Saturday Division Round Analysis

Here are some quick reactions to Saturday's division round games, with at least one statistical morsel from each game. Click the game header to see the graph and advanced box score.

NO at SF

This was the most exciting of the weekend's games, with a 7.6 Excitement Index. SF had an incredible 1st quarter, taking a 17-0 lead. They clung to those 17 points like life depended on it until Akers hit a 3rd quarter FG to make the game 20-14. Things didn't look so good for SF with 4 min left in the game. Yes, they were up by 6 against NO. But, they were only up by 6 against NO. That's when the explosion occurred. All football hell broke loose. There was one touchdown per minute in the final 4 minutes of the game, 2 per team.

Vernon Davis will be immortalized with his amazing catch on the goal line for the win, but he racked up huge chunks of WPA well before that moment. His 47-yard catch to get the NO 20 was a 0.45 WPA play, and his 37-yarder on the prior TD drive was a 0.39 WPA play. If I'm game planning for the Giants this week, I'm going to take away Davis.

Vernon Davis finished with an amazing 1.05 WPA and 13.6 EPA. Alex Smith shares the credit with a 0.98 WPA. This was Smith's career game to date. His next best game was a 34-20 game at Oakland in 2006, notching 0.46 WPA on 10.8 EPA. That looked like an odd game as a career best anyway. He only had one attempt beyond 15 yds that day, and totaled only 165 yds with 1 interception. He was simply methodically efficient against that season's 2nd worst defense in the league.

It was a high-drama end-game, to be sure, but it never needed to happen had Alex Smith slid before crossing the goal line. With 2:18 to play, SF was down by 1. Smith ran a naked bootleg 28 yards for the go-ahead TD. Instead of scoring the TD, had Smith taken a knee SF would have been able to run the clock down under 40 seconds before a chip shot field goal attempt. NO would have had a little over 30 sec and no timeouts to respond with their own FG. That might not be a 100% lock, but it's a better bet than leaving 2:18 plus a timeout to get a game-clinching TD.

NO didn't have that luxury. Being down by more than 3, they had to score the TD. But they scored too fast. NO took only 23 seconds to score the go-ahead TD, leaving 1:37 on the clock for SF. During games, I'm often asked, "Team X just scored a 30-yard touchdown, so why did their WP dropp according to your stupid model? It's obviously broken." No, it's not broken. It's that sometimes teams are better off burning clock on their way to a score than scoring instantly.

After scoring the TD, the 49ers smartly went for 2 to make it a 7 point game. A 7 point lead at that point is worth 0.86 WP, and a 5 point lead, should they fail, is worth 0.78 WP. Given a 45% conversion rate, that's a total WP of 0.82 WP for the TD. Running the clock down to 30 sec with a 2-point lead is worth a 0.91 WP--a much better bet. But hey, then we wouldn't have seen one of the most exciting endings to a playoff game in a long time.

Tebow at NE

The game was well over by halftime. Brady racked up 21.8 EPA in the 45-10 victory. It was apparent Brady had something to prove, and he was fighting for every touchdown well after the game was won, at least until that quick-kick in the 4th quarter.

Tebow's numbers were very poor: -5 EPA, 35% completion, 31% Success Rate, 5 sacks. His WPA was +0.01 however, thanks to the fact his worst plays of the night came well after the outcome was already decided.

NE's offensive line was stellar as usual, posting a  0.28 WPA while allowing zero sacks, zero sack yds, zero QB hits, and just 1 tackle for a loss. They had a 47% run SR on top of all that. How do you beat NE? You have to beat their o-line. In their 3 losses this season, they gave up an average of 2 sacks, 13 sack yds, and 4 QB Hits, and 2 tackles for losses. That's not a lot, but it's good enough to make Brady uncomfortable.

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

37 Responses to “Saturday Division Round Analysis”

  1. Anonymous says:

    "How do you beat NE? You have to beat their o-line. In their 3 losses this season, they gave up an average of 2 sacks, 13 sack yds, and 4 QB Hits, and 2 tackles for losses. That's not a lot, but it's good enough to make Brady uncomfortable."

    Okay, but can you also tell us their average O-line stats for wins? Otherwise this is just a glorified Bill Simmons level analysis.

  2. Anonymous says:

    "How do you beat NE? You have to beat their o-line. In their 3 losses this season, they gave up an average of 2 sacks, 13 sack yds, and 4 QB Hits, and 2 tackles for losses. That's not a lot, but it's good enough to make Brady uncomfortable."

