Doubling down is an interesting approach here.

"There are a number of readers who are unhappy about the way our numbers have judged the Giants over the course of this season, and they are not going to be happy now."

Should have added-- Dude, we won the Super Bowl. I think we are going to be quite happy!

"Sorry, these are the numbers."

And I thought the numbers were 17-14, and 18-1...

I would argue that VOAf would be the best judge of "who outplayed who on that given day", not adjusting for opponent. Aaron may not be willing to say "should have won", but I'll go ahead and say that.

Anyway, by that measure, three Super Bowls have been won by the team that played more poorly, with one too close to call. The Pittsburgh one doesn't surprise me too much, but NE/NYG and NE/CAR do.

Statistics are only useful for a large sample. To look at one game, it's really not that meaningful.

On the other hand, the numbers do reflect that the game was close - and you don't need to be a genius to know that Patriots almost won the game.

It's the Giants D that sets the tone of the game - you can't learn that from stats.

"Aaron may not be willing to say"

That is clearly what he said here.

Bill Belichick, for some inexplicable reason, decided not to challenge the decision to give the recovery to the Giants.
Aaron, you also echo this in Audibles, and I tend to disagree. While it did appear at first that Woods recovered the fumble, subsequent replays showed that, while Woods was indeed first to fall on the ball, Bradshaw did get his arm underneath Woods before Woods could have secured sole possession.
Therefore, while it very well may have been that Woods recovered and was down by contact, there is almost no way I could see Mike Carey overturning the call on the field. Had Belichick challenged, it would have been a classic case of a coach challenging a call he wished had gone his way, not a call he actually believed could get overturned.
------
As to the VOA for the game, I would have figured the Giants would have come out on top (the Patriots had too many three-and-outs, it seemed), but it doesn't really surprise me that the Patriots rated higher. What does surprise me is that the Giants defenses rates below average (above 0%). Understanding that there is no opponent adjustment, they still had five sacks and forced some three-and-outs. I imagine they are hurt by the fact that the Patriots were 2-for-2 in the red zone and had several long third down conversions.
All in all, I think this game may have proved more about the Patriots than about the Giants. It showed the Patriots were not the infallible juggernaut that so many painted them to be. And while I truly believe the Giants are a better team than their regular season DVOA indicates (but they were mediocre during the regular season - something "clicked" for them beginning in week 17), they are clearly not better than the Patriots. New England played either their worst or second worst game of the season, and the Giants still needed a last-minute touchdown to win. That is not the mark of a great team.
However, congratulations to the Giants and to their fans. I was rooting hard for them last night and was very glad to see them win. Hopefully we have more exciting Super Bowls like this one in the coming years.

I'm sorry, but how on earth do the Giants get a defensive 6% VOAf for Sunday's game?

They limited the Patriots to 17 points, forced a fumble, had 5 sacks, and a pretty decent 3rd down conversion %. And VOA thinks they were below average?

I'm sure they wouldn't get anywhere near the -43% DVOA awards them (which also seems low to me, but whatever), but 6% VOAf?

Something's wrong here.

re: 6

[T]he Giants [...] are clearly not better than the Patriots. New England played either their worst or second worst game of the season, and the Giants still needed a last-minute touchdown to win. That is not the mark of a great team."

I think that the Patriots played their worst or second worst game of the season *because* the Giants completely disrupted their offense. About the only time the offenses clicked was on their scoring drives; at the beginning, and when the defenses were tired. With the NY offense / NE defense, that's about as expected - they Giants have an above-average unit, but so do the Pats - but on the other side of the ball, it was a huge surprise. I don't think you can (or should) diminish that by saying that the Patriots "played their worst game". I think a more appropriate take would be "the Giants played their best game" - and yeah, there's a difference between them.

I think that, throughout the season, the Patriots were the ridiculously superior team. However, as of now (meaning not only yesterday, but should they play again tomorrow, for instance), the Giants are on par with this NE team.

I don't know. I don't mean to imply any tweaking of the numbers to purposely rate down the Giants, but if the numbers say their defense was below average after *that* defensive performance... It was impressive enough in and out of itself, before you even consider it was the freakin' 2007 Patriots getting stopped cold.

Perhaps in calculating Toomer's stats, the push-in-the-face play should have been subtracted?

Sorry, but if we're going to talk about "luck", we might also want to talk about flags never thrown. I just find it hard to stomach the notion that you are praising Toomer when his biggest catch was bs.

The weakness of DVOA/VOA is seen with the problem with Moss's numbers. Moss would have been wide open on the third down pass if Brady had not underthrown the pass to a point where the defenders could catch up. Given how easily Moss burned the Giants' D Week 17 (two plays in a row!), I was quite disappointed this option was not tested more by "genius" Josh McDaniels.

Oh, and re:3, I think that's just wrong. Of course you're going to play worse against a particularly great opponent; the whole purpose of DVOA is to take that (among other factors) into consideration, so what would otherwise be a fine, but not by any means astonishing, defensive performance, is taken with the proper perspective.

Anyone can hold the Miami Dolphins to 17 points, get 5 sacks, etc. Not every defense can get the same results against the most prolific offense in NFL history, and that alone is enough to say the Giants outplayed the Patriots yesterday, by a large margin.

Disclaimer-- if this post appears to be arguing with DVOA over the Giants, I apologize. I really couldn't care less if DVOA says the Pats outplayed the Giants (they didn't) or not.

