RAS.football

@MathBomb Top 7 NFL yards / game over the last 5 years and their RAS. Wasn’t expecting this much uniformity.

@Nfldraftregs Generally, but thats a top tier elite cone. Most guys aren't running that and its totally fine. Many guys don't NEED to run it, they dont win with agility. You shouldn't care in those instances, but people make claims about their motivations and work ethic anyway.

I’m not into assumptions but do know from watching an insane amount of football and studying film that guys who skip agility drills lack agility on film and go on to not be agile in the NFL. So, maybe they aren’t skipping the drills because they are scared, but if an RBs thought he could run a 6.6 3 cone, he’d do it in a heartbeat.

The idea that players are "scared" to test is ridiculous. That the only reason they could possibly skip is because they think they'd do poorly is both silly and not supported by data. That players MUST do all tests, or specific ones, is also silly and counterproductive.

@LARSN00TBAAR @draftjm13 @LanceZierlein Correct. I had planned on updating the size grades to better reflect what they actually mean, but I ended up not getting anything done this offseason instead.

@draftjm13 @LanceZierlein iirc the scores are pretty much percentile? (@MathBomb correct me if im wrong) so those are 45 (ish) percentile in each. I wouldn't consider that poor but 🤷‍♂️

Feels like one of the safer WRs in this class as a day 2 pick.

@JJPackfan Were getting way into the weeds, which is awesome, this has been a great conversation, but we'll have to pick it up another time. Im heading into a dead zone for a few hours, will have no phone.

@MathBomb How does this not lead to what I said? Maybe I’m not fully following.

@JJPackfan Lol, nobody is playing to my system. There are teams that use my metrics, but even those have their own internals that they use. The tests have remained unchanged for 40 years, aside from cone which was added later. Theyre playing to the tests, but also to the scouting process.

@MathBomb I dont know. Current method leads to a LOT of 9+ RAS. Likely because guys now only train at what they’re already good at - to test amazing at 1-3 tests, then skip what they’re not great it. They are playing your system. Which happens to work against teams too (real reason).

@JJPackfan Scouts saying something all the time doesn't make it true. Many of the players that have had that said about them in this class were simply injured.

@MathBomb So you have survivorship bias compounded by improved situation. Can’t use that to make a sweeping generalization of players who skip tests at combine. Anecdotally, scouts say all the time that given player skipped a test because they would be bad at it and they know from film

@JJPackfan Most players ONLY test at their pro day, while some only test at the Combine. You mention survivorship bias, but a player testing while knowing he has to beat his previous number is more likely to test better, they have kept trying to improve because they know they need to.

@JJPackfan Would ir surprise you to know that your first note is not true as stated? Several flaws in looking at the difference in times. For one, 90% of all testing is pro day, you'd be basing a metric on 10% coverage and applying it across the board.

@JJPackfan This isnt geared towards football heads, thats not the purpose of RAS. It works for them, but it is designed to be as simple as possible with the fewest assumptions that is easiest to understand. It accomplishes that very well.

@MathBomb People in FF community have been doing this for decades. It’s not as big of a deal as you’re assuming. Explain it, link it, and most people will understand. And those who don’t - that’s on them 🤷‍♂️ Greater accuracy should be the goal.

@JJPackfan Its not an unreasonable idea, but the data doesn't back it up. You end up with less fair results that are more difficult to explain, and because of who it impacts the most you would end up having to defend it more often with lower quality data, the alternative works better.

@MathBomb I’m not exactly proposing to add a penalty either. With very little thought on a solution, assuming a 5 (average) for non tests seems fair and more accurate. Assume athletes are average until they show otherwise. That seems reasonable.

@JJPackfan It IS based on objectivity and optimization. The public perception would be bad because its a worse design with more flaws thats harder to explain. Its a byproduct of making a metric based on an arbitrary penalty that makes more difficult to justify outcomes.

@MathBomb Basing any outcome on the public perception is a terrible process and not trustworthy. It seems you’re saying the results are knowingly biased and skewed. Objectivity and optimization >>>> public perception

@JJPackfan It is based on an average, the scores are averaged and then scored the same way so we get that equal distribution from 0 to 10.

@MathBomb So RAS formula is not based on an average of testing numbers OR based on %ile relative to size?

@JJPackfan It doesn't, though, because you oversimplified past the actual methodology. Not that you were intentionally doing it, but that created a straw man argument. I can tell that wasn't your intention, but that's why it's important to point out those details.

@MathBomb Feels like you’re overlooking my point because of a technicality here… Assume the players performed your min # of tests. My point still stands.

@JJPackfan I've done both 0s and 5s. Same problems for both, in varying degrees, but both are worse than the current method so far. Significantly, as you'd expect, from the first case.

@MathBomb Based on what proposed solution? Filling in 0’s for non-tests? I never proposed that 🤷‍♂️ I didn’t provide a solution. I get it’s difficult. I’m just pointing out that there IS a flaw currently.

@JJPackfan The math says that a player who tests well is significantly more likely to test well than poorly, and vice versa. So if you make those assumptions as you've described, it's a far better assumption than assuming across the board we should penalize players who score well.

@MathBomb It does not hurt players with lower scores, because RAS is an average or %ile, correct? So if a player does 3 tests and has a RAS of 4, that is assuming the other tests would be equally as poor. High RAS assumes non-tests would be equally as high. Which is a bad assumption.

@JJPackfan It doesn't make you stupid for thinking that it would work, you only know what you know and you're not looking at the data like I am. But it is stupid. It's worse in every way. People with no stakes at all would be theorizing how we could not penalize players instead.

@JJPackfan You're not the one that has to answer to it. You have no stakes, so you can safely say that without batting an eye. I have to hear it. And I've done enough research to know which way is going to get the most scrutiny, and the most DESERVED scrutiny. Adding penalties is stupid.

@JJPackfan The NFL does not care about the testing in the same way as fans, and players certainly do not, and they don't care about these things for different reasons. It's zero stakes for you or me, for them it's $$$. Doing a drill well that you don't need to do is nice, but that's it.

@MathBomb And if the players jumps and agilities truly ARE 95%ile like their 40, they WOULD participate. It would be stupid not to. The ONLY real reason to skip is because they are less than 95%ile in those tests. Better to let people’s imaginations believe they ARE that athletic…

@JJPackfan I'm not done doing my full review of the data yet, but here's one example of how bad applying a penalty is. Calvin Johnson did 7 tests. There are 250 WR who scored worse than him in EVERY drill he did who will score better than him overall. That's wholly ridiculous.

@JJPackfan Nope. A player who only runs the 40 does not qualify for a score. And you're literally doing what I just explained, only assuming it provides a benefit, disregarding a majority of players who score.