    I'm curious to know the mean number of rushers the O-Line faces on passing plays. I suspect teams leave a lot of people in coverage against all those NE receiving threats and bring fewer attackers after Brady, meaning the receivers and tight ends are responsible for a lot of the success of the O-Line.

  3. Anonymous says:

    Brian--the comments re SF-NO call out for a deeper WP analysis that I'm sure is coming. An SF field goal would have only given them a 2-point lead, and I'd be surprised if NO has a bad 2-min offense.

  4. Mike M says:

    Pats play on the field was no-where near as good as their 45-10 win would suggest.

    A deceptive final score does not bode well for the Pats next week, far more often than not, teams can not out-perform their stats in back-to-back weeks in the postseason.

  5. Brian Burke says:

    At least it's "glorified" Simmons-like analysis. Point taken.

    No additional analysis needed on the SF-NO situation. Simply put, the numbers say that a 2-point lead with 30 sec to play (and no timeouts) is safer than a 5/7-point lead with 2:00+ (and 1 timeout).

  6. Boston Chris says:

    Some solid analysis from Mike M. Wasn't he the guy who said there's no way the Pats have a better chance of winning the Super Bowl than Green Bay or New Orleans? Good point...that 45-10 score is really deceptive. Probably should have lost that game in fact. C'mon, you joker.

  7. Boston Chris says:

    Went ant found one of his idiotic comments...

    Mike M, "The Pats have the highest probability to win the SB with one-of the worst defenses imaginable ?

    I think it's time to go back to the drawing board with the model !"

  8. Anonymous says:

    Brian, I also think that the SF/NO go down at the 1 or score the TD needs more analysis. While I am sure that you are correct in stating that he should have gone down, I suspect that changing things slightly would alter the decision (like if NO had 2 TO's, I presume that he should score the TD).

    In short, while the "obvious" don't score/let them score situations are well understood, this sort of situation isn't. I for one would admit to thinking that he should score the TD in this situation.

    Also, I would love to know how close are the numbers for Jimmy Graham to go down at the 1 with NO trailing by 5? This must be somewhat close as well, no?

  9. Anonymous says:

    Brian, Did you see the comment that
    NO's coach Payton, expected SF to go for the field goal and over time so he planned his defense to prevent SF from getting into FG range. SF's coach decided to go for the win.
    Great idea by SF and bad idea for NO.
    Almost as bad as Pittsburgh's defensive planning against Denver.

  10. Brian Burke says:

    Cross-comment from Keith's Alex Smith post regarding whether NO should have also taken a knee at the 1:

    I don't think NO could afford to be that picky. They trailed by more than a FG. The risk is too big for a few reasons.

    1. SF allowed 1 rushing TD all year.

    2. SF knows you're running, because there would be no other reason to employ that strategy. Incomplete passes stop the clock.

    3. If NO gets its TD on 1st or 2nd down, the strategy doesn't buy you anything. SF still gets the ball back only needing a FG to win. So that's a lot of risk with the possibility of no payoff.

    The difference between NO's and SF's situation is that a FG is 99+% from that range, so SF can kneel 3 times and then kick. They would be guaranteed to burn all the clock they could, and at the same time be virtually guaranteed of making the FG. NO had neither guarantee.

  11. Rob F says:

    The analysis of those situations are interesting, but there is no way you could expect Smith or Graham to take a knee there. It is not reasonable to expect every player to know the win percentages for every situation late in the game. These guys are out there trying their hardest to gain yards, get first downs, and ultimately get in the end zone. You can't expect Smith to be evading the rush, scrmabling down the sideline, and thinking, "OK, down 1 with over 2 minutes to go and opponent has 1 TO, I should take a knee, but they have 2 TO's, so I better score!"

    If it's an obvious situtation like the Brian Westbrook play a few years ago (team was ahead with the ball and just needed the first down to run the clock out), then yes, let's make sure players are aware of the game situation. But if you're behind late in the game and you can get in the endzone, instead of thinking about number of seconds and TO's left, just get in the endzone.

  12. Rob F says:

    I see Brian's post now after I made mine. Thanks for clarifying the differences there.

    I still think it's tough to expect Smith to be thinking about taking a knee in his situation. Sure, it might have been the right thing to do, but there is no way he or anyone else was considering the possibility that he'd be running it in for a TD from the 28. So again, player heading toward the goalline to take the lead with 2 minutes left, I think you have to have the attitude of score a TD there.

    If we're just doing this as a hypothetical exercise, then fine, I get the point.

