Why Chris Boswell Is Ranked So Low... (1 Viewer)

[rockaction]

@Adam Harstad You are the man! Thanks for posting here. You wrote the single best ff article ever last year. Any time I bring it up around here 80-90% of the posters nail me to the cross. Your reference to predictions accuracy even more cements that article.

Bass, if you happen to know offhandedly, which article are you referencing? I try to keep up with Adam’s articles, but I haven’t read every one of them.

 

[Joe Bryant]

@Adam Harstad You are the man! Thanks for posting here. You wrote the single best ff article ever last year. Any time I bring it up around here 80-90% of the posters nail me to the cross. Your reference to predictions accuracy even more cements that article.

Bass, if you happen to know offhandedly, which article are you referencing? I try to keep up with Adam’s articles, but I haven’t read every one of them.

Hi @rockaction . Here are Adam's articles this year https://www.footballguys.com/articles?authorId=10768

 

[rockaction]

@Adam Harstad You are the man! Thanks for posting here. You wrote the single best ff article ever last year. Any time I bring it up around here 80-90% of the posters nail me to the cross. Your reference to predictions accuracy even more cements that article.

Bass, if you happen to know offhandedly, which article are you referencing? I try to keep up with Adam’s articles, but I haven’t read every one of them.

Hi @rockaction . Here are Adam's articles this year https://www.footballguys.com/articles?authorId=10768

Thanks, Joe. I really appreciate that. Bass contacted me with the article. It’s from last year, and it’s one that left an impression on me. It’s a good article. It’s something he piggybacked off of Chase Stuart and your VBD system back from when Chase wrote an FBG article in 2013.

Thanks, @BassNBrew , for getting back to me.

 

[OldMilwaukee]

@Adam Harstad You are the man! Thanks for posting here. You wrote the single best ff article ever last year. Any time I bring it up around here 80-90% of the posters nail me to the cross. Your reference to predictions accuracy even more cements that article.

Bass, if you happen to know offhandedly, which article are you referencing? I try to keep up with Adam’s articles, but I haven’t read every one of them.

The Best Draft Strategy for 2023 (And Every Other Year)

 

[Adam Harstad]

Yeah, unfortunately, I'm in the opposite category. My "gut call" was to draft Boswell. It was great until I noticed that he was a consistent "avoid" and Jake Elliott, who was a consistent "great" or "good" play was available.

I know it probably doesn't help, but that really sucks and I'm sorry. I'm always pretty keenly aware that perfection is impossible, that being a "good analyst" in this space means that maybe 56% of people who follow your advice wind up with more wins and 44% wind up with fewer. Intellectually, I can take comfort in the belief that I'm helping more than I'm hurting on net. But the 44% who wind up worse off certainly don't take any comfort in that, and nor should they. It's my job to care about my readers' successes. It's not their job to care about each others' successes, and any such appeals should ring hollow. The only teams you should care about are your own.

Anyone who reads anything I write is placing their trust in me. Anyone who actually HEEDS what I write is placing an even bigger trust. If you don't take that seriously, I wonder what you're doing in this industry-- it's certainly not trying to help people. If that 44% doesn't eat at you, what's going to motivate you to bust your *** to drop it down to 42%? To 40%?

I'm sure it's too little, too late, but I do genuinely believe that Boswell's performance to this point was a fluke and he won't sustain it down the stretch. Fluky points count the same as any others, so that's no help for now. But-- providence permitting and the creek don't rise-- hopefully this will be a recommendation that stings a bit less a few weeks from now.

honest question....is the bolded basically just based on models and stats saying it won't sustain or do you have other opinions on why it won't sustain that are based on something else....like you think Tomlin will get more aggressive.....game scripts will all of a sudden change for PIT....?....Boswell with start to succumb to the "elements" or things of that nature or are you just gonna kind of die on the hill of there has to be a regression to the mean based on models....

I mean, I write a weekly column titled "Regression Alert" where all I do is look at the highest and lowest performers in a metric that is historically fairly unstable and predict that both groups will regress strongly to league average and, as a result, the low performers will outperform the high performers going forward. Here's this week's edition (which tracks every prediction and links back to every other article from this season, and to the year-end roundups from all previous years, and from there to every article I've written and prediction I've made since 2017).