I can see the rationale for penalizing the Giants offense for the fumbles that they recovered all of. But honestly-- why would you credit the Pats defense with them?

Take the Bradshaw fumble. Did the Pats do anything at all to cause that fumble? No. While fumbling is a characteristic of teams and players, I find it hard to conceptualize that certain teams make it more likely that their opponents will muff handoffs. To my mind, that play should certainly count against the Giants offense because of the risk of turnover. However, somehow giving a bump to the Pats D there seems wrong to me-- they didn't really do anything but watch the Giants screw up, which was totally not in their control. Why would that count in their favor?

The other fumbles, the defense should get credit. They both came on sacks where the defender did something to put the ball at risk.

Same thing goes for the interception. Absolutely that was a bad play by the Giants offense, and should count against it when figuring out how well the Giants offense did. But if Smith holds on to the ball, as he should have, then the Pats don't get that interception. They did have to catch the ball, so at least they did something (unlike with the Bradshaw fumble), but they didn't have to do much.

I guess I am saying I don't think that the whole concept of it being a zero sum game between the offense and defense before opponent adjustments is necessarily the best approach. It just feels like it is missing something. Perhaps it is missing something that could not be applied reasonably within DVOA because the information is not always present, but that would just be a limitation in how DVOA could be calculated.

Same thing for drops and other plays that could/should have been made. Take, for example, the play where Manning avoided a sack and then lofted the ball 3 feet over the head of a wide-open Plaxico. Once Manning avoided the pressure, that was an absolutely wide-open pass. It should have been a huge gain, and wasn't merely because of something Manning did, not because of anything at all that the Pats D did. Similarly, why should a defense get credit for when a receiver flat out drops the ball?

The Pats defense played well yesterday. They did not play as well as the Giants defense. They were helped out by a few things the Giants offense did, things the Pats had no control over, that made the Pats defensive DVOA look better than it should have. In other words, I think the Giants offensive DVOA is probably right, I think the Pats offensive DVOA is probably right, but the Pats defensive DVOA is probably overstated.

And if the same sort of thing happens throughout a season, which was the case with this Giants team that had a penchant for dropped passes and for turnovers in their own end (many times unforced), then that might cause it to appear that their opponents often did better relatively to them by the numbers than in reality. Might even cause a team to be underrated by DVOA, and have their fans saying "I think DVOA is missing something about this team." Because if you take the VOA numbers for the offenses as-is but take the defenses, and don't credit the Pats fully for that interception, don't credit them for anything on the play where Bradshaw and Manning couldn't get the handoff, don't credit them for the dropped passes-- don't credit the Pats for things they really didn't do anything (and vice-versa)-- I bet that the Giants defense would come out pretty far ahead of the Patriots defense.

Which we all saw with our own eyes, anyway.

Aaron, I really don't much like using DVOA to decide who outplayed whom in a single game.

The thing is that the teams were playing to win, not to rack up DVOA. If you told the Giants they needed to have a higher DVOA, not higher score, than the Patriots, they might have been able to do it. But since that wasn't their goal, you can't really say the Patriots "outplayed" them.

I dislike the Giants trolls that keep hammering on how the Giants didn't get any respect, but I don't think that the VOA comparison is particularly useful or necessary.

My apologies for saying DVOA instead of VOA in my previous post. So it should read more like this:

"If you told the Giants they needed to have a higher VOA, not higher score, than the Patriots, they might have been able to do it."

I think it also applies to the Steelers. At some point, they had the game won, and they were happy. They didn't know they were behind in VOA, and they didn't care. It's absurd to go back afterwards and claim they were outplayed because they didn't make up that gap in VOA, and instead settled on hoisting the Lombardi Trophy.

VOA works in the long run because the situation where the winning team posts a worse VOA is pretty rare, and things tend to average out. But in the short run, you can't use it to assert that a losing team outplayed a winning team.

I don't know how anyone could have watched the game and not seen that the Giants had some good fortune. Bradshaw ripping the ball out from under Woods was a great play. It was also a lucky play. The pass to Tyree was one you'd expect to be completed about one time in ten, and luck was obviously part of it. Doesn't change the fact that Manning and Tyree made a phenomenal play. Underdogs often need luck to pull an upset. Hopefully people can get that saying a team was lucky does not mean saying that team didn't deserve to win. All the luck in the world wouldn't have helped the Giants if their defense hadn't played its best game of the season.

Not a Giants fan, but just wanted to point out: DVOA can only account for things that are obviously lucky based on the play-by-play. The interception off of the low pass that bounced up off the hands of a reciever to Ellis Hobbs seems pretty lucky to me, but it is not possible to tell that from the play-by-play or quantify how lucky.

Anyone who watched both the Patriots and Giants in the playoffs and paid attention would come away with several obvious conclusions.

First, the Giants pass rush was excellent and forced two of the top two quarterbacks in the league, along with Garcia, to make errant throws and be generally less effective then usual.

Two, Eli Manning was not throwing interceptions and was playing at a much higher level than he had earlier in the year.

As for the Patriots, they were not nearly as impressive in the playoffs as they were earlier in the year when they ran up the score on weak teams to earn their premature reputation as the greatest team of all time. The two teams they played in the playoffs were not nearly as tough as the three that the Giants faced. After all, the Chargers not only played with a cripple at quarterback, they were also missing Tomlinson, the best running back in the league and the engine behind their recent success.