  13. Brian Burke says:

    Yes, you can expect players to do exactly that. There are prior examples of players doing it. I don't expect players to consciously be aware of situations like this in the heat of battle, but I do expect the coaches in the press box to be aware, and to relay it to the field.

  14. Rob F says:

    I know what you're saying, but when a team is in the heat of the battle, as you accurately described it, unless there is a timeout just one or two plays prior to the play in question, there is probably not going to be enough time to analyze the in-game situation and come up with a play call, and then relay specific instructions to players on whether or not to score a TD, if they happen to get the chance. I mean, really, what were the chances the Smith was going to be running one in from the 28 there? 1 in 500? Sure, after the fact, we can cherry pick a spot in a game where, if we knew before the play what was going to transpire, we could tell Smith to take a knee at the 1 without going out of bounds (which, by the way, would have been pretty hard while running full speed down the sideline while being chased). You could also say something like, "anyone who is inside the 10, about to score, take a knee before scoring," so that way everyone is aware, but still, that requires time to analyze the down, distance, time, and TO's left, in addition to getting a play called to players who are actively thinking more about formations, routes, blocking assignments, etc. I don't know if that's a realistic thing to accomplish in 40 seconds, over and over again throughout the game, even with a computer and all the data right in front of you.

    I really don't know, so I ask this sincerely: have you, or anyone else, tried to do this, in real time? In other words, have you watched a game, input all the appropriate factors, and decide, before the snap, what the right thing to do would be, in a variety of possible play outcomes? I think it sounds like a pretty hard thing to accomplish in 40 seconds, over and over.

  15. Rob F says:

    I guess another way to put it would be, the default play for any player is to score a TD if able. A player would have to be absolutely certain that scoring a TD is not the best play if you expect him to take a knee.

    That first knee (Smith at the 1) would be hard to determine ahead of time. The second, third, and fourth knees (after getting the first, down, he takes a knee while burning clock to set up a 4th down FG) would be easy to determine. Those are extreme examples, obviously, but you can see how after taking the first knee, it is a lot easier to analyze the data quickly enough to make the right call.

  16. Anonymous says:

    Had Smith downed it at the one and Akers missed the subsequent FG, there might still be a site or two like this claiming it was smart, but unlikely he ever plays another down of organized football.

  17. Brian Burke says:

    Maybe. But I do know it's been done before, so we know it's certainly possible. I can think of three occasions, the most notable being Maurice Jone-Drew in November 2009. It happened the same day as Belichick's infamous 4th and 2, so it got lost in the tsunami of media coverage. But you're right--they had time to think, and were already well within FG range.

  18. Anonymous says:

    Your point that Alex Smith should have taken a knee is interesting, but shows how different statisticians are from football players. To suggest that a football player sprinting toward the end zone and the go-ahead touchdown should take a knee shows very little understanding of what it is actually like - in the real world - to play the game. Also, it was a play designed to get a first down. It's not as if the 49ers were planning on that play getting into the end zone. In addition, the 49ers defense had been dominant for most of the game, so I don't think it makes sense to take a knee given that fact.

  19. Anonymous says:

    "In addition, the 49ers defense had been dominant for most of the game, so I don't think it makes sense to take a knee given that fact"

    Umm.. no matter how dominant the 49ers defense was playing, taking a knee in that situation would have been virtually a guaranteed win. In fact, the Saints offense could have been swapped with a Division 3 college team the next drive and taking the knee still would have been the better move

  20. Anonymous says:

    Brian, I disagree with your post on NO going down at the 1.

    I will start with your 3rd point, that NO gains "nothing" if they score on 1st or second down. This is, IMHO, clearly false, as even forcing SF to burn their final TO improves NO's probability of winning; for example, if SF didn't have the TO, they would have lost about 10-15 more seconds running a spike during the final drive, which means that they would have kicking a FG instead of running a play to score a TD.

    So I think it is clear that, assuming that NO scores a TD, that it is better for NO to take a knee at the 1 and then score the TD no matter what down they score on or what type of plays they run. It would be even better to waste more time by running on 1st and 2nd down without scoring and then going for the TD on 3rd and 4th down (SF would have had less than 20 seconds left in these scenarios).

    Of course, the problem is that you are taking 100% chance of a TD and reducing it by a pretty large amount, especially if you then try to burn time on 1st down or 2nd down.

    So the real question is do they gain more WP by making SF burn there TO + the time it takes them to score than they lose from the chance of them not scoring the TD? Your response doesn't address this question, it merely hand waves this issue (though I wouldn't be surprised if your numbers are correct).

