offense want to play?

The sample here was 648 teams over the last five years. Running a simple linear regression (trying to predict the relationship between fantasy PPG and rush rate), we found that the R-Squared was only 0.0002, or translating, 0.02% percent of the point per game production can be explained by neutral game script rush rate of the respective offense. Essentially, there’s virtually no relationship between these two variables when using a linear regression model.

Using Campus2Canton’s Team Data, analyzing RB1 PPG production compared to Neutral Game Script Pass Rate.

I was surprised by this, to be totally honest. I assumed a higher rush rate resulted in more opportunity for the lead back, which ultimately led to better fantasy outputs. That was not the case, and even when we ran for both RB1 and RB2 production, our R-Square was only 0.16%, meaning production is unexplained by rush rate. Even splitting work between the RB1 and RB2 gave us a slightly better R-Square but at the margins. I have a handful of theories on why, but I think there’s an argument to be made that higher rush rates are less creative offenses or at least more conservative. Long-term, this results in fewer points and points per opportunity on a drive-to-drive basis. Regardless, there’s no strong correlation between neutral game script rushing percentage and rushing production. 

Why I chose this specific metric matters as well. Neutral game script rates are mostly about intent. How do offenses want to run when the game is not out of reach? This speaks to playcalling intent, and over the course of the five-year sample, it allows us to have enough data to make broader scale determinations. The calculation we’re using to identify RB1 production also helps negate the concerns with neutral game scripts, as when games do get out of hand, backups (non-RB1s) see more production.

A Quick Multiple Linear Regression

Relying on rush rate as a standalone value was a bust. I expected it to provide some level of correlation, but there is none. The next step was to run a multiple linear regression to see if we can identify some of the variables that can help explain running back production (the dependent variable), at least ones that prevent multicollinearity. Is there a way to perfectly prevent multicollinearity in this case? I don’t think so, especially when variables like offensive line performance are likely related to neutral game script rush rate (i.e., good offensive line play incentivizes more rush attempts.)

Running a multiple linear regression based on key variables assumed to be related to rushing production.

The variables chosen (below) were chosen after rigorous testing of dozens of others, attempting to find ones that did not display direct colinearity but gave promise as true unrelated independent variables. Next, testing for multicollinearity using VIF (Variance Inflation Factor), we found that the VIF for each variable tested as follows. VIF is calculated by treating each independent variable as a dependent variable and testing each, the calculation being: 1 / (1 – R^2).

Neutral Game Script Rush Rate: 1.114

Plays Per Game: 1.093

O-Line Yards: 2.523

Second-Level Yards: 2.106

Power Success Rate: 1.3756

Recruit Rank: 1.142

Mostly, there is some level of colinearity, but the only concerning number is offensive line yards and depending on a penchant for risk and comfort level with colinearity, 2.52 is a fine number that shows some level but not enough to be entirely problematic, especially given the sample size. Using these variables, the R-Square is .1345 (13.5%), a substantially better number than relying on just neutral game script pass rate, but still below what we’re looking for. These numbers, on aggregate, tell us that there isn’t a single bullet(s) using metrics that would traditionally attempt to identify high-end running back production. 

Finding the Secret Sauce

Let’s start with the obvious. Better offenses produce better fantasy production. Again, using a simple linear regression with RB1 PPG and Rush/EPA per play. As a singular variable attempting to explain production, our coefficient of determination is 0.146 or 14.6%. That’s still very good. Offenses with high expected points added in the rushing game produce good running backs. Somewhat of a chicken/egg question, but this is just stating the baseline obvious: good offenses = good production. Directionally, that’s what we’re looking for…duh.

So we know offenses that are traditionally good at producing efficient/explosive units, evidenced by a high EPA/play, are better at producing running backs. Chunk plays matter and, on a per-carry basis, increase points per opportunity. The kicker here is that Rush/EPA is substantially more important than overall offensive EPA/play. There is little correlation between the higher EPA/play overall (0.02%). Pass efficiency matters much less to running back production – at least on a per-play basis.

Opportunity matters. Once again, duh. Looking at the backfield rushing dominator, we found that the R-square for the relationship is 0.338, meaning 33.8% of the variation in running back ppg performance can be explained by the backfield rushing dominator. More carries compared to the rest of the room is obvious but important. Looking at the overall dominator (26.9%) or backfield receiving dominator (9.7%) actually provides worse outcomes depending on the offense. 

However, it matters more for Power 5 running backs with a .435 R-square value than the 0.251 R-Square value for G5 running backs. This distinction essentially means that the percentage of backfield opportunity matters A LOT more to Power 5 RB1 outcomes than to Group of 5 RB outcomes. If those touches are more valuable, you can get by with a lower opportunity share.