A big part of this is that the names involved are irrelevant. If you're at the top of a stat that's not sustainable, I predict you'll not sustain it, even if you're Justin Jefferson or Derrick Henry. If you're at the bottom of a stat that's not sustainable, I predict that you won't sustain it, even if you're Bryce Young or Mecole Hardman. And despite not cherry-picking names, Regression Alert's predictions have about an 80% lifetime success rate. It would be higher, but I try to avoid any easy wins. I'll often stack the deck in favor of the "high outliers" group and we'll still see flips where the high outliers go from outperforming the low outliers by 30% to getting outperformed by 30%.

I get into a lot of the theory and math behind this, but the upshot is that mathematically speaking, in any statistic with a measure of randomness, the highest and lowest ends of the spectrum are almost certainly disproportionately impacted by that randomness (in a way that will not be expected to carry over going forward). The more randomness a statistic is subjected to, the more strongly it will impact the past samples and the more noticeably it will disappear from the future. And as I showed above, kicker is a position that is heavily impacted by randomness (otherwise it would be much more predictable).

Beyond that, I always try to keep a keen eye on history. Football performance is quite variable, but that variability is largely bounded-- it is extraordinarily unlikely that a player will perform outside the bounds of what we've already seen. Not impossible, but if you bet against it, you'll be right way more often than wrong. It's very plausible that Pittsburgh this year is an outlier (the fact that they've "overperformed" by so much to this point makes it more likely, at least on a season-long basis-- though that's of no use for going-forward projections). It's much less plausible that they're a bigger outlier than the 2000 Ravens and 2002 Buccaneers. I.E. my biggest reason for believing this is unsustainable is that we've never seen anyone sustain it.

TL;DR -- Yes, I'm willing to die on the hill that there will be regression to the mean (not based on the models, but based on mathematical principles on how randomness operates). I have written a weekly column since 2017 specifically dedicated to dying on the hill of blindly predicted regression to the mean.

 

[Stinkin Ref]

Yeah, unfortunately, I'm in the opposite category. My "gut call" was to draft Boswell. It was great until I noticed that he was a consistent "avoid" and Jake Elliott, who was a consistent "great" or "good" play was available.

I know it probably doesn't help, but that really sucks and I'm sorry. I'm always pretty keenly aware that perfection is impossible, that being a "good analyst" in this space means that maybe 56% of people who follow your advice wind up with more wins and 44% wind up with fewer. Intellectually, I can take comfort in the belief that I'm helping more than I'm hurting on net. But the 44% who wind up worse off certainly don't take any comfort in that, and nor should they. It's my job to care about my readers' successes. It's not their job to care about each others' successes, and any such appeals should ring hollow. The only teams you should care about are your own.

Anyone who reads anything I write is placing their trust in me. Anyone who actually HEEDS what I write is placing an even bigger trust. If you don't take that seriously, I wonder what you're doing in this industry-- it's certainly not trying to help people. If that 44% doesn't eat at you, what's going to motivate you to bust your *** to drop it down to 42%? To 40%?

I'm sure it's too little, too late, but I do genuinely believe that Boswell's performance to this point was a fluke and he won't sustain it down the stretch. Fluky points count the same as any others, so that's no help for now. But-- providence permitting and the creek don't rise-- hopefully this will be a recommendation that stings a bit less a few weeks from now.

honest question....is the bolded basically just based on models and stats saying it won't sustain or do you have other opinions on why it won't sustain that are based on something else....like you think Tomlin will get more aggressive.....game scripts will all of a sudden change for PIT....?....Boswell with start to succumb to the "elements" or things of that nature or are you just gonna kind of die on the hill of there has to be a regression to the mean based on models....

I mean, I write a weekly column titled "Regression Alert" where all I do is look at the highest and lowest performers in a metric that is historically fairly unstable and predict that both groups will regress strongly to league average and, as a result, the low performers will outperform the high performers going forward. Here's this week's edition (which tracks every prediction and links back to every other article from this season, and to the year-end roundups from all previous years, and from there to every article I've written and prediction I've made since 2017).

A big part of this is that the names involved are irrelevant. If you're at the top of a stat that's not sustainable, I predict you'll not sustain it, even if you're Justin Jefferson or Derrick Henry. If you're at the bottom of a stat that's not sustainable, I predict that you won't sustain it, even if you're Bryce Young or Mecole Hardman. And despite not cherry-picking names, Regression Alert's predictions have about an 80% lifetime success rate. It would be higher, but I try to avoid any easy wins. I'll often stack the deck in favor of the "high outliers" group and we'll still see flips where the high outliers go from outperforming the low outliers by 30% to getting outperformed by 30%.

