Well, if nothing else that game should give Easterbrook something to put in his column. In the third quarter, Washington punted from around midfield on 4th and 1, 4th and 1, and 4th and 2. And now they’re 0-1 :).
…And when Zorn didn’t have them go to the no-huddle when they were down by nine with five minutes remaining, TMQ wrote the words “Game Over” in his notebook.
In what way? If you mean you were incorrect when you said it, agreed. If you mean you later realized the error of your ways and retracted it, works for me. If you mean you didn’t say it…
Good point, and a logical explanation.
I was never debating about the future carries of “a RB.” A was debating about the future carries of one particular RB: Tiki Barber. The guy who led his team in rushing 80 consecutive games. What’s the statistical odds of that happening?
I watched the guy play. I saw how elusive he was, and how his greatest strength was in turning almost every crushing hit into a glancing blow with his use of angles and body control. And how his body didn’t even have that much wear and tear to begin with due to his rank on the depth chart for the early part of his career.
Ah, I see. I thought you were referring to the post I quoted and just mis-read the year, because I mentioned Week 9 in that post and you said I predicted he would fall apart in the second half. (Where did you get the part about the second half, then?)
Anyway, “crumble” may have been overstating what I believed (as should be obvious based on the post I quoted). It was a question of possible outcomes. Say there were four possibilities for Tiki’s 2006, his age 31 season, coming off one of the greatest RB seasons ever: improve, maintain, some decline, and collapse. Looking at history, what percentages would a reasonable person apply to these possibilities? Without once again pouring over comparable RBs or doing arithmetic, off the top of my head those percentages might look like this:
Improve: 1% (nowhere to go but down)
Maintain: 42% (he had 35 fewer carries in '06 but was basically the same player)
Decline: 42% (as in, say, 1300/4.4 following his 1860/5.2)
Collapse: 15% (as in 1000/3.9, or a season-killing injury [the latter being perhaps more likely])
If I had to pick a single scenario as most likely, it would have been something between Maintain and Decline, a 1500/4.8 type of thing. I was hardly being a hater (or even a pessimist) because I didn’t expect him to have the greatest age 31 RB-season ever and you did, even if you were ultimately correct.
In the long run I think you get into trouble with this kind of analysis. Every now and then you’ll correctly predict something unlikely – and many of those times you’ll even be correct for the reasons you cite instead of coincidence – which is cool, but over time you’ll wreck your batting average this way; we’re biased, our eyes deceive us, and we have incomplete information. In contrast the numbers are what they are, and used correctly they can tell us how likely or unlikely certain developments are. They often don’t tell us why something is so likely or unlikely, and that’s where scouting can help you find exceptions, but, as fans, the more often we predict apparent statistical long-shots to come through based on what we see on TV, the more often we’ll be wrong.
I honestly haven’t a clue where I got the “second half” part.
I’m not convinced that statistical analysis is particularly reliable when applied to the NFL. You yourself explicitly stated in a recent thread that you would like to see the schedule increased to 18 games because 16 games simply isn’t enough to provide reliable statistical analysis.
Also, NFL football is incredibly variable year to year, so I consider statistical analysis of multiple years fundamentally flawed. There is so much roster turnover, and even without changing personel the injury factor makes the same roster radically different week to week; the year to year differences are staggering.
That’s why when I hear something about there is no statistical correlation with a specific “ability” like, say, recovering fumbles, I shrug and say “so what?” It’s a completely different team on the field the next season, so I wouldn’t expect the same results to begin with. Even when broken down within the same season, injuries and familiarity/comfort with scheme vary so wildly I don’t think there is much legitimate analysis to be made.
As an example, look at the difference between the first two Giants games in 07 compared to the rest of the season. Bet that fucked up all the defensive predictions for the better part of the season, even though subjectively it was clear as day that they had come together by the end of the third game.
Sort of; I said I wanted a bigger sample size to determine who the best teams are for the playoffs, and that such would be especially important if parity continues to grow. For individual stats there are definitely sample-size issues, but they’re not as bad as only having 16 data points to determine who are the best 12 teams in football. A full time QB might have 500-600 plays in a season for evaluation, about the same as a starting position player in baseball; A RB might have 400 plays, a WR 150, and so on.
We’ve been down this road before, you and I.
The thing about statistics that you I’m not sure you appreciate is that they can not only tell you how probable something is, but they can also tell you how likely that probability is to be accurate; the fumble recovery thing, for example, is pretty much settled fact at this point. Yes, roster turnover exists, but so does roster continuity. There’s a very strong year-to-year correlation for virtually any stat or activity on the football field, in spite of all the different players out there every week, because most of the players are the same. In order to maintain that fumble recover is a repeatable skill, you have to come up with a reason why it’s the only such skill not to correlate from year-to-year or week-to-week; i.e., why is it that being good at it one year makes a team no more likely to be good at it next year, and why is the skill unique in this regard?