Despite all of this, the Patriots still deserved to be favored, however it was foolish to extrapolate data from early in the season and use that as proof as to why the Patriots were the best team of all time and the Giants the worst. Apparently the Giants realized one key fact which seemed to elude many self-appointed experts; the playoffs are a second season where the results of September and October mean very little.

Out of curiosity, did you choose to attribute the fumble to Bradshaw out of belief on sight that it was his fault or were you unsure and decided to give Eli the benefit of the doubt?

Eli's DPAR seems low for the first 3 quarters. I thought driving into the redzone, then petering out got a high DVOA ala the Eagles. It seemed like Eli converted a number of 3rd downs that should have got him a higher DPAR.

Re 9:
They can't adjust for non-calls because they use the NFL's official play-by-play. So unless you want that catch officially ruled incomplete, it happened and DVOA\DPAR has to take it into account. Also, I don't see how Aaron and co can account for bad calls and no-calls because it's not random every time. Fumble recovery has been proven to be random every time, so they can say recovering 3 in a game is good luck.

Re Tim #17 - While I agree with you to some extent, Aaron showed that the DVOA of the Giants and the Patriots since week 17 was virtually the same - despite the common fan's view that the Giants were playing really well and the Pats were struggling. Couple that with the data from weeks 1-16 and it is hard to argue that the Giants win was one that could have been foreseen (but I give you credit, you still say the Pats should have been favourites, despite the emphasis on the early season DVOA by Aaron et al).

Interesting too, that Brady still beat Eli in DPAR but they are virtually identical in DPAR/pass.

"Giants Had Luck On Their Side". LMAO!!! I KNEW that's the way Aaron would play it.

And then, to make it even more hilarious, there's a link to one of his recent columns "Here's How The Patriots Will Blow Out The Giants". It simply doesn't get any better than this. I'm enjoying the various lame attempts to explain this away even more than the game itself.

Allow me to point out the obvious here. If the "worst team to ever reach the Super Bowl" is playing evenly with "the greatest team of all time" then it's really NOT an even game is it? The heavy underdog is really outplaying the overwhelming favorite if both sides are even. I personally thought it was a crime that the Giants were somehow behind after the first half because it was obvious even then they were severely outplaying the Pats.

I especially love the way Aaron stays away from the INT that hit Steve Smith in the hands, having 12 men of the field to extend a Pats drive, the illegal bat penalty that canceled what looked like another scoring drive, and the Eli misfire to a WIDE open Plax that was more incredible than any of Brady's incompletions. You want to lament missed opportunities, you need to look at it for both teams, not just the one you're a fan of.

But I can't say I didn''t completely expect it, and instead of being upset this sort of rationalizing actually makes me feel even better :-)

No, I'm not quite ready to let this go just yet. Maybe after a month or six :-) But seriously, this needs to be a lesson to Aaron and everyone else running the site to not be so arrogant and dismissive of *any* team's chances in *any* playoff game after what's transpired not only this season, but in the last three. It's a lesson I hope they put to good use in next year's postseason.

Oh BTW in 8 quarters this season it was Pats 52, Giants 52. Just trying to destroy this silly notion that the Pats blew the Giants away the first game or were the vastly superior team of the two. And maybe, just maybe, launch a new and supplemental way of thinking that compares the two teams in a vacuum instead of measuring them by how they did against teams they won't be playing. For example, how the Giants did against the Vikings was completely meaningless because they weren't PLAYING the Vikes in the Super Bowl. So why was that info somehow considered more relevant than how they played against the Pats--the team they WERE playing Sunday?

Okay, *really* sorry to be hijacking the thread a bit here, but what I'm trying to say is that going into the Super Bowl, did it really matter how superior the Pats were to every other team in the league when, based on actual on-field results they weren't that superior to the Giants, if at all? They weren't playing all those other teams, after all.

Based on results purely between the 2007 Pats and 2007 Giants I'd say each team would take 5 games in a ten game series. I could care less how much the Pats would've whipped say, the Bills or how badly the Giants would've lost to, say, the Redskins. Because the last I checked, those teams weren't in the Super Bowl.

21:
Even when the Patriots lose the superbowl, their opponents whine. Very gracious and classy.

Nevertheless, congratulations to the Giants on a historical upset. A very exciting game, and an excellent game plan, well played.

As VOA (an unbiased statistical measure - there is no room for a "homer" to fake the stats here. Take the whines elsewhere, please!) shows, the Pats slightly outplayed the Giants, if you assume they had recovered the expected number of the fumbles, but were slightly outplayed considering how the fumbles actually turned out. That seems a perfect description of a game that came down to the last three drives, two of which featured desperation heave-ho passes into coverage. It does not get any closer than this.

Fumble recoveries are - in the larger picture of DVOA stats - treated as a non-repeatable skill, and rightly so. But in a particular game you can see the specific actions that lead to recoveries.

In the fumble-recovery-that-wasn't, it was a brilliant play that saved the ball for the Giants. First, Bradshaw had to get in contact with the ball under Woods quickly enough to leave a smidgen of doubt about Wood's control of the ball. Once he did so, the refs could not blow the play dead, but he still had to pull the ball away from Woods.

In another fumble-that-wasn't-recovered, it was a heads-up play to bat the ball forwards to where a Giant could recover it. It turned out to be a penalty - correctly called - but was still much better than the Patriots recovering. Quick thinking.

So, this was an exciting game that came down to which team won when the ball was up for grabs (including fumbles, the "great leap" catch, missed or made interceptions, etc.)