I get into a lot of the theory and math behind this, but the upshot is that mathematically speaking, in any statistic with a measure of randomness, the highest and lowest ends of the spectrum are almost certainly disproportionately impacted by that randomness (in a way that will not be expected to carry over going forward). The more randomness a statistic is subjected to, the more strongly it will impact the past samples and the more noticeably it will disappear from the future. And as I showed above, kicker is a position that is heavily impacted by randomness (otherwise it would be much more predictable).

Beyond that, I always try to keep a keen eye on history. Football performance is quite variable, but that variability is largely bounded-- it is extraordinarily unlikely that a player will perform outside the bounds of what we've already seen. Not impossible, but if you bet against it, you'll be right way more often than wrong. It's very plausible that Pittsburgh this year is an outlier (the fact that they've "overperformed" by so much to this point makes it more likely, at least on a season-long basis-- though that's of no use for going-forward projections). It's much less plausible that they're a bigger outlier than the 2000 Ravens and 2002 Buccaneers. I.E. my biggest reason for believing this is unsustainable is that we've never seen anyone sustain it.

TL;DR -- Yes, I'm willing to die on the hill that there will be regression to the mean (not based on the models, but based on mathematical principles on how randomness operates). I have written a weekly column since 2017 specifically dedicated to dying on the hill of blindly predicted regression to the mean.

honestly you could have just said "no" and then "yes"...lol....thanks for all you do...

 

[Runkle]

Just wanted to add I'm a big fan of @Adam Harstad 's work. Consistently some of the best info available. And it doesn't matter to me if he's off on a particular kicker... it's not the kicker himself but the situation and opportunity I'm interested in week-to-week. Yeah, some guy might have a run of fluke games, or, even might be personally better than the situation suggests or might be outscoring his sample size, but, that's not what I am in the market for. Instead I want to hear about who, overall looking at the whole league, has a shot to have a lot of scoring chances on a particular Sunday. So for that, @Adam Harstad is the guy who I want to hear from.

Overall, to me, it's like poker. A lot of people can give poker advice. And when you're drawing to a flush for a made hand... someone like @Adam Harstad says "the implied pot odds on a call here are +EV given the stack sizes and your bankroll"... that's useful, actionable information that good independent of the result. A lot of other guys? Their advice boils down to "Hearts are hot tonight! The last four hands drawing to a flush in hearts hit, while the ones drawing to clubs are cold! So if you're drawing to a heart flush push those chips in! (and if it's clubs, fold!)" That might be correct information looking backwards, it may be accurate and true, but it's also useless and misses the point. The former info is real while the latter is a mirage. Gimme something based on well thought out numbers in a useful model any day.

 

[BassNBrew]

@Adam Harstad - Research request that I'm not even sure is possible.

My theory on kickers in H2H leagues is that I just want to acquire the kicker from the same team of my quarterback. I draft better than average following the principles of your article last year. I use the information of the the experts here to make draft, lineup, and waiver decisions. Let's say that over time that makes me a favorite against 66% of the teams I play and that my leagues on average award playoff spots to 33% of the league. Based on this, most weeks my RB/WR/TE position are going to give me a nice advantage. The only thing that stands in my way is my QB laying an egg either by his RBs getting most of the TDs that week or drives stalling and the team kicking FGs. My premise is that on team with decent offense, if I own the Qb and kicker, I'm generally getting a bit of the apple every scoring drive with either a FG or XP. I'm now trying for max points every week like a national contest or total points leagues, I'm trying to give myself the best chance of scoring more points than my opponent most of the time.

Side note that I think many here will get a kick out of. Your reference to Chase Stuart made me wonder what he's up to and in the course of digging I came across this....

I began playing fantasy sports in the late ’90s, which was a year-round hobby as fantasy basketball and fantasy football rose in popularity (I started off with fantasy baseball). I was quickly hooked on fantasy football, but it took a couple of years before Footballguys came across my radar. The articles were terrific and opened my eyes to the intricacies and strategies of the game. But the real treasure was the site’s message board. I could post on the board and minutes later someone would reply. That was my first introduction to the value of reader feedback. I didn’t think of “posting” on the message board as writing, but it was there that I learned the appropriate ways to craft an argument. The board also helped me develop a pretty thick skin for internet criticism, the sort of armor every blogger needs.