Expanding to the larger point, you say that statistical analysis of multiple years is fundamentally flawed; well, ok, but then you have to explain why it works, and why you yourself use it. Even if you’re just saying “Dallas scored a lot of points and won 13 games last year, so they’re likely to score more points and win more games than San Francisco.” Points and wins are stats; if using them to predict the likelihood of future results works (which it obviously does), why wouldn’t you expect it to work if you dig a little deeper and consider more facts in your analysis?
Now that doesn’t mean you can’t color your statistical predictions with some scouting and common sense – after all, there will always be information that the model isn’t considering. You might say, for example, “Sure, 75% of RBs with careers similar to Tiki’s decline either a lot or a little heading into their age 31 season, but the Giants are returning almost all their offensive lineman and Tiki doesn’t have a history of nagging injuries, Tiki hasn’ taken a lot of hard hits in his career, and Eli will be better so opposing defenses won’t be able to stack the line as much, so Tiki is less likely than most RBs his age to decline.”
That’s fine. It’s just those factors are all very hard to quantify, and they might be countered by even more unquantifiable factors, so it’s also very reasonable (and probably, for most fans, a lot more accurate in the long run) to focus on the most glaring factor: “Tiki’s a 31 one year old human being, and one thing we know for sure about 31 year old humans is that, no matter how good they’ve been up to that point, they tend to have a very hard time playing Running Back in the NFL.”
Not if you did them right. FO, for example, provides numbers for the year to date, but for purposes of predicting future success early in the season they’re still weighting their preseason projections much more heavily, as after two games you don’t have a lot of information to go on, and there’s a lot even about those two games you don’t really know yet (for example, that Dallas and Green Bay were two of the best offenses in the NFL). And also, since teams do improve or decline over the course of the season, they use weighted rankings from the midway point on that increasingly discount early games as the season progresses.
It’s not perfect, but neither is saying that they obviously came together by the end of the 3rd game: did they, or did their opponent choke and make them look good? Or did they just have an unusually good second half (like Omar Vizquel hitting a home run)? Just as we didn’t have enough information to say that the Giants’ defense was bad after the first two games, we also didn’t have information to say that the defense was good after three games.
How’s the correlation on defeinsive interceptions, or are those random too? How about field goals? Dropped passes? Penalties? Just how much of the game is “random”, anyway?
You know what? Ignore that last post of mine. It’s just a gut reaction to something that my head might accept, but offends me on a visceral level. I think I understand what bugged me so much about fumble recoveries, and it’s not that they aren’t random. They are most definitely random, and that is by design. The ball is intentionally shaped in a way to ensure random bounces.
What bugs me is the context under which you brought them up. You don’t want fantasy scoring to include fumble recoveries, and your stated reason is because they are random, and therefore aren’t earned. To my mind, with what goes on at the bottom of a pile, fumble recoveries the most earned thing in all of football.
I concede and accept that not only do statistical tendencies carry over from week to week and season to season, but also I personally cite them as valid in exactly the way you describe.
What bugs me about the statistical tendency debate with you in particular is that you seem to rely solely on them as evidence. Take the Tiki Barber debate. I would bet that most people, after seeing him in 2005 and 2006, would reasonably have expected him to be great again in 2007. They might not have been able to put their finger on exactly why they would lean that way like I did, but not everybody is a Giants fan so they can’t be expected to have watched the team as closely as I do.
And that’s exactly what bugs me. For a Giants fan, you seem to not watch much Giants football, or at least not watch it very closely. Tiki’s skill going to crumble by 2006? Let’s reserve judgement on Justin Tuck despite him being an absolute menace in every opportunity he’s gotten over three seasons? It’s things like that that drive me bonkers.
So far, You know what? Ignore that last post of mine. It’s just a gut reaction to something that my head might accept, but offends me on a visceral level. I think I understand what bugged me so much about fumble recoveries, and it’s not that they aren’t random. They are most definitely random, and that is by design. The ball is intentionally shaped in a way to ensure random bounces.
What bugs me is the context under which you brought them up. You don’t want fantasy scoring to include fumble recoveries, and your stated reason is because they are random, and therefore aren’t earned. To my mind, with what goes on at the bottom of a pile, fumble recoveries the most earned thing in all of football.
I concede and accept that not only do statistical tendencies carry over from week to week and season to season, but also I personally cite them as valid in exactly the way you describe.
What bugs me about the statistical tendency debate with you in particular is that you seem to rely solely on them as evidence. Take the Tiki Barber debate. I would bet that most Giants fans would reasonably have expected him to be great in 2006. And that’s exactly what bugs me. For a Giants fan, you seem to not watch much Giants football, or at least not watch it very closely. Tiki’s skill going to crumble by 2006? Let’s reserve judgement on Justin Tuck despite him being an absolute menace in every opportunity he’s gotten over three seasons? It’s things like that that drive me bonkers.
The last two guys I called for breakout status before it became the general concensus were Eli and Tuck. I have one to add this season: Steve Smith. Unless he radically improves against the Rams, I think Toomer should already lose his job to Smith on the depth chart. I was torn on which one of those two to draft in the auction league, but I just couldn’t believe Toomer would fall off the cliff like he seems to have. All those “should’ve been picks” on Thursday? Toomer was the intended receiver.
