The one thing I wish analysts would step away from is using total yards to determine the best defenses and offenses. That is a dubious metric at best.
Oh, I know the yardage stat is useless. The Saints did after all win 7 games and actually managed a shut-out vs Tampa Bay. To me the comments sound like they're coming from someone who used to be king of the locker-room and who could audible and do weird shit cause Williams loved him, I'm guessing Vilma or Harper. It'll be interesting to see what happens, I'd be surprised if we got rid of Spagnulo, it'd be a lot cheaper to get rid of the clearly over-paid and discontent players instead.
No, I just picked Seattle, like I picked Morris, as examples of how their output seems flawed. But since I can't see the actual inputs or what they input the numbers into...all I can do is look at the output. And then ask questions. Like Morris getting dinged for being stuffed on 3rd down. Most of us know short yardage conversion is a lot on the offensive line; FO ranks the Skins offensive line 24th in power runs (defined as 3rd or 4th down short yardage). Is their own offensive line ranking taken into account into Morris's 3rd down rating? Who knows? What is the weight they give to different statistics in coming up with DVOA? For example, how much does an INT count in their ratings versus a TD? Who made that determination and what is the reasoning for that? I don't have a problem with new statistics. But anyone having any basic knowledge of statistics knows...you need to understand the variables and formulas for the outputs to have any real utility. We'll never know as long as they keep this stuff hidden. But as I said in my first post, I would be careful with their conclusions. You just don't know hiw they're reaching them. And I think I'm done on this subject.
I do. You would too, if you bothered to actually read up on how they do their work. It's pretty obvious you don't intend to, so probably you're right that it's best this conversation come to a close.
Ack I can't resist. I'm honestly glad you do, and I do appreciate you discussing with me. But see: How did they arrive at those figures? Why are interceptions, on average worth a greater negative than the "worst" fumble? They just give us their ideas, but as far as I can find, never explain why they arrived at those values. Hopefully you can help explain or point me where they explain this.
Because the other team necessarily gets the ball on an interception, whereas a fumble is not always lost. Is this really so hard?
No. Read the quote again. So necessarily, "always lost" is worth -4. Why is it worth -4 then whereas an interception averages -6?
There's that; while giving up a fumble is bad, you've got somewhat decent odds of recovering it. There's no chance to recover an interception. As for how they get at the numbers: from what I gather it's basically historical analysis: you try and figure out what impact giving up interceptions has on the overall ability to win games. Regression analysis comes in handy here, along with some level of subjectivity (interpreting statistics is an art, after all!) Certainly there's room to disagree with the specific success points they assign to various events (we agree an interception is bad, but should it be -5 bad or -6 or -3 or what?), but IMO that's not a reason to throw the baby out with the bathwater.
Also an interception is always preventable by making a more accurate throw and/or a more sure catch, whereas even a well secured ball is still at risk of fumble. Plus perhaps fumbles are effected more by the weather. We don't know the factors no, but the way they arrive at these decisions is by studying football games with statistical analysis. If their numbers say interceptions are more likely to alter the chance of a win than a fumble, then it makes sense that they have a heavier penalty. If making 3rd down conversion receptions are more likely to influence the score than running for five yards on 1st & 10, then it makes sense for that to be considered a greater success.
That's not actually referring to if the fumble is being recovered, but rather to the type of fumble that took place. The idea is basically that something like an aborted snap fumble that offense usually recovers is not as bad as say a fumble by a wr that the defense recovers. The whole premise behind that being that fumble recovery is random (well, depending on the type) so it doesn't make sense to punish a player who had particularly bad luck with getting his fumbles back versus one who had good luck. http://www.footballperspective.com/the-definitive-analysis-of-offensive-fumbles/
You know we actually agree here, jeff. Go back to my original post; my point all along is be suspicious of their values because they're not transparent in how they get to them. Their stats are interesting, but again, not dispositive. Even you, who understand these stats better than most, still are working under an assumption of how their formula work. As you aptly point out, there's some level of subjectivity here...but we don't know what level, and we don't know how subjective their analysis is.
Pretty much. Most basically: we can create a very simple regression analysis that expresses win percentage as a function of interceptions per game. From there we can start adding all sorts of control variables in order to try and isolate the effect of an interception. It's not perfect, but you'd be surprised how quickly you can get a fairly decent 5% confidence interval for how much each interception affects your win percentage.