In the summer of 2002, Dodds announced that he was requesting freelance articles from fans of the site. At the time, Pro-Football-Reference was in its infancy, which still made it the best resource for the casual fan to access statistics on the NFL. I spent a few hours copying and pasting some of PFR’s data and used that to create my freelance article for Dodds.

Every bit of success I’ve had can be traced back to that response by Dodds, and the decision by Dodds and Joe Bryant (the other co-owner at Footballguys) to keep me on staff every year since (this was written in 2013). Once I became a staffer, I was able to graduate from incompetent writer to novice, but more importantly, I made some fantastic connections. I was able to convince Doug Drinen, then and now one of the most important people at Footballguys, to mentor me. That’s probably what happens when you e-mail a guy 1,000 times.

Yeah, one of my favorite uses of the kicker and defense spot in fantasy is tweaking correlations. Start a defense against the opponent's QB when you're a big underdog, say. Or pairing QB/PK (although I think the correlations there are actually positive-- a big day from the QB is more likely to lead to a big day from the kicker than a small one). I know the League Dominator factors all of these cross-positional correlations into its recommendations already, which is super cool. (I also think if people are worried about consistency-- and they shouldn't be, but I know some are-- the best way to achieve it is by looking at cross-positional correlations rather than seeking out "consistent" or "inconsistent" players.)

As for Chase, he has a kid now! I mean, it's less surprising to me since I'm essentially the same age and I've had kids for ages. But I know many who still think of him as 18 in perpetuity.

Thanks for the reply and the link.

Days like today are why I like the QB/PK tandem.

Goff 9 pts
Bates 10 pts.

 

[Mister CIA]

When our league used to require kickers, my first selection filter was to target kickers on what i preceived to be "high-end, mediocre offenses," and then field conditions (dome, then weather). Don't know if it was a good idea, but it made sense to me and I did not question it.

 

[Adam Harstad]

I was getting a head start on the season-ending evaluation for Rent-a-Kicker's performance, so I took a look at Boswell.

My original position was that Boswell wouldn't maintain his FG accuracy (100% at the time), nor would he maintain his share of Pittsburgh's offense (45.9% of all points scored). League average on both values was 85% and 32.7% (and even among the most defense-first teams of the last 20 years the kicker only scores around 35.3%) and my model expects all kickers to perform right around both marks. I also said-- based on the fundamentals-- Boswell probably "should have" been expected to rank around 7th in scoring to that point in the year.

Since the time this thread was started, Boswell ranks 9th in total points (behind Aubrey, Ryland, McManus, Sanders, McLaughlin, Elliott, Lutz, and Bates), falls to 10th if you combine both Vikings kickers (Reichard and Parker Romo), and possibly falls to 11th if you count Fairbairn (who scored 7 fewer points, but also had his bye-- he leads in points per game). He's making 83.3% of his kicks-- right in line with his career average-- and has scored 35% of Pittsburgh's points-- right in line with expectations based on their 20ppg average.

There was also a question raised at the time of whether hyper-aggressive coaches like Nick Sirianni and Dan Campbell were especially bad for their kickers. As mentioned, since this thread was posted Elliott ranks 6th in points and Bates ranks 8th. Elliott has scored 31.5% of all points for Philadelphia, which is just a hair below league average (as expected for kickers in top offenses), while Bates has scored 27.4% for Detroit, which is fairly low and somewhat concerning-- though as mentioned, he's still 8th in total points because a rising tide lifts all ships. (He's also seen his accuracy decline significantly-- down to 72.7%-- though one of those misses was from 65 yards, so I don't think that's especially meaningful.)

I'll try to post final numbers next week. I think it's worth revisiting not because my hypothesis was vindicated (when your hypotheses are wrong, it's even *more important* to revisit so you can learn and correct going forward), but because this is how regression works. There are always case-specific factors one can point to suggesting that this is the instance where regression isn't going to apply. Chris Boswell is uniquely good, Mike Tomlin is uniquely conservative, his offense is uniquely prone to stalling out in field goal range, he is uniquely positioned to sustain levels that for anyone else would be unsustainable.

But those case-specific factors are virtually always a mirage. And it's good to document that so the next time we have someone like Chris Boswell and we start to convince ourselves that This Time Is Different, we can remember that we thought last time was different, too-- and it wasn't.