To offer context, while I like Gerris Wilkinson a lot, I’m not expecting him to be a flat-out stud. Just very good. Similar to a Gibril Wilson. Smith, OTOH, I expect to be great.
Interceptions I’m pretty sure I can answer another time when I’ve got the right book with me, but there’s a meaningful positive correlation (as there is with fumbles and forced fumbles; it’s just the recovery rate that’s random). The year-to-year correlation for field goal percentage is actually extremely low – obviously it’s a skill, but the difference between good field goal kickers and bad ones in the NFL isn’t very large, and there’s so much variance in the stat even for the best of them that, when picking a kicker, a team would do better to pay more attention to his abilities as a kickoff man, since those are extremely consistent from year to year.
I’m not specifically aware of anyone running a correlation for team penalties, but I have to imagine there’s it’s pretty high. I remember growing up that the Dolphins always seemed to have the fewest penalties, and the Raiders the most. (Fact: until recent years, there was a actually a small positive correlation between penalties committed and winning percentage; apparently the teams that were getting caught the most were also getting away with it the most.)
Not much, as far as we know. Fumble recovery rate. Probably blocked kicks. Turnover returns aren’t exactly “random,” but they are non-predictive events – they’re so rare, and so dependent on context (having blockers wake up in the right position, recovering the ball in the right part of the field, etc.), that you just can’t use them to make any predictions about how a team will do returning fumbles and interceptions in the future.
It occurs to me that it’s mostly special teams. Field goal kicking and punt / kickoff returns.
Yeah, I see what you’re saying. Part of it is that I habitually maintain a robust agnosticism; not just about sports, but lots of things. Seriously, talk to me about anything in real life for 15 minutes and I’ll drop “I’m not sure” and “it’s not that simple” a half dozen times. Indecisiveness has become a personal philosophy at this point.
That’s what interests me and that’s what I know a lot about, so that’s definitely the angle I tend to take. I’ve made scouting-based predictions in the past. After his rookie year I predicted Jacobs would be excellent but injury-prone; the stats didn’t back me up on that, but it’s looking good so far. My feelings about Kiwanuka are based solely on visual inspection, as the numbers certainly aren’t there. Of course, back in the day, I also thought that Kent Graham would be an above average starter and that the Giants were crazy not to play him.
Believe me, I’m watching the games, but I’m not watching each one three times from multiple angles and taking notes, and it’s so complex I know I’m missing 90% of what goes on during the game. I’m also missing 99.9% of what goes on off-field – generally speaking, I don’t know if a guy is lazy and stupid or brilliant and hard-working, I just know he can run fast (or not) and tackle well (or not).
So, I don’t know if Tiki has good genes that will allow him to age gracefully, or if he’s a master of avoiding hard hits in bad places. But I do know that RBs tend to slow down in Year X and he’s in Year X+2, so if I’m going to make a prediction I’ll go with the percentages and say he probably won’t be the same guy, because it’s hardly ever happened before. I don’t know if Tuck is great or merely good, but no one outside of his coaches has seen him play DE much, and I do know that players who excel in part-time roles often come back to earth as full-time starters (e.g. Lamont Jordan), so I’ll take a wait & see approach, with the caveat that I think he’s been great so far and has the potential to be great in the future. (By the way, go easy on Football Outsiders – they put your boy Tuck on the cover of this year’s book, with the caption “The new pass rush prototype.”)
They’re absolutely earned on the field – they’re just not earned by fantasy owners, is my objection. Since predicting them is completely futile, it’s nothing but dumb luck on our end.
Fair enough, but so are field goals and return touchdowns.
Other than the year before that in the very same stadium, sure, super rare.
Heh, touché.
True, but there had been 30 year old RBs with super-fantastic seasons. There has never been a RB with an age 31 season as good as Tiki’s (Walter Payton and Curtis Martin come close.)
I hate to interject on a small point, but field goals are predictable enough to not be anywhere near random. It’s impossible to determine what lengths, but certain teams have far better red zone defenses than others (the “bend but don’t break teams”) and when they go against offenses with bad red zone efficiency, you can usually count on a couple field goals. It’s actually pretty simple to find the right kicker matchups.
Kicker FG percentages vary so wildly from season to season that you can pretty much ignore them and judge kickers based solely on kickoff distance, which is pretty consistent.
Division is looking tough, with Dallas and Philadelphia getting impressive wins today. It’s nice to see the Giants start off the season in sole possession of first place, as it should be.
I told you guys the Eagles are no joke. Then again, maybe the Rams are. Hopefully the Giants won’t lay an egg against them next week.
They are.
Hmm, it appears that my less than stellar review of the Iggles, may in fact be baseless. I still don’t think the Eagles receivers can keep up that kind of performance all year though.
Ha! Satire is the sincerest form of fla . . . well, ok, no it’s not, but this is still pretty cool IMO.
(Actually, I’ve always wanted to try a league without kickers, just because they decide so many games even though no one does fantasy football to match their kicker-evaluation skills against the competition. Another option might be to give points for FG attempts, since those are much more predictable than FG accuracy.)