The Last Time You’ll Have To Think About The D/ST Position (Fantasy Football)
The D/ST (defense and special teams) position is, in addition to Kicker, one of two common ‘auxiliary’ pieces in your fantasy football roster. To put it politely, these positions don’t get that much respect. You’ll spend all season analyzing WRs and RBs, strategizing to snag one of the breakout TEs, and poring over QBs for streaming options (especially in Superflex leagues). Then, a few hours before kickoff, you’ll set your D/ST and Kicker (probably at random).
In this article, we’ll be diving deep into the D/ST. I think defense is the more interesting of the ‘auxiliary’ slots because it can help win you a week: it’s certainly uncommon, but not that rare for your defense to put up 20+ points, or even breach 30 in a really special game (while kickers can have stellar performances, there just isn’t that much fantasy upside). Furthermore, with the fantasy football playoffs coming up, it might be useful to start thinking ahead about which units will yield you the juiciest matchups. In addition to this article, you can check out the Stream Finder tool on The Fantasy Footballers website to help you decide which defenses to target.
With this in mind, what’s the best way to approach the D/ST position? As we enter the crucial part of the season, should you be eschewing depth at your skill positions and rostering multiple defenses? Or, on the other end of the spectrum, should you opportunistically stream the position from week to week? Fantasy football is a complicated (very complicated) game. However, after doing this research, I am convinced that the D/ST position alone is not so complicated; I am convinced that all you need to know about the position can be found below. All data is from nflfastR, and usually through 2020 (so that we aren’t biased by the year we are in the midst of). Without further ado, let’s get into it.
I’ll be using default defensive scoring with a ‘yards allowed’ option added (similar to points allowed). To start, we can look at the average fantasy PPG per week by positional rank; that is, what the top D/ST will score on average, the second-best D/ST will score on average, etc.
Naturally, we see some skew here: since we are taking an average (and looking at the ‘number one overall’), the top ranked D/ST unit will score far more than the field. In this chart, we see that the top D/ST scores are just over 12 PPG, the second-best scores are about 10.75 PPG, etc. This is similar to what we’re seeing in 2021, with the New England Patriots and Buffalo Bills atop the position.
It’s important to note the drop-off in value: getting one of the top defenses can be very valuable, but there is not that much of a difference further down the stack. For example, the difference between the first and fourth defense is 2.0 PPG, but the difference between the fifth and eighth defense is less than 1.0 PPG! This will be important to remember in later analysis.
We can also look at a fun historical chart of D/ST points scored (which measures how good a defense is) and allowed (which measures how good the offense is, since this is D/ST points scored by the other team) for teams since 1999. The Ravens and Steelers, known historically as defensive stalwarts, stick out on the right side, whereas offensive juggernauts like the Colts, Saints, Chiefs, and Packers round out the bottom of the chart (low amounts of D/ST points allowed to the other team). The Patriots are elite in both phases, while the Browns have been, simply put, awful.
Projecting D/ST Performance
We’ve seen that top-scoring defensives can give your fantasy team an edge; let’s turn to the task of actually predicting how good defensive fantasy performances will be.
To do this, we will look at the current ‘form’ that the defense is in, which we will define as the most recent four weeks. This gives us a general sense of teams that are playing well going into a week. Since matchups are important for D/ST start-sit decisions, we’ll also consider the form that the opposing offense is in.
We can then look at crucial statistics – interceptions, fumbles, yards allowed, NFL points allowed, sacks, etc. – attributed to both the defense and the opposing offense over that four-game window. With these statistics, it’s straightforward to build a regression model to predict fantasy D/ST points scored. Here is the output, where the numbers are the regression coefficients:
Before getting into the numbers, let’s interpret the first row to make sure everything makes sense. The ‘Interception’ row tells us that, for every extra interception that a defense is forcing per game (in the past 4 games), the D/ST is expected to score .29 more fantasy points in the next game. For every extra interception the opposing offense has thrown per game (in the past 4 games), the D/ST is expected to score .44 more fantasy points in the next game.