Yeah I acknowledge that, I just don't think they're shit, which, I recall, is the term you used. :) That may be because I've got a friend who works for them so I've got at least a bit of insight into their process. If you go over to their forums, you'll also find they're fairly open about discussing things. They might not be willing to tell you what their numbers are, but they'll probably go into some detail about how they arrived at them if you ask. Also note: their system is better than most subjective power ranks. The fact that we're having a conversation about the relative importance of fumbles vs. interceptions puts us in a far loftier realm than just "Seahawks Rule!"
Good link. However, there's quantifiable data on fumble recoveries...so why doesn't FO use that as well? Why do they count a fumbled QB-center exchange the same as a fumbled QB snap, when your link shows historically, the recovery rate is significantly different (delta of 25%)?
With respect to the shit comment, no they're not shit so if I said that, retracted. They're infinitely more interesting than QBR, for instance, which is shit. But without more info, interesting is all they can be. With respect to subjective power ranks, I don't put any value in those, either. In the end, all that matters is how the season ends. Who really cares if ESPN ranks your team 7th before the season when you end the season 4-12 (hi Eagles), or if you're ranked 1st at the end of the 2007 regular season if it doesn't end up in a Superbowl win... Hell, Mel Kiper just named Andrew Luck on his all-rookie team ahead of Wilson or RG3, and I (and you, I suspect) couldn't care less. All that matters to either of us is that we have our own QB and we're happy about that. Edited: I guess this is a bigger deal for me simply because what I do for a living. I work in the pharma industry, where our clinical trials are run through extremely sophisticated and open statistical analyses, to avoid any appearance of subjectivism. Because the other word for subjectivity in statistics is bias
I'm guessing the trolling comment was meant for this: You probably meant sack and not snap. It confused me too.
No, we don't care about what Mel Kiper thinks because he's known to have bad methodology. History shows us he's a poor evaluator. If some other random no name guy shows up and claims Luck is the best rookie QB and provides a reasoned argument, then that's interesting. I would like to read that. I might be skeptical but I want to explore these kinds of interesting questions rather than shouting that every other opinion is wrong 'cause I say so.
If you think that's what I've been trying to do, than either you have misconceptions or I didn't communicate it well. Go back to my original post on all this which says be suspicious of their data, as they present these as dispositive statistics but no one in their right mind should be accepting them fully until their methodology is transparent. I've only posted that like oh...6 times, including a few posts above this, so at this point I'm thinking you're just being willfully obstreperous. Or just a troll that likes to throw cusswords in all of his posts, because you know, thug life on the internets is awesome.
Less arguin', more footballin'! Ray Lewis intends to play Sunday, and has announced that, either way, he's retiring after the playoffs.
Where did anyone say you should accept them fully? Yes. I am obstreperous with people who act like trolls and refuse learn about what they claiming.
Some interesing head coaching rumors: Andy Reid is apparently likely going to end up in Kansas City of all places. With pretty much full control, which is ironic. Reports are that the Chiefs offered him everything and so he cancelled his scheduled interviews with the Cardinals and Chargers and is working on the details of his contract. Scott Pioli will likely be canned and someone Reid likes (Heckert probably) installed as GM. Bill O'Brien (just finished his first year at Penn State) was a hot topic but is now saying he's not going anywhere. This is after he supposedly interviewed with the Browns. Turns out Penn State put a giant buyout clause in his contract that could be as big as $15m. So any NFL team that hired him would have to give him a competitive salary and give him the cash to cover the buyout. Sounds like nobody is willing to do that.
Speaking of our long national nightmare being over, Tony Siragusa has been benched for the playoffs by Fox.
YOU TAKE THAT BACK! Andy Reid is friendly and cuddly and we will love him forever until we lose a game.
How can you not love that man? You just want to wrap that mustache around you and feel all warm and protected. Feeling protected would be a new thing for QBs in Kansas City, so this is a pleasing turn of events!
Aw, the way you phrased that led me to believe Kenny and Moose would be doing the games without Siragusa, which is my personal dream. Billick is good though, too bad we're still stuck with Aikman and Buck every week. If Reid can replicate what he did in Philly, Chiefs fans will be overjoyed. Meanwhile I am praying that the Eagles sink into a cycle of replacing one incompetent coach with another, bad drafts and 4-8 win seasons. Philly fans have no idea how good they had it with Reid. Even if Jim Harbaugh never wins a Super Bowl, I'm not going to be calling for him to be fired because it's a hell of a lot more fun watching my team now than the previous decade. There are a lot of things worse than losing in the Super Bowl or NFC Championship game, and I hope Philly fans get to experience all of them.
I know you are, I'm just wondering if you'll still be thrilled watching the Mike Nolans, Steve Spurriers and Josh McDanielses of the world.
