One of the biggest mistakes we can make when trying to project upcoming football performance is relying too heavily on past stats.

That's especially true if we're looking at only one season -- just 16 games (at most). A lot can change in such a small handful of games. A pass or two that bounces off the fingertips in the end zone or a shoestring tackle that prevents a goal line score can really inflate or deflate touchdown totals.

That's why -- outside of a select few players -- touchdown totals can be pretty drastically different year over year.

One way to try to find a more realistic touchdown outcome than looking solely at a player's tally from a past season is to compare the number of touchdowns he scored to his Reception

This shows that a receiver with a Reception NEP of 100.00 should score about 7.5 times. A player with more than 7.5 scores and a low Reception NEP is going to be flagged as someone who should score fewer times in 2018 because of the model.

Backtesting: 2017 Red-Flag Candidates

Let's backtest. There were conveniently 20 receivers who "should have" had at least 1.5 fewer touchdowns based on our model -- if we look at players who had at least 25 catches in 2017 to monitor.

Just one of the 20 scored at a higher rate (in terms of catches per touchdown) in 2017 than in 2016, and that player was 20th on the list (with just 1.54 expected touchdowns to lose), and he's also an elite touchdown scorer who didn't play many games in 2017.

Yup. Just Odell Beckham improved his reception per touchdown rate in 2017, and part of that can be attributed to the fact that he had only 25 catches to examine.

Another elite player, Antonio Brown, still couldn't build on or replicate his 12-touchdown performance from 2016 in 2017, despite the fact that he led the league in Reception NEP (131.06) this past season and built on his Reception NEP from 2016 (109.70). Simply put, double-digit touchdown seasons are rare, and even the league's best players can't consistently find the end zone at that high of a rate.

You may also note that not every player had terrible touchdown totals. Brown still had 9. Tyreek Hill still posted a decent touchdown number despite being a regression candidate.

And the primary regression candidate, Davante Adams, still scored 10. Adams' 90.08 Reception NEP in 2016 meant that his 12 touchdowns should've been 6.68, so he did dip, and his catch-per-touchdown rate worsened.

Okay, so the method pretty much checks out, and these receivers are designated as the red-flag options for 2018.

2018 Regression Candidates

There are 16 receivers who outperformed their expected touchdown total by at least 1.5 scores this season. Here they are.

At the top, we have Will Fuller, who scored touchdowns like they were being given out on Halloween. His 42.39 Reception NEP means that he should've scored fewer than 3 touchdowns. He had more touchdowns (7) than red zone targets (5).

The next three receivers on the list -- Michael Crabtree, Sammy Watkins, and Jarvis Landry -- all changed teams, so that's a wrinkle to be factored in for certain. However, each outperformed their expected touchdowns by at least 3.8, which is a big number for our purposes. Remember how a Reception NEP of 100.00 should net around 7.5 touchdowns? Well, none of them even got close to that Reception NEP but scored at least 8 times.

Crabtree, at least, had 11 targets from inside the 10-yard line, and that type of usage will generate outliers via this method. End zone threats can outperform their Reception NEP when it comes to scoring touchdowns. Watkins, on the other hand, had 10 red zone targets and 5 from inside the 10, yet just one of his scores came from outside the red zone.

Landry had a whopping 14 targets from inside the 10 and 23 from inside the 20, both third-most in the league. Usage like that can lead to big touchdown numbers, but whether he gets that type of goal line work with the Cleveland Browns remains to be seen.

Amari Cooper, drafted as the WR20 on average (via PPR average draft position data [ADP] from FantasyFootballCalculator) for 2018, does lose Crabtree's touchdown ability but gains Jordy Nelson. Nelson scored 6 times in 2017 despite a Reception NEP of just 47.72 (meaning an expected touchdown number of 3.19). Both are due to regress in the touchdown column, but the change in variables matters a lot.

DeAndre Hopkins' 13 touchdowns helped him rack up the WR2 season in PPR formats, but he should have scored 9.68. That's similar to Antonio Brown's number in 2016, and he dropped to 9 touchdowns this past season. He'll cost a first-round pick this year, per ADP, and the anticipated dip in touchdowns could result in paying top dollar for someone expected to regress.

Davante Adams, the WR7 via ADP, again made the list but could just be a player who perennially outperforms his expected touchdown tally, given his role in an efficient passing offense. Still, he'll run you a mid-second-round pick.

Alshon Jeffery, the WR20 in terms of ADP, should've scored about 66% of the touchdowns he actually did. Even with the expectation that he'll score fewer than nine touchdowns in 2018, the draft cost is low enough for him to outperform that ADP.

Again, the goal isn't to blacklist any of these receivers but rather to keep in mind that their touchdown numbers should dip this season. We should account for that when drafting them in 2018.