Right off the bat, we can breathe a sigh of relief: all of the numbers make sense. The more interceptions, fumbles and sacks that a defense causes (or an opposing offense creates) over the last 4 weeks, the more D/ST points are expected this next week (positive coefficients). The reverse is true for yards gained and (NFL) points allowed. In addition, all of these numbers are highly significant, with the exception of ‘defensive fumbles forced’; this is almost significant but, intuitively, fumbles can be a pretty random stat.
What’s more interesting, though, is that the numbers in the offense column are larger than the numbers in the defense column! For example, each sack that the opposing offense has allowed in the past 4 games is nearly three times as impactful on D/ST points than each sack generated by the D/ST in that same window! This is the same effect for every statistic outside of ‘Points Allowed’: the opposing offense has a bigger impact on the D/ST point projection.
This points us to an interesting tentative result: opposing offense matters more than defense. We’ve certainly seen offensive prowess being more important in predicting offense output, but it’s a bit surprising that offensive prowess also is more important in predicting defensive output. Let’s look further with this heat map:
Each of the boxes here represents a group of 5 teams (or 6 teams in the middle boxes), while the number in the box represents the average D/ST points per game since 2012. So, for example, the top left box tells us that when a top 5 defense plays a bottom 5 offense, the D/ST scores 9.9 points on average; the bottom right box tells us that when a bottom 5 defense plays a top 5 offense, the D/ST scores 5.8 points on average. Remember, we are measuring ‘top 5’, ‘bottom 5’, etc. based on the most recent four weeks of performance going into a game.
This chart confirms what we saw in the regression model: the numbers, in general, increase more when we move from down to up (offense gets weaker) than when we move from right to left (defense gets stronger). Clearly, weaker offenses are more important than strong defenses. For example, this chart tells us that a bottom-5 defense playing a bottom-5 offense (top right) scores 1.2 PPG more than a top 5 defense clashing with a top-5 offense (bottom left). It also tells us that even starting bad defenses against terrible offenses (top row) is usually significantly better than starting any defense vs. a top-10 offense (bottom two rows).
Let’s now circle the wagons and think back to our D/ST scoring distribution. The top left box here – 9.9 PPG for a top 5 defense vs. a bottom 5 offense – is enough to score the same as the fourth-best D/ST overall. What sort of strategy can we glean from these results?
Remember that, in the chart above, ‘top 5’ means ‘top 5 over the past four weeks’. This designation is actually not that difficult to hit; in fact, in 2020, most of the league (18 teams) ended up as a top-5 defense between Weeks 5 – 17 (remember, this requires 4 previous weeks of performance):
What’s more, teams that crack the top-5 usually will stay there for a few weeks (with the exception of the Falcons and Cardinals, who snuck in at the end of the season). In addition, notice that the Rams and the Washington Football Team – two of the top three D/ST units last year – only came on strong at the end of the season.
The point here is that quality streaming options will usually be available. Unlike the skill positions, where fantasy teams roster significant depth (and the landscape is more top-heavy) you have a very good chance of finding one of these D/ST units each week on the waiver wire. How many people, for example, had rostered the Seahawks D/ST at the end of the season in 2020 after they started out the year as a defensive sieve?
Let’s see, then, how effective streaming defenses can be. We will go through each week and pull out all of the top 12 defenses that are facing against putrid (bottom 5) offenses; there are usually a couple per week, and since we are considering the top 12 defenses – a pretty generous amount – it’s likely that at least one is usually available in free agency. I did remove the Patriots in 2019 since, after the first couple of weeks of dominance, they were likely rostered in almost all leagues. Anyways, we can average these points and see how they perform (the ‘NFL’ logo) in 2019:
And in 2020:
In both cases, our ‘streaming strategy’ came in as the seventh D/ST overall. This isn’t great, but there are a couple of caveats:
- In both cases, the strategy was within less than 1.0 PPG of the fifth-best D/ST overall, and quite close to the sixth-best D/ST overall.
