342: Mailbag 1/24/20

00:00:02
Speaker 1: This is the Label podcast with your host JJ zacher Resa, JJ zacher Reason. What's up everyone. It's JJ Zacharyson, the editor in chief at Fandle and at number fire dot com. And this is episode three hundred and forty two of the Late Round podcast with many shows that are part of the FanDuel podcast network. Thanks for tuning in. I'm tired of the football analytics debates on Twitter. They're exhausting. A few weeks ago, I made a thread of tweets after Dave Gettlman said quote, people say it's a passing league. I get that, but the top four teams were not in the playoffs. The top four rushing teams were in the playoffs. The fact that an NFL GM is using this type of logic is ridiculous. That's why I made the thread. I wanted to explain to people that it's not about cumulative numbers. It's about efficiency because, as we know, game flow will dictate how pass or run heavy a team actually is. If a team is leading, they tend to run more. If they're losing, they'll pass. That's why since twenty eleven, and this is what I talked about in the thread of tweets, but since twenty eleven, seventy six percent of teams with nine or more wins had a lower overall pass to rush ratio than they did in neutral game scripts. To put this another way, winning teams were more run heavy in total than they were when games were close, and as you increase that win total, the percentage of teams being more run heavy win not in neutral game scripts increased as well. At eleven wins, eighty five percent of teams were more run heavy in total ratio than in neutral scripts. At twelve wins, ninety three percent at thirteen wins, Every team since twenty eleven with thirteen or more wins had a higher overall pass to rush ratio than the ratio that they had in neutral game scripts. The idea for that thread wasn't to say that running the ball is irrelevant. Instead, it was to say how absurd it is that a freaking NFLGM is using this type of logic to make a point, and in his case, to potentially build a team around it. And now that we've seen the forty nine Ers and Titans and Raves and others run effectively and make it past the wild card round of the playoffs. Twitter battles around analytics have been crazy over the last month because, as a lot of you know, those who study analytics always talk about the importance of passing efficiency and pass defense efficiency. But people who may not be as open about the role that numbers play in football, they'll fall back to the idea that you have to be able to run the football and stop the run, especially in the playoffs. That's how you win championships. So let's look at the forty nine ers. They obviously play good defense and they can stop both the pass and the run. But look at what they've done offensively in their two playoff games, two decisive victories. While Jimmy Garoppolo has twenty seven pass attempts, the forty nine ers non quarterback rushes have equal to eighty one. Now, aside from the fact that playing with a positive game script can increase the number of rushes that a team runs, the forty nine Ers have been effective running the ball, and that's what's caused nerds to fight on social media. One side says you need to throw the ball effectively to win, while the others said, see, I told you that you need to run the football. As someone who analyzes the game with a pretty numbers heavy approach. I really think a big part of the problem is communication. It's the communication of these ideas. Analytics don't say that you should never run the ball. In fact, there are a lot of good analysts who provided a lot of instances where running the ball is preferred. What's being missed, I think is the aspect of sustainability. It objectively is smarter to build your offense to run an efficient passing game because of what we know wins in the league. Passing efficiency correlates to wins at a much much higher rate than rushing efficiency, and the same goes with pass defense efficiency and rush defense efficiency. And again I'm using efficiency because of game flow. Game Flow is a big reason that cumulative data is less relevant than efficiency data. What I think also should be recognized is that a single elimination playoff has a lot of variance. Why do you think the NCAA tournament is such a spectacle. It's because you're getting all these different schools in one bracket and literally anything can happen because its a single elimination tournament. It's called march madness for a reason. March variants just doesn't sound as good, but Virginia losing to a sixteen seeded UMBC doesn't mean Virginia is worse than UNBC. If they had played one hundred times in twenty eighteen, Virginia might have won ninety nine of those games, but that doesn't matter in a one and done playoff. So what that means is that crazy stuff can happen in the NFL playoffs. The objectively best teams aren't always going to win, and that also means that variants can hit in a big way. In a small sample size, teams can win in a lot of different ways, like in a three or four or even five game stretch. A team can convert on third down at a really high and unsustainable rate. Teams can run the football more efficiently than most teams pass the football. On a small sample size, it can happen, and that's what happened with the