Touchdown Regression: What It Is and How to Use it for Player Prop Bets, Fantasy Football

Football, by nature, is a chaotic sport.

And for those of us who care about the statistical outputs of NFL players, we don't always get the results on paper that we may expect.

With 22 players on the field every play, there are, frankly, dozens of reasons for that variance every single NFL down.

One of the easiest ways to think about a misalignment between data and expectations is when it comes to touchdown totals.

Sometimes, a player will have a strong red zone role and just not get into the end zone at a high rate in a given season. Or a player starts ripping off long touchdowns out of nowhere and is a top-10 touchdown scorer on modest volume.

Logic would dictate that, over a longer sample, touchdown rates tend to even out. But in order to get some facts on paper, I first want to run through how big a factor touchdown regression can (and probably should) play into our decision-making processes.

What Is Touchdown Regression?

Touchdown regression, effectively, is the idea that a stat like touchdown rate (touchdowns per pass attempt, per carry, per target in my case) may fluctuate over certain stretches but will tend to level out close to the league average over the long-term.

Does this mean that every player should be scoring, say, five touchdowns a year no matter what? Of course not.

You have to be on the field and get the ball in your hands in order to score touchdowns, so usage is a massive indicator for expected touchdown rates -- and actual touchdown outputs.

Is Touchdown Regression Real?

Yes, it's quite real as you'll see in a minute, and while it's not guaranteed that a player who overachieved in one season will underachieve the next season, the rates at which players regress (and progress) are high enough that we'd be a bit stubborn to ignore the long-term trends.

What Goes Into Touchdown Regression?

This one depends on who you ask.

There are a lot of different ways to generate expected touchdown (xTD) numbers.

You can simply match yards to touchdowns and draw a trendline. You could do that with EPA numbers and touchdowns, too. Or first downs. You can account more for red zone opportunities or even look at field position locations for every touch and derive an xTD number that way.

I've been looking into touchdown rates for years now, and there's no one way to get on the right path, so I just started to use it all (assuming there's a strong relationship between a stat and touchdowns) and average it out.

So, I like to use myriad metrics for each of the four positions and get a blended view of expected touchdowns.

Touchdown Regression Sample Sizes

You can skip this part if you want, but in order to study touchdown regression, we need a season (n) and a follow-up season (n+1) to see if a player's touchdown rate increased, decreased or stayed the same in order to draw any conclusions.

I've gathered all QB, RB, WR, and TE seasons from 2016 to 2024 to do just that.

I'm looking at players with the following thresholds met in consecutive seasons:

• Quarterbacks:
  • Passing TD Regression: 100+ pass attempts
  • Rushing TD Regression: 35+ rush attempts
• Running Backs:
  • Rushing TD Regression: 100+ rush attempts
  • Receiving TD Regression: 50+ targets
• Wide Receivers:
  • Receiving TD Regression: 75+ targets
• Tight Ends:
  • Receiving TD Regression 50+ targets

I'll also be using a few terms to simplify the analysis:

• Overachiever: A player with a touchdown rate (TD%) higher than his expected touchdown rate (xTD%) in a season
• Underachiever: A player with a TD% lower than his xTD% in a season
• Outlier Overachiever: A player with a TD% at least 1.0 percentage points higher than his xTD% in a season
  • For example, a touchdown rate of 4.5% and an expected touchdown rate of 3.4%
• Outlier Underachiever: A player with a TD% at least 1.0 percentage points lower than his xTD% in a season

Let's dive into touchdown regression a bit more by position.

Touchdown Regression By Position

Quarterback Touchdown Regression

Quarterbacks have two ways (commonly) to score touchdowns: through the air and on the ground. The vast majority of QB touchdowns are passing touchdowns, of course, so we'll start there.

Using more than 40 stats to test, I find a lot of simple stats do the best job at explaining touchdowns:

• Completions
• Passing Yards
• Successful Drop Backs (plays with positive EPA)
• First Downs

We can use these numbers to derive a measure of expected touchdowns as a passer.

Now, it's only of use to us if these numbers tend to do a good job of listing out expected touchdowns, of course. If these spit out xTD numbers in the hundreds, then we don't have a good baseline.

As for that question, the average difference between total passing touchdowns and expected passing touchdowns among our sample (again, passers with 100+ pass attempts from 2016 to 2014), is -0.01. It's hard to do much better than that, so that checks out. (Put simply: xTD numbers are -- over the full sample -- very close to actual TD numbers.)

But now to test to see if the regression part of this all actually exists.

As for that question, I think we have a good answer.

We've seen 118 QBs overachieve their xTD% to any degree and have a qualified follow-up season.

The next season, 72.9% of these overachievers saw their touchdown rate drop the next year, and of those, they dipped an average of 0.9 points.

Among the 52 outlier overachievers (i.e. those with an xTD% differential of +1.0% or greater), 90.2% saw their touchdown rate drop the following year -- by an average of 1.5 points.

So, basically, 9 out of 10 quarterbacks who really have a great season with regards to touchdown "luck" experience a dip in their touchdown rate the next year.

We see similar numbers for the underachievers in terms of touchdown growth.

Here's a quick chart summing up the high-level results:

This shows that 88.1% of the big outliers in the negative sense (or the QBs who got unlucky with their touchdown rate in a year) see a positive increase to their TD% the following season.

