Identifying 2023 Touchdown Regression Candidates: RBs (Fantasy Football)
If you won a fantasy football championship last season, you likely rostered one or multiple players who scored double-digit touchdowns. Whether it was Jamaal Williams, Stefon Diggs, or George Kittle, their endzone production often gave you a weekly advantage. As we head into the 2023 season, is it fair to assume that they will repeat that performance? Is it possible their touchdown efficiency regresses to the mean? To answer both of those questions, we can use a metric called Expected Touchdowns (or xTD) to determine if a player’s endzone production is sustainable year over year. In this article, I will outline how this metric is calculated and how it can be used to determine touchdown regression candidates (positive and negative) at the running back position!
Play-by-play data used for this metric was provided by nflfastR.
Calculating Expected Touchdowns
To calculate Expected Touchdowns, I created an xTD model that uses historical play-by-play data to determine the likelihood of a player scoring on any given play. This is based on a variety of variables such as the down, distance to the goal, and the type of opportunity (rush attempt, pass attempt, or target). For targets and pass attempts, I also factor in the depth and direction of the throw. Once we factor in each of these variables, my model will calculate an xTD value for each opportunity on a scale from 0 to 1.
For example, we can take a look at the following play for Austin Ekeler against the Denver Broncos in Week 6:
- Type of Opportunity: Rush Attempt
- Yards to Go: 2 Yards
- Yardline: Denver’s 6-Yard Line
- Down: 3rd Down
If we plug this play into my Expected Touchdowns model, I arrive at an xTD value of 0.1309. This tells us that, historically, around 13% of rush attempts in this specific scenario would have resulted in a touchdown. Because Ekeler scored on this opportunity, we calculate his Touchdowns Over Expected (TDOE) as follows:
- Touchdowns Scored: 1.00
- Expected Touchdowns: 0.1309
- Touchdowns Over Expected: +0.8691
As you can see above, TDOE is the difference between the touchdowns scored and their xTD. By contrast, if Ekeler did not find the endzone, his TDOE would have been -0.1309. Keep in mind that this metric is calculated for every opportunity, of which only a small percentage will result in a touchdown. In fact, this was the only touchdown that Ekeler scored against the Broncos despite receiving 30 total opportunities. He finished that week with 1.07 xTD (the total value of his 30 opportunities) and a negative TDOE of -0.07.
As you may have already surmised, Expected Touchdowns are synonymous with opportunity and usage. The more opportunities a player receives, and the closer they are to the endzone, the more likely they are to score – leading to a higher expected touchdown value. On the other hand, Touchdowns Over Expected (TDOE) is a metric that highlights efficiency, which is both volatile and subject to regression.
Ekeler’s Week 5 and 6 performances are the perfect example of regression. Against the Browns in Week 5, he scored twice on only 20 opportunities – resulting in an efficient +1.69 Touchdowns Over Expected. Averaging nearly two touchdowns above xTD is usually unsustainable, and a sign that Ekeler would likely regress the following week. And as we outlined in our initial example above, he regressed significantly to the mean against the Broncos, totaling a negative TDOE value (-0.07) in Week 6.
Interpreting xTD and TDOE
After calculating the xTD value for every opportunity, we arrive at the chart above. The line represents the projected point of regression for each player. The farther away a player is from the trendline, the larger the potential dropoff would be if they regressed to the mean. On the other hand, players that are located below the trendline could improve their efficiency the following year, resulting in more touchdowns scored relative to their total opportunities. While a variance of two or three touchdowns may seem minimal, keep in mind that there are rushing and receiving yards associated with those plays as well.
With that in mind, how likely is it for a player to regress to the mean? If we dig a little deeper, running backs who scored above their xTD value experienced an 86.2% decline in Touchdowns Over Expected the following year. As for players who finished their season with a negative TDOE, their efficiency improved by an average of about 93%. In other words, regardless of where a player is located on this chart, they are very likely to regress toward their Expected Touchdown value this upcoming season.
Negative Regression Candidates
Derrick Henry is another year older and has shown few signs of slowing down, ranking as the RB3 in half-PPR points per game (17.9) last season. Even more impressive, he accounted for over 45% of the team’s total opportunities as he once again operated as a focal point for the Tennessee Titans. As a result, it should not be a surprise that Henry was also a top-12 running back in Expected Touchdowns at 9.2 given that he totaled 390 opportunities in 16 games. However, there is plenty of room for regression as he also scored +4.8 touchdowns over expected last season. Interestingly, we have already seen Henry regress to the mean after averaging an even more efficient campaign in 2021 (+6.4 TDOE pace). And if these trends continue, Henry projects to score even closer to his xTD value as he approaches his age-29 campaign in 2023. Regardless, Henry’s tremendous usage will continue to carry his fantasy value. So as long as the Titans are willing to feature him in their offense, he should remain an RB1 even if his efficiency takes a slight dip.