Tennessee Titans offense. Before that Kansas City game last weekend with Ryan Tannehill under center. This year, the Titans had scored thirty one red zone touchdowns to just one red zone field goal, meaning when the Titans were scoring in the red zone, they were essentially scoring a touchdown That gave them a massive, massive advantage. But that's not sustainable. That's not how basic logic works. You can't expect the Titans to continue to score in that way when it's like twenty five times as efficient as an average NFL offense, just like you can't expect the forty nine ers to just continue to run at a six and a half yards per carry clip week over week. The difference in these debates is looking at what's probable versus what's possible. It's possible that the forty nine ers do what they did against Green Bay here in the Super Bowl. That's in their range of outcomes, but that shouldn't be the expectation. That's not what's probable, and it's also not what's sustainable. If you look at year over year data and see what's sticky or what tends to remain the same, it's passing efficiency. Rushing efficiency has very little year over year correlation, as does most defensive performance. So to put that another way, the way a rushing offense performs in one year typically has no predictive value in how they're going to run the next year. There are obviously going to be some exceptions here, but this intuitively makes a lot of sense. Quarterbacks drive the game, so when the same good quarterback is under center for a team, that team will see similar results. Now, I don't want to tok your ear off here, because I can go on and on about this stuff after having read a ton of great research by people who are smarter than I am. There are a ton of things that analytics have shown that have a chance to alter the way that the game of football is played, but at the very highest of levels. As a communicator, first and foremost, I think it's really important for both sides of these arguments to understand that the core fundamental difference seems to be about what's sustainable and what's not sustainable. In other words, there probably needs to be more discussion around variants that For the very pro analytics side, it's okay when Derek Henry and the Titans offense do what they did for a stretch of time. And for the people who are more hesitant about analytics, it's good to not use these small sample sizes to overshadow the bigger picture. Now, maybe this is a stupid thing for me to bring up. Maybe I shouldn't care about the back and forth beefing that's been on every Football Fans timeline over the last two and a half weeks. I just think it's important to move the conversation forward, and the best way to do that is to be better communicators. Now, I know that that was a long ramp, but today is a mailbag day, so I've got some questions to answer. But before doing that, I want to let you guys know that with some of the NBA's biggest stars headed to Chicago for the twenty twenty NBA All Star Game, Fandel is giving you a chance to join them. Just enter any NBA Road to All Star free play until February second, and compete for your chance to see the game in person. Your top three scores will count toward your spot and the leaderboard, with the top five users on the leaderboard after February second, winning a trip to Chicago. For more details was at www dot Fandel dot com, slash NBA, or download the Fandle app today. Eligibility restrictions apply. This first question says there's been such pushback against analytics and football. Despite the use of data conveniently proving itself useful, fantasy has seemingly embraced the use of data in forecasting much better than real football. What's it going to take to show people that data is here to help well? First off, I think we probably take the analytics versus non analytics thing to the extreme. I would probably bet that the majority of people are totally cool with the use of analytics. It's the question of to what degree. And I do think how we talk about this stuff is important. It's important to have inviting discussion and to not be condescending. That's hard to do sometimes when you're putting in a lot of work and then someone tries to discredit it through anecdotal evidence. But I do think that it's important. It's something that I definitely want to work on. But the reason it's embraced in fantasy is obvious. It's clear it's a numbers driven game. If you're not analyzing with numbers, you're losing a very, very obvious edge. This has questions from Twitter from at Panazelia. You said that your rookie model loved Nikhil Harry out of the draft. What did your model love about him? Do you think that your model failed with him or is it too early to tell? Is there anything about your model that you plan to adjust for this year regarding the outcome with Harry. And then I also got an email from Brian that was along the same lines that said, good afternoon, JJ. Love your pod. Thanks, listen to a multiple times a week on my lunch break. I'd like to ask about Nikhil Harry. Do you have an example of a player who started his career like him and went on to be very successful, particularly a player that was picked in the first round, was injured a bit, but also failed to impress as a rookie. Okay, so to the first question