Well played, Giants. In a close game, you played best when it mattered most.

Congratulations.

#24: No "whining", just pointing out all the things that went wrong for the Giants. If Aaron can do that for his team then I should be allowed to do the same for the other team, no?

And I'm not pretending to be classy. Although I am being more civil than a lot of other people who share my point of view regarding the Pats.

Aaron and the boys need to learn a lesson from this and I think they will.

And again I must insist that if an overwhelming favorite is playing even with a heavy underdog, it really means the underdog is outplaying the favorite badly. They'd have to be to even have a chance to win, right?

Don't give me this revisionist "oh the Pats weren't that great the last 7 games" history--before the game, when people should have said this, it was the juggernaut Pats against the puny Giants. And this site took that angle and ran with it as much as anyone.

re: 11

Is a 6% differential in 3rd down percentage worth 9% VOA? If it is, Aaron, I think the formula is giving it WAY too much weight...

re: 12

I think you're on to something here. Aaron, have you given any thought to a tiered weight on turnovers? I'm not sure how much more you could adjust on fumbles (aside from noting 'reason' in charting), but the INT data is already on the game charting files. Perhaps a lower penalty for 'accidental' INTs (not quite the word I'm looking for, but you know what I mean) would improve the formula?

It makes sense: perhaps a team punished harshly by fluke plays suddenly has its luck evened out, to the point where it looks like a miraculous and inexplicable turnaround. The whole "fumble luck evens out" might extend to INTs too, provided you're talking about fluky ones.

Perhaps the problem I'm seeing with VOA is that, since we rarely see it, we can only go by what DVOA points out. I think perhaps the opponent adjustments are doing a very heavy lifting here, effectively compensating for VOA's limitations (which is, after all, their job) but clouding them at the same time.

Thanks, Sergio. I mean, the idea that some things are not zero sum is already in VOA-- for special teams, I believe. Unless I am mistaken, for example, Gostkowski's OOB kickoff counts against the Pats ST but not for the Giants ST. Why should it? The Giants did nothing there.

Well, why should the Pats get a defensive adjustment for a half of a fumble recovery on the Bradshaw fumble? Why should they even get credit for stopping the Giants on that play? They did nothing. It was purely a negative play for the offense, not a play for the defense.

Anyhow, DVOA is a wonderful way of distilling a lot of the distortions from games to getting to team quality, but it isn't perfect. After all, by the DVOA numbers, the team that gained more yards per play in the passing game (and had just 5 points lower in completion percentage) and more yards per play in the running game and who had a better combined third-and-fourth down conversion rate, who in addition to sacking the other quarterback more often also hit him significantly more frequently and more violently, who had the same number of forced fumbles, and despite making quite a few bone-head unforced errors was actually outplayed. And was actually outplayed on defense.

Nope.

Giants fans shouldn't be upset with the VOA (not DVOA-- DVOA showed the Giants significantly outplaying the Pats) numbers. Aaron should be upset with the DVOA numbers because in this case they show that while the system is good, it's still flawed.

"The Giants converted 8 of 17 on third down which is also pretty damn impressive."

The Giants converted 8 of 16, plus 1 of 1 on 4th down.

Just re-checked the play-by-play. Counting Watson as a conversion, Giants were 8-16 on third and 1-1 on 4th. The Patriots were 8-15 on third and, not counting Blackburn, 0-2 on 4th.

Another VOA oddity-- it is saying that the team that managed to get into the red zone twice (scoring both times) outplayed a team that managed to get into the red zone five times (scoring twice). All while turnovers were even and while the Pats supposedly had slightly better special teams.

That is more than a tad bit counter-intuitive.

Pardon me-- scoring thrice, two touchdowns and a figgie.

Bah. "red zone five times" Make that four times. Still twice as often. And it wasn't starting field position that did it. They had one starting on the 40 (thanks to the OOB kick) which pairs up nicely with the Pats' starting near midfield after Maroney's runback.

But the other three times the Giants made it to the Pats' red zone, they started on their own 23, 20, and 17.

The other time the Pats made the red zone, they started from near their own 20.

So two other times more than the Pats, the Giants took the ball from crap field position and move it into the red zone.

Yet the Pats defense outplayed the Giants defense, according to VOA.

Holy cow, I didn't even realize this before.

Do you realize that the Patriots only got inside the Giants 40 yard line a grand total of three times in the entire game?

What's great about this site is watching Schatz pervert numbers to satisfy his worldview. If he ever admitted being wrong, I probably would lose interest.

Aaron, consider this: Bellichik is never going to consider hiring you ( as Epstein did with Voros McKracken) unless you create value going foward, instead of rationalizing mistakes of the past. Your bias is marginalizing your future.

"And I think we all know that this is more about you crowing that any desire for FO to derive any sort of lesson from this."

Doug, take a step back and re-read my posts.

I don't think that is a fair interpretation of my posts. At all.

I'd be interested in seeing a breakdown of the numbers by down.... all game long I thought the Patriots were awful on first down, average on second and good on third, but their inability to get a good down and distance on the first downs that wore on them. Yes they converted two third and thirteens, but they got into third and thirteens in the first place.

Regarding fumble luck... there seems several categories of fumbles. There are the ones where the fumble bounces right to someone who grabs it and/or falls on it without much contesting. These seem to be pure luck based on the random bouncing of the ball.

