387: Team-Level Regression

00:00:02
Speaker 1: This is the Late Podcast with your host JJ zacher Retha JJ Zacherson. What's up everyone, It's JJ Zacharyes and the editor in chief at fandle and at Number five dot com in this episode three hundred and eighty seven of the Late Round Podcast all the many shows that are part of the FanDuel podcast network. Thanks for tuning in. So over the last few weeks, I've been doing shows on how to find breakout players during your fantasy football drafts. I did one on running backs, I did one on wide receivers, and I did one on tight ends. I was planning on doing a quarterback study, but I did all the research and I couldn't really bring anything together. There were some trends that I found, but nothing was so obvious and nothing was super super actionable. My advice remains the same when aiming for a late round quarterback, look only at upside, since we know that our worst case scenario in most leagues is streaming the position, and you can do that effectively. The episode two hundred and forty eight explains how so because of that, I'm shifting my focus. I do these episodes each year where I look at team level regression and then I dig into individual players on a separate show. It's a helpful exercise to walk through before drafts start, so you can get an idea of what's bound to change from one year to the next. I think team level regression is often overlooked in fantasy football circles too. We're always concerned with individual players and what's sustainable and what's not, But a lot of those numbers are the product of how a team operates overall are in general. So today I'm going to talk about some regression candidates at the team level, and of course I'm doing this with a fantasy angle. I'm basically asking what did these teams do in twenty nineteen that impacted fantasy football that's bound to regress and effect fantasy football in twenty twenty. Before digging in, though, I wanted to send another reminder about fanduels football multiverse. Each week, you can make FanDuel lineups based on Madden simulations, and those Madden simulations have pretty fun what if scenarios associated with them, like, for example, what if the Cowboys didn't give Ezekiel Elliott his big contract? What would that offense look like with Tony Pollard as their starting running back, or what if Sam Darnold was drafted by the Giants instead of Saquon Barkley. You can see how those scenarios play out this week on Fandal's YouTube page, where my colleague Jim sonis and I will be calling the games. The first game this week is Wednesday at eight pm Eastern, and then there's one on Thursday two. So check out fandel dot com slash NFL Multiverse, enter a free lineup, and have a shot to win a little bit of cash. It's all free, So head to fandel dot com slash NFL Multiverse, make a lineup, and then tune into fandl's YouTube page on Wednesday. Okay, so regression to the mean can happen in a positive in a negative way. The term positive regression is usually thrown around in the fantasy football world. Is a very quick and easy way to say a player or team is going to regress the mean in a positive way, their numbers are bound to get better, and then negative regression is the opposite. Now, keep in mind, positive regression is not the same thing as progression. Progression tells us that someone or something is actually getting better, like a player who progresses is one who becomes more talented and then accumulates better statistics as a result. Positive regression is just math that a team is gonna get better within a statistic, not necessarily because they get better, but because they basically become less unlucky. I look at a lot of different statistics the spot regression at the team level, but I think there are two main buckets that those statistics generally fall into, touchdown scoring and play calling. With touchdown scoring, I'm not just looking at raw numbers, but I'm looking at how many passing touchdowns teams had versus how many rushing touchdowns, And I'm looking at how well teams were able to convert plays and drives into scores. Generally speaking, teams who do things at a crazy high or low rate one year will look different than next year. Let's just focus on touchdowns to kick off the analysis on today's show. I talk about pass to rush touchdown ratio a lot on here, but it's just the ratio between passing touchdowns to rushing touchdowns that a team sees. So if a team has a ratio of one, that means that they're passing and rushing for the same amount of touchdowns. It's very rare to see teams with a ratio under one. By the way, I don't think that's surprising, especially in today's NFL, But since twenty eleven, only thirteen teams ran for more touchdowns than they threw four. That's just four and a half percent of all teams over the last nine years. The average pass to rush touchdown ratio over this time has been one point nine to four. Teams basically regress to That mean a team with a ratio well above that is likely to see it more run heavy the following year, and vice versa. A good example from twenty nineteen and our first regression candidate of the day is with Jacksonville. Last year, the Jags had a passing to rushing touchdown ratio of eight. They threw for twenty four passing touchdowns, and they had just three rushing touchdowns. This is a big reason why I kept saying last year Leonard Fournette is going to regress eventually, even though he didn't totally regress. That's gonna happen in a small sample size, but it would be shocking if the Jags continue this pace into twenty twenty. Since twenty eleven, only Seattle in twenty seventeen had that high of a ratio in a single season. The Seahawks saw that ratio go from eight point five to two point three the following year. And when you look at the ten teams who have had a ratio above five and a half over this time with next season data, each of them saw their ratio fall the next season, with an average fall of about three point seven. The thing is, Jacksonville's really not projected to be very good in twenty twenty offensively. When you have a smaller pie that you're working off of, a shift in ratio doesn't do crazy things, doesn't do dramatic