Let's move to the ground. The following stats are used for rushing xTD.

• Carries
• First Downs
• Successful Rushes (carries with positive EPA)

Here are the results.

I know these samples are small (the cutoff is 50-plus carries for each consecutive season), but it's more of the same, where we see roughly 75% of overachievers drop their TD% and roughly 75% of the underachievers raise their TD% the next season on the ground.

Running Back Touchdown Regression

Running backs also score two different ways (rushing and receiving), so we'll need to dig into both.

But we'll naturally start with rushing scores.

The stats used for blended touchdown rates are as follows:

• Carries
• Rushing Yards
• Rushing First Downs
• Red Zone Carries
• Successful Rushes (carries with positive EPA)

And getting right to it, here are the splits for the four buckets of performers.

As we can see above, 76.5% of overachievers saw their touchdown rate drop the following year, and 71.1% of the underachievers increased their touchdown rate the next year -- a number that jumps to 77.8% for the outlier underachievers.

I think most notably here pertains to the outlier overachievers. As we can see, 88.6% of them had a TD% decrease the following year plus an average drop of -2.2 percentage points.

That'd be a drop of over four touchdowns for a 200-carry back year-over-year.

What about for receiving scores at the position?

(Note: not all values will add to 100%. This is because there are numerous instances of RBs going consecutive seasons with no receiving scores and thus maintaining the the same touchdown rate, signaling neither an increase nor a decrease for the follow-up season).

Congruent with other findings to this point, we see trends going as expected.

Most notably (to me) are two things.

First, underachievers don't increase at a rate on par with other touchdown categories (e.g. rushing touchdowns). The band of receiving touchdowns is tight for RBs, and no running back has double-digit receiving scores from 2016 to 2024, and only six guys have had more than six. Therefore, it's not that uncommon for a running back with a decent receiving workload to get blanked in the touchdown column in consecutive years.

In a sense, receiving touchdown regression is a little "less real" in terms of buy-low spots.

The second thing, though, is that touchdown regression is quite real from the other end of the spectrum. Outlier overachievers tend to fall off quite a bit (-4.1% on average). That's an aggressive downswing for RBs who are coming off of efficient receiving seasons.

Wide Receiver Touchdown Regression

While wideouts can get some rushing scores, the vast majority of WR touchdowns since 2016 (around 97%) have come through the air, so that's where we'll focus our time.

With regards to touchdown expectations for WRs, these stats carry weight:

• Targets
• Receiving Yards
• Receiving First Downs
• Receiving EPA
• Red Zone Targets

Let's get to the results.

Okay, you know the drill by now: 75.7% of overachievers and 85.3% of outlier overachievers regressed in terms of touchdown rate in their follow-up season.

The 2.7-point drop for outlier overachievers is quite drastic and a great reminder of the volatility that exists with receiving scores (which we see across RB, WR, and TE).

Conversely, 68.4% of outlier underachievers saw a positive correction on their touchdown rate the following season, doing so by an average of 1.3 points.

Tight End Touchdown Regression

Touchdowns are vital for tight ends from a fantasy football standpoint, as the yardage numbers are suppressed for virtually everyone outside of a few of the elite each year.

The same stats are on the list for tight ends as we get with WRs -- but they are weighted differently based on the regression analysis.

• Targets
• Receiving Yards
• Receiving First Downs
• Receiving EPA
• Red Zone Targets

Let's get right to it.

These are some juicy mathematical red flags.

A 93.3% decrease rate for outlier overachievers and an 80.8% decrease for all overachievers is quite substantial, especially at -3.5 and -2.5 clips, respectively.

The correction rate for the underachievers isn't necessarily as strong as we may expect (58.2% versus a 40.5% further decrease the next year). However, the outlier underachievers tend to correct in roughly two-thirds of the instances (67.2%).

This is yet another reminder of the variance that exists with receiving touchdowns.

How to Use Touchdown Regression

Okay, so we should be able to trust the math and buy into touchdown regression by now, but how does it come into play for fantasy football and player props?

For season-long touchdown player props, we still need to blend projected volume and xTD% numbers to ensure that anticipated changes in touchdown rate are significant.

That is to say: a player who underperformed his TD expectations isn't an automatic over target if his role is set to change drastically or the touchdown prop is already listed at a high mark.

Matching up expected touchdown rates and NFL player projections is crucial.

For daily fantasy football and in-season touchdown props, you can keep an eye on expected touchdown numbers and matchups each week to find potential buy-low options in touchdown scorer markets.

For season-long fantasy football, we can really leverage this type of data leading into the year in drafts (targeting positive touchdown regression candidates and downplaying negative regression touchdown candidates).

Touchdown Regression Candidates for 2025

I'll be further exploring these findings to pinpoint players who may be due for noticeable touchdown regression in 2025 as well as maintaining expected touchdown numbers for the 2025 season once it begins.

Be sure to check back for future updates.

Futures Day is coming August 26th. Learn more here.

The above author is a FanDuel employee and is not eligible to compete in public daily fantasy contests or place sports betting wagers on FanDuel. The advice provided by the author does not necessarily represent the views of FanDuel. Taking the author's advice will not guarantee a successful outcome. You should use your own judgment when participating in daily fantasy contests or placing sports wagers.