The others are the kind where multiple people are there before anyone can secure the ball, and then it becomes a struggle in the pile-up. I don't know if these are as random, though it certainly does not seem a controlled scenario. I can't buy into the cliche that it is recovered by the team that "wants it more".

#9:

The weakness of DVOA/VOA is seen with the problem with Moss’s numbers. Moss would have been wide open on the third down pass if Brady had not underthrown the pass to a point where the defenders could catch up. Given how easily Moss burned the Giants’ D Week 17 (two plays in a row!), I was quite disappointed this option was not tested more by “genius” Josh McDaniels.

Moss was about 60 yards downfield at that point. Brady didn't underthrow it-- Moss just didn't get any separation until he was outside of his quarterback's range. The Giants' DBs played that pass well. Moss is always going to win a footrace eventually, but Webster and Wilson got enough of a headstart that he couldn't pass them before he was so far downfield that his QB couldn't reach him.

I'm sure that there were more plays called to test the Giants' deep coverage, but they ended with Brady either checking down or going down. The OL didn't hold up its end of the bargain.

My bad, Doug. Sorry for losing context in your reply.

Is Dolfan a Giants fan or Dolphin fan?

#40 - I think he's STEPHEN YANG!

#21: "the illegal bat penalty that canceled what looked like another scoring drive"

You're joking here, right? The illegal bat penalty prevented a turnover and kept the Giants' drive alive. Without the illegal bat penalty it's, at worst, Patriots' ball at the 30, instead of NY ball at the 39.

Aaron,

It's time you stepped up and stated exactly the difference in the meaning and use of DVOA and VOA. Otherwise, we have to put up with statements like "DVOA showed the Giants significantly outplaying the Pats". It's getting boring to explain again and again that DVOA doesn't mean that. VOA is the stat that compares two teams on the day they played, not DVOA.

Still, here goes.

DVOA is how well your team played, if and only if you assume the other team played at their season-long DVOA level. For example, if you assume the Patriots played at their usual high level, the Giants had a great game. If you assume the Giants were their usual average selves, the Patriots reverted to near-average play. So: both were great or both were average or both were somewhere in between - either way a close game.

VOA is how good your team's play-for-play results were, assuming your opponent was a typical, average opponent. It can also be interpreted as how well your team did compared to that opponent on that day. It makes no statement on how well you had to play to get those results.

The best use of these stats is to use VOA to compare two teams on the day they played, to use a range from VOA to DVOA to describe the overall quality of play in that game, and to use that range or the season-long or weighted DVOA to anticipate future results against other opponents, depending on whether you think your team has fundamentally changed in nature.

As for VOA with and without fumble adjustments, that can be used to show how much of the quality of play on loose balls factored into the VOA comparison. In this superbowl, it's the difference between the Patriots outplaying the Giants by 9 to 12% (not 21%, that would be double counting most plays) in VOA (assuming typical fumble recovery play), versus the Giants outplaying the Patriots by 3 to 5% in VOA.

It is fair and accurate, and a correct interpretation of DVOA and VOA, to say that the Giants made up most of the perceived gap in the teams through improved play in normal situations, and put it over the top by being the better team when the ball was loose.

Congrats to the Giants. It was sooooo close, decided in the final minute. In the end, you had the better game, and deservedly won.

Congrats to the Patriots for a great season. You were less than a minute away from 19-0. No team has come close to that. That's small consolation, but not a small accomplishment.

I love FO; it's my main source for football analysis. That said, I think my opinion on the extent to which it is appropriate to use numbers (traditional or sabermetric) over what we see is a little different from that of Aaron.

I think the world of DVOA and DPAR (I mean, don't you guys have the best record at predicting teams records?). But I think a majority of their value is as a predictive indicator. I just can't make myself care very much which team had better numbers in a given game, if I can't use that info to help me guess who will win the next one.

And I think the idea of "fumble luck" is the epitome of this. In my opinion, there is very little luck in fumbles. Somebody has to strip that ball, or drop that ball. Someone has to see it, and fall on it. I don't think that the Giants were especially "lucky" in their recovery of fumbles on Sunday (I am a patriots fan). I applaud FO for recognizing that fumble recovery is a NON-PREDICTIVE indicator, and that if the Patriots and Giants had a rematch tonight, that I should pick the Patriots. But citing fumble luck in a discussion of the last game of the season just seems kind of a waste of time. Sometimes I feel as though the numbers are treated as the end and not the means. An I just don't see the end citing "fumble luck" serves here.

I've loved quick reads all season. I just guess that when there is nothing left to predict it doesn't seem as useful.

Anyways, thank you guys for all your great work this season. And Aaron I respect you going out there and showing your numbers even when you know you're going to get taunted a lot. I've gone by your numbers pretty steadily all season... just seems like right now is when they carry the least weight.

On further reflection, I also wonder about the VOA special teams numbers.

Take away the drives that started because of a turnover and the last kneel down one. The average Giants starting field position was the 30 yard line (it works out the same if you count the turnover on downs after 4th and 13 or not). Meanwhile the Patriots average starting field position was the 22. That is pretty significant. If the Pats were outplaying the Giants both offensively and defensively and to a near-draw on special teams, how could that be?

There were three fumbles, all recovered by the Giants, but how much did they really help in the grand scheme of things? If the Patriots recovered Brady's fumble, it would have been second and seventeen on the NE 49 with only 22 seconds left and only one timeout. Possibly they could have gotten in position to score-- but I think even with the Patriots awesome offense it is not probable.