Profiling the Teams

Over the last five years, a profile of teams who continue to put good running back production out at the RB1 spot emerges. Leading the way is Arizona State. Herm Edwards, a conservative (too conservative) coach, relied heavily on his RB1, evidenced by a nation-leading 85.6% backfield dominator and a future 3rd round pick in Rachaad White for two years. I don’t want to focus on the Sun Devils because they hired a new head coach and because the reliance on the RB1 was so far above average they were a significant outlier.

Using the Campus2Canton Team Tool to average the production of RB1 from 2018 to 2022.

However, clear patterns emerge in production when looking at the top 24 (RB2) teams over the last five years. First, these teams at least rely on the RB1 at an above-average rate. Utilization of the RB1 (backfield dominator rating) averages 49.4% nationally. The top teams are almost 57%, over 8% higher than their counterparts. As mentioned before, rushing production has a stronger correlation to points per game than receiving production, and with a 60% backfield rushing dominator, they outpace their counterparts again by 9%.

Again, a lot of this is intuitive. A higher opportunity share means more points per game. However, some teams are better at this than others, and the top 25 here are a good example of which offenses to target. For example, Minnesota under PJ Fleck loves rushing the ball and giving a substantial workload to the RB1, nearly 65% of the RB carries. Clemson and Alabama both produce quality rushing production by virtue of focusing on a single back AND being in better game scripts down the stretch than most other teams.

Using the Campus2Canton Coach Tool to average the production of RB1 from 2018 to 2022.

Team trends over time tell a good story as they, generally speaking either have the same coach for a four-to-five-year window or if they do switch coaches, keep a similar mindset towards offensive production. However, while that’s not always the case, looking at head coach production for running backs matters too. Using the same sample size (five years), above are the coaches, by year, who provide(d) the highest points per game in the sample.

Players to Target

Based on the data above, we have a handful of players worth targeting. First, whoever emerges in the UCLA backfield will probably be an excellent option. Chip Kelly has had an RB average over 16.8 points per game every season since 2018 despite a low backfield dominator (38%). However, the UCLA offense is efficient and runs more plays than average, ranking 13th in plays per minute (2.66). TJ Harden outpaced Carson Steele in the spring and given the history, he’s going way too late in CFF drafts.

Fresno State, a team that’s averaged over 19 fantasy points per game and returns Jeff Tedford in his second stint, should not go under the radar. Kalen Deboer was part of those productive years and was the offensive coordinator under Tedford. In 2023, Malik Sherrod was used in both phases of the game and will likely outperform his RB47 CFF rank.

Further, down the list, two G5 coaches/teams stand out. Jason Candle’s Toledo has traditionally relied on an RB1 both with Bryant Koback and before him Terry Swanson. Jacquez Stewart is probably the RB1 who could take on a larger role in 2023, although there isn’t a clear path to a majority of touches. Still – knowing what Candle has delivered in the past piques my interest.

The “Eli Drinkwitz RB” should probably be a thing by now, and I never hear it discussed as such. Tyler Badie was one example, but going back to Appalachian State, Drinkwitz has provided productive options. Last year they were never able to find a true bell cow. Cody Schrader emerged as the primary option but as a committee lead rather than a true lead back. I expect that to continue, but in dynasty and C2C formats, Tavorous Jones remains interesting and is viewed as the future of the position.

Honorable Mentions: Matt Campbell/Iowa State, Jamey Chadwell/Liberty, Jeff Traylor/UTSA

Final Thoughts and Takeaways

First, the biggest takeaway is not conflating intentional (high neutral game script rush rate) rush-heavy offenses with highly productive RB1 seasons. In fact, there is essentially no correlation between the two based on the analysis done above. Just because an offense runs at a higher pace does not portend fantasy production. 

Second, opportunity remains king, but even more so among Power 5 programs. There is a higher correlation between the percentage of RB touches going to the RB1 among the better teams in the nation. Each opportunity amongst better (Power 5) opponents is worth slightly less than their G5 counterparts. Over time this is expected to change with realignment and conference schedules changing. For now, we know that Power 5 programs require a higher share of the backfield to achieve high-end production. 

Finally, identifying the teams and coaches who consistently produce RB1s at a high rate remains relatively consistent over time. It is still the best way to identify what players and situations to target. These teams utilize their primary running back at a higher rate than their counterpart and often hail from offenses with more plays per game or an efficient rushing attack.

Get Our Newsletter

Get notified periodic notifications about our content and future subscription deals.

You May Also Like

College Fantasy Football – Week 4 Waiver Wire Pickups

Could a #GoCougs receiver be THE MAN in Wazzu? Josh Kelly is one of the waiver adds we recommend in Week 5 as you chase the TITLE in CFF!

College Fantasy Key Takeaways: Week 4

After a week 4 full of action, Austin (@devydeets) breaks down his thoughts on the slate and how it impacts fantasy teams moving forward.

Flip the Switch Ep. 1: When is it Time to Buy or Sell in C2C?

My NFL team is old and it sucks. College team is lurking in fourth and good, but not good enough. Here’s how I attacked this strategy and trades made over a three-hour period to fully arm my college squad!