things to projections. Just as an example, if Jacksonville scores twenty seven touchdowns, again, if they see a decrease of three point seven in ratio, which could easily be much larger, but let's say they see a decrease of three point seven, then they're still only going to rush for like five touchdowns this year had they gone for fifty offensive touchdowns, and if they were projected for fifty offensive touchdowns, they go from like forty five passing touchdowns in that scenario down to forty, which would raise their rushing touchdowns by five. That's much more dramatic. Basically, what I'm saying here is projections should show regression. Regression is baked into a logical good projection, and while someone like Leonard Fournette looks decent, if we were to assume a bellcow workload, it's still hard to project them for significantly more touchdowns because the offense he's in doesn't project to be very good. So, yes, regression is coming. If Fournette has a similar workload, he'll almost definitely score more than he did last year. The real problem knows that regression isn't created equally. Not all regression is the same. The Jags positively regressing in pass to rush touchdown ratio, they not do a whole lot to the team's bottom line because that bottom line is probably not gonna be very good now along those same lines as Carolina, or should I say, along the same lines as being someone of an outlier and passed to rush touchdown ratio, because Carolina had the opposite situation as Jacksonville. Last year, the Panthers had a passed a rush touchdown ratio that was just zero point eight. Five of the teams with a ratio under one since twenty eleven, all but one of them saw that ratio surpass one the following year. For a lot of reasons. The expectation should be that the Panthers throw for more touchdowns this year than they did last year, and the percentage of their touchdowns will skew more pass heavy than they did a season ago. Those were the two biggest extremes in passed to rush touchdown ratio this past season. But another way that you can look at this is the percentage of drives that end in touchdowns. Basically, if teams don't score any points on a large number of drives, but they still have a pretty high touchdown rate per drive, then we should expect them to not be able to maintain that week after week and year after year. The best example of this from last year is Tennessee. Last season, the Titans ranked twenty third in percentage of drives that ended up in a score, so either a touchdown or a field goal, but they ranked fifth in percentage of drives that ended in a touchdown. I've mentioned this a little bit before on the show, but Ryan Tannehill, he took over at quarterback in Week seven last year in the regular season. From that point through Week seventeen, the Titans were eighth and percentage of drives that ended up in scores, but they were second in percentage of drives that ended in touchdowns. The Titans scored thirty nine touchdowns to just four field goals over that time. Things aren't going to be so efficient and so perfect next year, you guys. I don't think it means that you have to fade Titans players, though not at all. Actually, regression doesn't mean something becomes irrelevant. I think it's important to spot regression and know where regression is coming, but you also have to know how the market is looking at that regression. Market for Ryan Tannehill, for example, in my opinion, is overstating the regression. So I can say the Titans are going to regress, that they won't be as good at scoring touchdowns next year, and I'd bet on that. But what matters is how the market is acting, and for Tennessee and the players in that offense, the market hasn't assumed that they would keep up that pace at all. Now. I've talked about how teams score touchdowns by looking more at ratios so far, but generally speaking, the cumulative number of touchdowns that a team scores on offense also regresses. In order for a team to do really well and score fifty or sixty offensive touchdowns in a season, things had to go right. Whether it's a lucky bounce here or there, or just injury luck. Things went right, things went well for a team to score like fifty five touchdowns in a season. That's the case for Baltimore. Last year. The Ravens led the NFL with fifty eight offensive touchdowns. Of the seven teams to hit that marksans twenty eleven, none of them increase their total. The fallow up, and the average touchdowns scored among those teams the next season was forty seven. Even if you look at the twenty five different teams with fifty or more touchdowns since twenty eleven, twenty three of the twenty five saw their touchdowns drop the following year. We can't assume Baltimore is even close to us good this year at scoring touchdowns as they were in twenty nineteen, and then naturally, the opposite side of this is true too. The two worst teams at scoring touchdowns last year were the Steelers and the Jets. Both ended up with twenty five. We know Pittsburgh is going to score more this year with Ben Roethlisberger back, but the Jets should too, potentially a lot more. Only nineteen other teams have scored twenty five or fewer touchdowns since twenty eleven. Only two of those teams saw a decrease in touchdowns the following year, And given what we've seen from teams who score in the twenty five range, like we saw with the Jets, we should expect them to increase their total based on what we've seen historically, by almost ten. This is a really big reason why we can't just use last year's results and plug them into team level projections for this season. I don't have the Jets scoring thirty five touchdowns in my projections, but I do have them for thirty two. That's a good bit higher than where they were at last year, which is partially because Sam Darnold is healthy, but it's also because of natural regression. So, Okay, we've looked at touchdowns and some really easy ways to spot touchdown regression. At the team level, we can and should do the same exact things for play calling. Two, Because think about it, if you're building projections or thinking about what goes into a player's stat line, it's really just how many times will