Had the Bradshaw-bat fumble been recovered by the Pats, what would the net result have been? Assuming the Pats would have fallen on it rather than running it back (and this is nothing but an assumption) they would have had the ball on their 29. Two plays later they got the ball on their 11-- a net difference of 18 yards. Meaningful, but not overwhelmingly different.

So yeah, the Giants may have had good fortune recovering 3 of 3 fumbles. But the only one that made a huge difference was the fumbled handoff between Manning and Bradshaw. That was huge, lucky and impactful since the Pats would have had a short field. But then, their INT was pretty huge, lucky and impactful.

One of the things that makes it impossible to know what "should" have happened is the sequentiality of it.

Yes, the Giants got lucky recovering the fumbles. Does that mean if they had not recovered the fumbles, they'd have lost? Of course not.

Perhaps they'd have made OTHER big plays to compensate for the fact that they were losing fumbles. Perhaps the Patriots would have made other BAD plays after getting overexcited about recovering fumbles.

There's just no way to know what SHOULD have happened if in some cosmic statistical plane, the Patriots outplayed the Giants or vice-versa.

What did happen is what happened. This piece, fanboyism or not, says something true about what DID happen: The fumbles recoveries helped the Giants win this game.

What would have happened if they hadn't recovered the fumbles, we'll never know. The statistics can't tell us, either. As the saying goes, that's why they play the games.

Otherwise, it's just a battle of spreadsheets, and who would watch that?

This game was Manning's season in a nutshell. Through three quarters, he was erratic and unable to sustain drives, completing just 50 percent of his passes with -3.7 DPAR. In the fourth quarter, he was the great quarterback he was throughout the playoffs, completing 65 percent of his passes with no turnovers for a 7.8 DPAR.

That's one way to say it. I'd present it this way: He was very sharp in the first quarter, not good at all for the next 30 minutes, then electric in the final period. And that opening period (albeit it's just one drive) was very critical - lots of third-down conversions, points, kept New England off the field, set a tone. When you're a heavy underdog, the beginning of a game has major importance attached to it.

i think the problem the entire year with the giants ratings is that, in general, they've done a very good job of not letting turnovers hurt them. It results in things like the fumbles dragging down the VOA. Throughout the season, Eli threw a lot of picks where the defense stepped up and didn't let up points. The resulting ugly wins made a bad game for VOA. You can pick points where both teams made mistakes or had a bit of luck, but i think the general idea is the game was razor close (like a SB should be) and the giants just made the plays when they needed to.

Unfortunatly, there is a bit of sentiment where the giants mistakes were forced by the pats, but the pats mistakes were their own. The game was close, and you have to give the giants their due for shutting down such a historically prolific offense and making the plays they needed. I doubt it's bias, the current formula is what it is, but i don't think it's given a wholly accurate description of the giants most of the season. Or, the ugly way they won is just an aberration that's hard to sum up in numbers.

I'm so glad the Super Bowl is over. Frankly, the incredibly petty level of discourse has really hurt my level of interest in this site. Not from the writers, whom I think are basically the same as always - the commenters have simply been obnoxious jerks. If I never see the word "bias" again in someone's post, I'll be a happy man. A part of me hopes that FO just starts banning people next year.

On the game: Interesting that VOA thinks the Patriots outplayed the Giants (slightly). I agree with some of the others who suspect that it can't accurately measure the degree of non-sack pressure that Brady was receiving. He got sacked a number of times, but was heavily pressured on nearly every throw. That credit should probably go to the Giants defense, but it's impossible to pick up from the play-by-play account.

Apropos of nothing, did the NFL Network ever do its "Four months ago" ad shtick this year? And if not, wtf?!

Fumble Luck?

On the muffed handoff, it was not luck that Bradshaw got his arm in there and ripped the ball back either. He's a much tougher man than those he opposed. That's just one reason why the Giants are better than the massaged numbers can describe.

42: The patriots were not going to get that ball either. And that was, again, Bradshaw beating the defenders at controlling a loose ball.

Bradshaw is rookie who has dropped a few. He was a very high risk of doing it the SB. I'm sure he was well coached/prepared to both minimize the risk of fumbling and, if it happened, to act aggressively to prevent the opposition from getting it. Call me crazy.

Apropos of nothing, did the NFL Network ever do its “Four months ago” ad shtick this year?

That spot played Sunday. It wasn't as clever as usual. They had a scene dissing Eli (punctuated by the speaker doing a giant nasal snort), a scene dissing Moss (uttered by a chubby speaker who was wearing clothes much too small for him), and a wedding groom talking about the Bears going back to the Super Bowl. The first two scenes were curiously constructed; it's almost as if the league were making fun of its fans.

The Giants outplayed the patriots. The patriots had an 'off' game because brady got hit 23 times, 5 of them sacks. He had the ball ripped out of his hand on one occasion. He was confused and had happy feet after the first quarter. The only drive he looked good on was his final touchdown drive.

The 'Best Offense in the NFL Ever' was held to 14 points. So kind of you to remove the fumble from Eli Aaron, now how about the INT? How do Eli's numbers look without the interception that was entirely Steve Smiths fault? Count it as incomplete, I'm interested if Brady still outplayed Manning.

I could agree the Patriots offense as a whole outplayed the giants offense. However the Giants defense made up for that.

Pats fans will blame this loss on Brady's ankle, 'poor luck', anything but the simple fact that Brady got rattled. He could have won that game with any one of his errant throws. Fortunately, he was listening for Tucks footsteps instead of focusing on the play.