about my prospect model, someone actually hit me up on Twitter to talk about this as well, in a pretty trolly way, I should add, but they hit me up on Twitter because of what I talked through with Dante Pettis versus Deebo Samuel. I listed off the receivers who ranked highly in my prospect model over the last two years, and there were some hits, but there were some misses. Some of you may have thought that there were a lot of misses, but I want to remind people about a couple of things here. Number One, I don't think people realize how difficult prospecting actually is. Like you thought the average draft position in regular redraft leagues was bad at predicting things. If you're in a dynasty league, you know that rookie drafts are way, way worse. Number two. When it comes to a prospect model like this, it's really important to not let one or two or even a handful of results approach. The idea here is to connect to players measurables, their body and their college production, but mostly their college production. But you connect those things and compare them to players who have been successful at the NFL level. You're finding shared attributes in order to increase your probability of finding the right player. My process didn't have the best results in twenty nineteen. I'm not gonna lie to you, guys. You can listen to my podcast last year, and I got the wide receiver group pretty wrong. I was fine at running back because they're generally a little bit easier to prospect, but wide receiver was tough. Changing everything up though, because of those results wouldn't make any sense. I can continue to try to work on the model in order to make that R squared value stronger when comparing prospects to successful fantasy wideouts at the NFL level. But I'm not gonna scrap it because I like jj R, Sega, Whiteside, and Nakill Harry. Now, regarding the other parts to this question, I do think it's too early to tell, or at least definitively tell, if Nikil Harry's a bust. It's certainly not a good thing that Nikil Harry had very little production during his rookie year. Though when you look at the best wide receivers in the league, nearly all of them were productive in some way as rookies. And if you look at rookie drafts since twenty eleven, among first round wide receivers who didn't hit one hundred Fantasy points, you're looking at basically two or three guys out of eleven with multiple seasons worth of data who hit those being Mike Williams, who's not even an elite asset, DeVante Parker who took forever to break out, in Alshon Jeffrey Otherwise, you've got Corey Davis, John Ross, Laquon Treadwell, Josh Doson, Corey Coleman, Breshad Perriman, Nelson Agalor, Kevin White, and John Baldwin. It's not great, and the opposite list of players who hit one hundred points during their rookie season. It basically includes every elite wide receiver in football right now. There were some misses, but there were a lot of hits. Now. I still have at least some faith in the kill Harry because the answer part of this question. He had an unbelievable college production profile. He had an elite breakout age, he had incredible market share numbers. He had exactly what you look for. If it doesn't pan out, that's gonna happen. There have been some busts with him profile over the years, but this is about probability. You're never going to get everything one hundred percent right, and even actual NFL teams don't either. This next question is an email. It says, hey, JJ, I was wondering is there a way to spot individual players that might still have good seasons on teams due for regression like the Ravens, Niners, and Titans next year? Thanks Marshall. So I'm assuming this is in response to a chart that I tweeted this week showing team offenses that really outperformed in the touchdown column based on their yardage totals this year. Because those three teams were the teams that I highlighted, and yes, all three of those offenses are due for some regression next year. But I think the important thing with this is to remember that regression doesn't mean a relevancy. If you're building projections, for example, or if you're getting projections from somewhere, those projections are likely using a top down approach where they build out the team level first and then the player level second. So if the team level shows regression, it'll automatically be sort of embedded into that player's projection. So think of it this way. Spotting those players is going to be no different than how you'd spot any other player. Their situations aren't as good as they were the previous year, but that doesn't mean that they won't be good for fantasy purposes. And honestly, guys, sometimes the industry takes regression too far. Sometimes you can get discounts on players because of the market. Is viewing regression a little bit too harshly. I know it's not a super direct answer, but I think it covers what you're asking for here. This nice question is an email says JJ, thanks for another season of great hashtag content. You're welcome looking forward to all of the off season information in pods to gain an edge heading into next year. With all the young, dynamic quarterbacks currently in the NFL, the increasing popularity of selecting these quarterbacks earlier in drafts, and the overall importance of