- The ‘streaming strategy’ never has a bye week; all of these other options do have bye weeks, which is a nuisance and can clog your roster if you hold onto the defense on bye
- Taking all of the ‘week-by-week factors’ into account (injuries, matchups, picking the best defense available, etc.) would likely slightly improve the average performance we presented here, potentially even producing an average that breaks into the top 5
All told, I think the implication is clear: you can likely get better than average D/ST fantasy performance by entirely streaming the position.
Many fondly remember the Patriots D/ST of 2019 (through, at least, the first half of the year). However, on most platforms, this wasn’t even a highly touted unit coming into the year, being drafted outside the top-10 at the position. The Rams, the best D/ST of 2020, mostly went undrafted! The Patriots and Bills, so far the top units of 2021 barely made it into the top-10 of D/ST ADP.
While having the top D/ST in the land is valuable, it’s very hard to predict who will come out on top. This fact, combined with the result that opposing offenses are more important in dictating D/ST scoring than the D/ST unit themselves, leads us to this simple result:
You should almost always be streaming your D/ST.
This is important, especially because I think it’s different from how many managers consider the position. The results in this article show how much D/ST fluctuates, both as the season progresses and with matchups. That means it is not advisable to stick with a defense through a rough patch even if they might be one of the best ‘real-life’ NFL units (good advice for the Patriots and Bills, two great defenses that face each other – both solid offenses – twice in December) – you can make your bread with shoddier defenses facing shoddy offenses. In addition, if there are other valuable options on the waiver wire (high-upside WRs or insurance RBs) you should almost never clog your lineup with multiple defenses. Just let your extra/underperforming D/ST go: you can get great performance by streaming the position!
Now, don’t hear what I’m not saying. If a defense is playing well and they have a stretch of good matchups, then by all means keep them on your roster. Further, if your defense is playing outstanding and putting up week-winning performances – you might very well have the #1 overall unit on the year and it might be smart to not release them (at least for the time being). However, in most cases, it is not worth it to stick with a great defense when they go against great offenses or to roster multiple defenses while your ‘starting D/ST’ goes through a funk.
Another edge case is something that Jason preaches as we approach the end of the season: if your team is doing well and coasting into the playoffs, it could make sense to roster a couple of D/ST with favorable matchups in the fantasy playoffs. Although this is basically the opposite advice of what we’ve been saying, it actually makes sense in the right situation. Of course, this approach only advisable if there aren’t other good options on the waiver wire (and if your team doesn’t need more depth or insurance at the skill positions). The upside to this strategy is that it capitalizes on the fact that opposing offenses are the most impactful for D/ST when it matters most: the fantasy playoffs.
Still, the main point remains: become a noncommittal D/ST manager. Disrespect the position. Sometimes I will even roster zero defenses early in the week while I’m waiting for injury reports; I’ll stream my starting D/ST once everything else is cleared up. Don’t let D/ST’s clog your roster, don’t grow attached to them, heck, even try to trade them for skill players if you can! You will be better off streaming the position and finding value elsewhere.
Did I miss something? Message me on Twitter.
@Jack: Pretty much everything I use is from nflfastR: https://www.nflfastr.com/
@Brian: Glad to hear! Yes, I’ll be taking a deeper dive into that with my matchups article this week…
@John: Great question…I did not change up the window size, which could certainly affect the results! Potential for a future study…
Where can I get some of these raw stats from? I would love to be able to use this for predictive analytics on my own end.
I basically have adopted this position, so it is good to hear some solid rationale behind the thinking. However, it would have been good to have some defenses to target rest of season for this year.
Good analysi here. I like how you even included the weights in your regression equation. Just curious — did you experiment with window sizes besides four weeks? Bottom line: your conclusions (streaming is generally the way to go, and the best streaming options are opponents of lame offenses) are useful guidelines. Thanks!