Their first game was decided by three points and so was the second. Seems to me that by the end of the year the Giants, with good Eli, and Pats were equals. Maybe the Pats older LBs and Harrison can't handle an entire season, who can say.

The best thing (for me) is the Giants are a young team with a solid offensive line and a dominant defensive line. Our new GM produced a draft class including Aaron Ross, Steve Smith, Kevin Boss, and Ahmed Bradshaw. We're only going to get better, assuming Manning remains an above average QB.

The real luck that mattered in this game came on the Manning-Tyree play. While there was undoubted skill involved, Manning was lucky to break free, lucky that when he spun out he did to where there was a clear area, and Tyree was lucky to hold on to it while falling backwards and losing grip of it with one hand.

Skillful? Sure. But also damned lucky. They redo that play, including having three defensive linemen out of four obliterating four out of five offensive linemen, and having decent coverage on the top wideouts for the Giants, 1000 times they might get the completion once.

And they got it.

BillB: The concept of fumble luck is that nobody has control over where a fumble goes - the ball bounces in fairly random directions, comes out at fairly random times that determine which mob it's in, etcetera.

Ahmad Bradshaw made great plays really on both fumbles - yanking the ball out from underneath a man nearly twice his size and slapping away the one he couldn't get - but just like a field goal returned for a touchdown, they were great plays with a huge luck aspect attatched by VOA, and rightly so as it isn't predictive of future events.

Overall I think the VOA/DVOA numbers seem to correspond with my subjective impressions of the game. My question might seem a little arcane, and goes back a few years to 1999-2000. I seem to recall the StL-Tenn Super Bowl as a tightly fought defensive battle with a fairly low score, especially considering how explosive the Rams offense was that season (IMHO those Rams, the 1998 Vikings, the Colts of the last few seasons, and the current version of the Patriots are clearly the best offenses of the 16-game-per-season-don't-let-the-defense-even-breathe-on-the-wide-receivers era.). So I wonder how is it possible that Tennessee's defensive DVOA for that game could possibly be +14%??? For that matter, Tennessee had a pretty effective offense that season too. How could the Rams' defensive DVOA be +19%? I know time fades memories, but those numbers don't at all jibe with how I remember that game.

54: For the record, I, as a Pats fan put the blame on the Offensive line which forgot how to block, and on Ashante Samuels for dropping a game-winning Int.

While I will agree that, on average, fumble recoveries are random and interceptions are the result of skill, the actual interception in the game was as lucky and not skill-based as any fumble recovery. This is one reason why I agree with Yaguar in #13. VOA is designed to be very accurate on average over the long term. To do that it relies on adjustments that, while they improve accuracy in the aggregate, may distort the results in any one game.

Way to stay above the fray, FO. Your stats are obviously not biased. Commentary will always be biased because that is human nature. This is not because of some dark conspiracy by FO, but rather the fact that everyone is biased. Even when we try hard not to be, which I am sure FO does, there will always be some bias in everyone's commentary.

FO's defensive rants about this, while stating the truth, just makes them look bad. You guys are professionals who will undoubtedly get insulted and accused of bias sometimes. I come to this site for the great analysis of football, both by the numbers and the commentators. Seeing this site sink to Aaron and the FO guys having a month long argument with the posters has greatly diminished my opinion of FO.

Matt,

So nobody has control over which way the ball bounces. But, as you saw, players can impact where it winds up.

To FO editors

You could have mentioned VOAf (or whatever) at any time during the season. But
playing the "luck" card after one of the greatest SB ever is not great judgement. Its like Bellichik leaving before the game is over. Getting defensive isn't going to help. You make editorial decisions. Live up to them.

Another point on VOA and non-predictive events:

The Randle El to Ward touchdown in XL is not the type of play that's rated highly in VOA because it's not very predictive or repeatable.

But in a one-game sample, you don't care. Do you think the Steelers gave a damn that it's not a very repeatable event? They needed that play to work once, and it did. Maybe winning on a end around deep pass isn't correlated with winning other games, but the Steelers didn't have any more games to play.

You can't fault teams for winning in a non-VOA-friendly way on a single game sample size.

Aaron,

Thats it! You are really onto something. I briefly reran the VAOf/DVOA numbers through a million iterated 60 minute intervals. I then added the VOAkp (VOA karmic payback) of 0.983 and 1.023 for NE, NYG respectively, and I got:

Giants 17 Pats 14!

Another example why the "luck" adjustment in VOA is not necessarily accurate for any particular game is the fumble recovery before the half. That probably has a significant impact on the VOA but it had little impact on the actual game. Had NE recovered the fumble, it would've had the ball on its own 49 with 10 seconds left and no timeouts. It's possible that NE could've gained 15-20 years on the next play AND stopped the clock, but it's unlikely. On average fumble recoveries at mid-field will tend to have a large impact on the game. Just not in this particular circumstance.

Umm, 15-20 yards, not years.

"Patriots fans will blame this loss on..." yadda yadda yadda.

Practically every Patriots fan post I've read (I haven't read them all, but I've read a fair number) has Patriots fans blaming the loss on the Giants "Four Aces" totally destroying the vaunted Patriots O-line. Please don't attribute things to fans that they didn't say.

The only people that seem to be whining are Giants fans. Which I don't get... your team won...

I'm surprised Toomer's dpar is higher than Welker's. For my money, Welker was the best player on the field.