selecting a quarterback with the Konami code, do you recommend drafting these types of quarterbacks earlier in the draft in one quarterback leaguess, trying to pin which non Konami quarterback will have a breakout season at the end of the draft seems not so ideal. Thanks again, Mike, So, I think there's a question that's still pretty hard to answer without any sort of average draft position data, because we're not really seeing an increase in popularity of selecting young quarterbacks earlier in fantasy drafts. At least that's not how things have trended. If anything, we keep seeing quarterbacks going later, at least that's what average draft position is telling us. Now, I understand Lamar Jackson is going to go early, but you're still gonna probably be able to get Kyler Murray in the middle to late rounds, And the same could be said for like Josh Allen. And then there's Russell Wilson, there's Dak Prescott. There are more than just two or three guys who are giving you that cheat code effect. But my answer to this question is probably obvious. I wouldn't be approaching the position any different in twenty twenty unless there's a clear value play. We know what the data says, and it's also not absolutely necessary to get a rushing quarterback. It's important and you should aim to do that, but players can have outlier seasons and touchdown rate as well, and you're really just hoping for that in a single quarterback league because that's the main way to get an edge to draft an outlier. There are just so many usable quarterbacks that it's become complete unnecessary to reach the position in a single quarterback league. Arguments can be made for Lamar Jackson if you think that he's gonna repeat what he did in twenty nineteen and depending on Mahomes's landing spot ADP wise. I get the allure there too, but you should be drafting upside late. We have mountains of data and evidence showing that that's the smartest thing to do. This last question is an email from Shay, and it's another one about one quarterback leagues in fantasy, it says Devil's Advocate question on quarterback draft value, Mahomes wasn't worth a third round pick, but if you drafted him instead of Adam Thielen or Damian Williams or Antonio Brown, you were probably happy. Right, Sam goes for Travis Kelce over those guys. Is there anything to the idea that an elite onesie player offers less risk than a wide receiver or running back at a similar ADP, and therefore the opportunity cost won't be as bad? Obviously hard to factor in injury for Thilan and Williams and Brown had their own risk that made them potential avoids. But still curious on your thoughts. Thanks, love the pod, and don't worry, I'll always wait on quarterback. Yeah. So I talked about this in detail on that Superflex episode that I referenced earlier in the week. It's episode one hundred and thirty nine of the Late Round podcast. But to reiterate a little bit of what I said earlier this week, each draft pick that you make is not about safety. It's about value. It's the same deal in the real NFL draft. Saquon Barkley was a safe pick. We knew Barkley would be a good running back, but that doesn't mean he was a smart one. Given the impact that running backs have in the NFL, he wasn't a valuable pick. The same goes for fantasy Patrick Mahomes was safe, but he wasn't valuable because of the replaceability of the position and the fact that you can get usable guys late in your draft and you can test this. I talked about this on episode one hundred and thirty nine. When you look at preseason average draft position to postseason result through the years, quarterback ADP is no more predictive than running back or wide receiver ADP. Seriously, it seems like they would be, but they're not. Replaceability is high. When there are so many usable quarterbacks in fantasy football, and so many quarterbacks who are capable of posting good numbers, that means the competition is tougher. It's harder for a quarterback to finish in the top twelve or the top six at the position, and to this point, you're only starting one of those players in your fantasy league. What that means is that with one small slip up or one stretch of bad games, all of a sudden because of the competition, that quarterback is no longer giving you any significant edge. It's very hard to be a good fantasy quarterback these days. You can sit here and pick off individual players and say, would you have rather had Patrick Mahomes than this guy? Yeah, I'd rather have Patrick Mahomes than Antonio Brown. Antonio Brown wasn't on an NFL field for ninety five percent of the season. Of course I would rather have Patrick Mahomes. I'd also rather have Leonard Fournette or Aaron Jones or Derek Henry or Amari Cooper, Kenny Galladay or Chris Godwin. We can play this game all day. It does us no good to do analysis at strictly hindsight and anecdotal. When we're predicting and forecasting, we're looking at expected value, and the expected value for early round quarterbacks is not great. That's gonna do it for today's show. Thanks to all of you for listening. If you've got to subscribe to the Late Round podcast, make sure you are by searching for it pretty much anywhere podcasts can be found, and don't forget to follow me on Twitter at Late Round QB. Thanks everyone, enjoy your weekend,