Tony Pollard will enter the year as the unquestioned RB1 for the Dallas Cowboys as they opted to release Ezekiel Elliott earlier this offseason. And after a campaign in which he finished as the RB8 in half-PPR points per game (14.3), Pollard projects to be even more involved this year – especially as a rusher. As you can see above, he was also the RB2 in touchdowns over expected, scoring +4.7 above his xTD. For comparison, in 2021, Pollard was much less efficient with a -4.51 TDOE. While both of those numbers present a drastic swing, I expect Pollard to finish somewhere in between this season, likely scoring much closer to his expected value. The good news is that his xTD of 7.3 is bound to improve as he will likely take on a more significant role in Brian Schottenheimer’s offense. For context, our Ultimate Draft Kit has Pollard projected for 304 opportunities, which almost guarantees that he finishes as an RB1 even if his TDOE regresses to the mean.
Jerick McKinnon was the definition of a league-winner as he finished as a top-16 running back in four of the final five games of the fantasy season. However, his performance screams touchdown regression as he scored nine of his ten touchdowns in his final six games. For context, that equated to an impressive, yet unsustainable +5.7 Touchdowns Over Expected from Weeks 13 to 17. While McKinnon projects to once again be involved in the passing game in 2023, his usage offers plenty of volatility as he likely will not average enough opportunities to provide consistent fantasy value. For best ball leagues, McKinnon will remain extremely valuable as you will always benefit from his multi-touchdown games. However, in standard redraft leagues, he will be much more difficult to trust unless he strings together another five straight games of elite fantasy numbers.
Positive Regression Candidates
Alvin Kamara’s fantasy value is somewhat clouded with uncertainty as legal troubles could force him to miss part of the 2023 campaign. However, there is room for optimism in his production if he remains the focal point of a revamped Saints offense this season. As you can see in the chart above, Kamara was one of the most inefficient touchdown scorers in 2022, finishing with 4.62 Touchdowns Below Expected. In other words, he left about 30 fantasy points on the table, scoring only four times on 300 opportunities. With the addition of Jamaal Williams, I expect Kamara’s rushing usage to decline this season. However, if his touchdown efficiency regresses closer to the mean, he could still provide RB2 value despite likely averaging a smaller share of the team’s offense. Regardless, his fantasy production clearly hinges on his availability as a suspension could be on the horizon for Kamara.
After a foot injury forced Travis Etienne to miss the entirety of his rookie year, he made a sizable impact in 2023, ranking as the RB23 in fantasy points per game. And while he was relatively efficient from a yardage standpoint (5.7 yards per touch), he significantly underperformed as a scorer, finishing 4.6 Touchdowns Below Expected. To put that into perspective, had Etienne scored anywhere close to his xTD value, he would have instead finished as the RB15 in points per game. Keep in mind, this is likely understating the fantasy impact of regression as he would have gained more yards as well. So as we head into the 2023 season, I would expect Etienne to be more efficient from a touchdown perspective, potentially making up for any opportunities lost to Tank Bigsby. And if Etienne were to retain or even exceed his 265 opportunities from 2022, he has the potential to finish as a borderline RB1 in what projects to be a more explosive offense with Calvin Ridley joining the team.
Jonathan Taylor had a tumultuous campaign in 2022, battling through multiple ankle injuries that eventually forced him to miss the final three games of the season. But despite dealing with several injuries, he still averaged an impressive 35% opportunity share for the Colts. This equated to an elite 23.1 opportunities per game before his season-ending injury. As a result, Taylor was on pace for an elite Expected Touchdown Value of 13.4, which would have been tied for RB1 right next to Austin Ekeler. However, as you can see in the chart above, he was also on pace to severely underperform as he finished 3.9 touchdowns below expected in only 10 full games. With all signs pointing to Taylor returning fully healthy, I expect him to once again be the lead running back for the Colts. Assuming his usage remains mostly unchanged from last season, and his touchdown efficiency regresses to the mean, Taylor has the opportunity to once again finish as a top-three running back in fantasy football.