Here is the thing. The point of statistics is to have a wide window - to try and draw useful conclusions from an enormous amount of information, more football games than a single person could every watch. But by its very nature, that information is not going to be as detailed as that that can be extracted by a person who looks closely at single games, with a narrow window.
I watched every giants game this year, and not a lot of other games.They were a frustrating team that shot themselves in the foot with untimely penalties, drops, blown coverages, unlucky or stupid turnovers. (In other words, classic giants football.) They weren't physically beaten often (except in the secondary). You got the feeling they could be good, but the longer they kept screwing up, the less likely that seemed.
And Tom Coughlin, mister minimize-the-penalties-and-turnovers, focus-on-execution, turned out to be right. When they did that, and gave the effort of their lives, they won the superbowl.

There is no way the kinds of broad statistics that can be extracted from the play-by-play can tell the difference between a bad team and a team playing badly. I had my doubts as to which the giants were. I am very happy I was wrong.

Some of these posts are cracking me up.

To make these people happy I propose a new statistic:

BVOA

Bias-Adjusted Value over average.

It adjusts the VOA for each game based on the :roll: obvious bias of the FO staff. It further adjusts on individual play for players known to be FO faves being involved in the play.

Note: I'm a Pats fan.

Despite that, I'm giving the Giants all the props in the world. The Patriots were, going into the game, by far the better team, at least up to that point, and yet the Giants had a better gameplan, made better in-game adjustments, played with more intensity, for the most part executed better, and hence leveled the playing field. As Aaron said, the two teams played essentially to a tie--a point where a bounce one way or a break the other would have given either team victory. And both teams got a number of bounces and breaks. The Giants got one more, or got theirs slightly better timed, and won. Congratulations, and I'm happy for all the Giants fans, and the Giants players who deserve a ring. I'm reasonably mellow about the Patriots loss--they had a fantastic run this season, and came within a hair's breadth of winning out despite getting outplayed in the biggest game of the year. Karmic payback for some previous years.

But one thing I don't understand, and will not accept, is the notion that Eli Manning played well in that game and deserves the MVP. Yes, he wasn't horrible. He had some good drives, and put enough points up to win. Yes, the one INT wasn't really his fault.

But he was erratic and unable to score on a defense that has been notoriously soft against the pass all year. Through about 50 minutes, he had led his team to a grand total of 3 points, despite good pass protection, amazing play by his defense, and a near consistent field position advantage from his special teams.

Then, in his final two drives, he led two TD drives. But on the first (I think...memory is a little hazy) he made a horrible decision and tried to force a ball, a ball that went right through Brandon Merriweather's hands of stone--a more sure handed safety catches that and the Giants never take the lead.

Then, in the second, when they were desparate to get down the field but still had plenty of time, he heaves an ill advised pass over the head of his receiver right into the hands of Asante Samuel--who drops it. Manning almost never got a chance for his heroic sack escape and Tyree's heroic throw, because a criminally stupid, or possibly inept, Grossman-like throw almost--even should have--cost the Giants the game right there, despite the teriffic play of almost every other player.

The fact that Asante Samuel couldn't quite hold on there, as Merriweather couldn't earlier, doesn't change the risk that Eli ran right there. Had either of them held on, then we wouldn't be talking about Eli leading a game-winning drive because he was clutch--we'd be talking about how the old, interceptiion-prone Eli cost the Giants the game right at the end. All because of a random bounce of Samuel's usually sure hands.

The MVP should have gone to Tyree, or possibly to the entire Giants D-line (if that's allowed) or to Tuck or Osi if you have to pick one person. But not to Eli. He was a decent QB in that game, but not a great one.

MJK-- the problem was that there was no obvious MVP. Tyree had the awesome one catch, but that awesome one catch was only half of a miraculous play; Manning's half of the play was pretty special as well, and he did more the rest of the game than Tyree did.

The Giants defensive line was dominant, but that's several players. Did any one of them really stand out over the others? Tuck had the best first half, but was he really that much better over the course of the game than Osi? Than Strahan? Hell, Alford made plays. Cofield made plays. As a unit they were balls to the wall.

Take the voters and have those who would be inclined to give it to a Giants D-line split their votes among the choices, and the result will be to give it to the Giants offensive player who did the most. Eli probably fit that bill.

(And since the secondary didn't get torched, and the LBers didn't get torched, it is odd to see a neutral metric like VOA come out suggesting the Pats D played better, which is a side issue. I think that it was pretty obvious to most people's eyes that this was not the case, but figuring out how to improve the FO metrics to be more accurate over the course of an entire season using the inputs available (mainly gamebooks) is not easy. I think it is probably sufficient to say this is one case where reality was different than the numbers.)

And I should have added...

"The fact that Asante Samuel couldn’t quite hold on there"

That greatly overstates the chance Samuel had on that ball. Even if he had been able to jump high enough from where he was on the field to get his hands on the pass better, it would have been virtually impossible (and I say virtually only because pro athletes sometimes pull things out that seem to be impossible) for him to get his feet down in bounds. Had he managed to make that pick, it would have been one of the best interceptions in Super Bowl history, as impossible as Tyree's catch.

Okay, I don't get something.... Giants fans are upset that no one gave their team any respect for being capable of winning the big game. Giants fans are upset that this site's stats didn't back up the superiority of their team.

Huh? Doesn't that make the victory all the sweeter? Does beating the absolute best by your team's playing the game of its life make it incredibly sweet?

It sure did for m