this offense score to go along with how much volume is going to be there for this player. Of course, efficiency in such matters too, but at a basic level, we're looking at volume and touchdowns. On the volume side, I tend to look at pass to rush attempt ratio and the number of plays run by a team. Let's start quickly by looking at the latter, which is plays run. The average number of plays run by NFL team since twenty eleven has been oneenty and twenty two, so you'd probably expect any team way under that mark to increase the following year, and any team way above it to decrease, and that's generally what we see. Again since twenty eleven, the bottom twenty teams in plays run saw their plays increase the following year on average by fifty nine. A couple of teams saw a decrease, but most had an increase. And then among the top twenty teams, so basically the top two per season, we had a decrease of about fifty four plays across the season. So what does that mean for twenty twenty? What's that mean? For this year will Teams that ran very few plays last year included Washington, Pittsburgh, Tennessee, Denver, and the Jets. Washington is the one team who really really stands out on that list because they only ran eight hundred and eighty five offensive plays. That's the second fewest in our two hundred and eighty eight team sample. Expect more volume in general for Washington, and I think that you could say the exact same about Pittsburgh. But Washington is the one team who really stands out in the plays run department, whereas no one released stands out dramatically on the other end. Now, let's talk about pass to rush attempt ratio real quick, since it plays in the whole volume thing. Overall, pass to rush ratio regresses. I say overall because you can look at past to rush ratio in different game situations. Game script dictates how pass heavy a team is going to be. If a team is really, really good, they may not have a super high pass to rush ratio because when they're leading, they're running the ball more to kill clock. When a team sucks, they're playing catchup, so their ratio is higher. They're throwing more. This is why neutral script ratio, or how pass heavy a team is when games are within six points. That's why neutral script ratio is more predictive for next season pass to rush ratio than overall ratio is. Let me explain that another way. I'm talking about three separate numbers here. One is the overall pass to rush ratio a team has in a single season. So last year, for instance, the Falcons had a pass to rush ratio of one point eight nine every one point eight nine pass attempts they ran the football in a neutral game script for Atlanta last year, that ratio was one point six y nine nice, So when games were closer, the ratio was lower, which makes sense because they weren't very good last year. Now, the third number that I'm talking about is the following year's overall ratio, and based on history, that number is going to be closer to the previous season neutral script ratio than the previous season's overall number. So the Falcons are actually a good example to talk through. They ranked first in overall ratio last year, but they were fourth in neutral script ratio. We should expect them to be a pass heavy team this year because that's what they do, and it's not like they're projected to be a spectacularly good team, but we can't assume that they'll be as pass heavy as they were in twenty nineteen. Now on the flip side are three teams who were bound to regress positively to the mean. One of them is Baltimore. The Ravens ran just twenty one point six percent of their offensive plays last year while trailing. That's easily the low lowest rate in football. They ended up with the lowest pass to rush attempt ratio that we've seen across our nine year sample, and they're gonna have a hard time repeating it. Among the bottom ten teams in pass to rush ratio since twenty eleven with next season data, the average increase was point two to two, So Baltimore in that scenario would go from a point seven to four ratio to a point nine to six to one. They'd still be run heavy, but it does change the way that we'd project them a little bit. My projections don't have an increase by that big of a number, but they should be more pass heavy. Another one on the flip side is San Francisco. They, like Baltimore, had a lot of positive game scripts last year, and as a result, they had more rush attempts than pass attempts. Only one of the twenty seven teams since twenty eleven with more rush attempts than pass attempts saw their ratio decrease the following year, So again, it's just hard to maintain being that good. They're bound to be more pass heavy this year. And then the last team is kind of interesting. It's Minnesota. They have some changing pieces with Kevin Stefanski and Stefan Diggs now gone, but the Vikings had a ratio that was under one last year as well. They ran more than they threw it. But what's more interesting about them than someone like San Francisco is that Minnesota ranked twenty third in neutral script pass to rush ratio last year. San Francisco is still twenty ninth. So Minnesota is low key in the spot too, where it would be surprising if they were able to maintain such a low rate of passing. Now, I know that I just went over a lot of random examples today, but as usual with this show, I'm just trying to get your mind to think about fantasy football and ways that it might not have before. Everything starts at the team level, how a player performs depends on factors that aren't always in that player's control. And if you can understand how team regression just generally works at a super high level, you're gonna have an advantage over your league mates. I guarantee it. But that's gonna do it for today's show. Thanks to all of you for listening. If you get subscribed to the Late Round podcast, make sure you are by searching for it pretty much any where podcasts can be found, and don't forget to follow me on Twitter at Late Round QB. And if you could do me a favor, leave a rating and review on Apple podcast. It goes a long way to helping the show. Thanks everyone. I'll talk to you later in the week with the Weekly Mailbag episode.