    So, once again, I make a request for some more detailed analysis on end game "when not to score/when to let your opponent score" questions.

  21. Anonymous says:

    This is the problem with a purely quantitative analysis of a situation that has significant qualitative variables. For Alex to have taken a knee on that particular play would have been problematic for several reasons: (1) Harbaugh would have excoriated him for that type of unconventional move; (2) it would have been perceived as a betrayal by his team mates; and (3) it goes against the ethos of playing football and scoring points. Also, what football player is going to take a knee as the end zone comes into view after two perfectly executed blocks? Do you have any idea how hard it is to stop in the middle of a full sprint?

  22. Anonymous says:

    Anon above - I was thinking the same thing, but then I remembered SF didn't even use their last timeout on the final drive. They used it with 3 seconds left when Brees was attempting a hail mary to reset their defense.

    That being said, the outcome shouldn't change the hypothetical probabilities.

    It comes down to an option between two choices. Would you rather have:

    a. A guaranteed touchdown, with the opponents having 1:35 and a timeout to respond


    b. 1st and 10 on the opponents 1, with about 1:35 and the opponent with 0 timeouts.

    Again, its funny that in the real game the timeout was meaningless because SF didn't use it, but still, you can't take that outcome bias into consideration.

  23. Anonymous says:

    My post above was referring to Anon 2 posts above (the NO / Graham final drive scenario, not the obvious SF / Alex smith scenario), someone responded while I was typing my comment.

  24. Anonymous says:

    Also should be "1st and goal on the opponents 1", not "1st and 10 on the opponents 1".

  25. Brian Burke says:

    #1 I'm not a statistician. I'm an ex-fighter pilot. I'm well aware of what's possible in heat-of-the-battle situations. Spare me the lecture.

    #2 I'm not calling on Alex Smith to take a knee at the one unilaterally without any coaching being involved. I'm calling for the SF coaching staff to be aware of considerations like this and instruct their players if they feel it's to their advantage. The comment about this being "a betrayal" of the team misses the point entirely.

    #3 SF was already well inside "FG range" (oops, I meant "FG attempt range"), so it's entirely conceivable they would have a gameplan for chewing time off the clock.

    #4 What moron would expect someone to stop in a full sprint? Don't QBs slide all the time?

    #5 Lastly, I'll say it again: It's been done before and it has worked. So spare us all the sanctimonious 'ethos of football' nonsense. The ethos of football is winning, and in so doing, making decisions that give your team the best chance of winning. Ethos my a$$.

  26. Anonymous says:

    Brian, only thing about the other players going down is that in all cases the team could run the clock down to zero; this situation was different in that SF could have kicked the FG and still lost. In short, my guess is that the SF would want him to score the TD in this situation (even though that want is wrong).

  27. Boston Chris says:

    In response to the comment way above about trying to do it in real time...I was...I was yelling at Alex Smith to go down almost immediately after he got the first down and it was clear the path to the end zone was wide open. Of course....he didn't seem to hear me...

  28. Rob F says:

    That's great, but did you KNOW ahead of time that you were actually correct, or did you just think that was the right play? What we're saying here is that we're trying to maximize win potential on every situation, but in order to do that, you have to KNOW for sure that what you were suggesting was correct.

    Would you have been thinking the same thing if NO had all 3 TO's left? How about just 2? Or none? What if there was 3 minutes left instead of 2? 3:30? 4? 8? Where are the cutoffs for when sliding inbounds is the right move instead of actually scoring the TD? This is what I mean when I say it's going to be hard to KNOW these things presnap and convey them to the players on field in a timely fashion when they're under the constraints of the play clock.

  29. MG says:

    What if Alex Smith had indeed taken a knee at the 1, and then cast an absentee ballot for Mitt Romney in South Carolina?

  30. Anonymous says:

    Great post! Seems the suggestion that Alex Smith take a knee at the 1 yard line has clearly hit a nerve.

  31. Anonymous says:

    Regarding Alex Smith:

    Brian, your contributor said that he essentially "ditched his Trent Dilfer label" in the game.

    But isn't the actual truth that if Harbaugh coaches properly and has Smith take that knee, the Niners win 26-24 (almost surely), and Smith has his typical average game (33 passes, 165 yards, one key run, essentially 10/14 points scored due to anything he did). He had a very nice first TD drive, then had a 3 yard TD drive, two 3 and outs after turnovers that resulted in field goals to reach 20 points, then a drive that was one huge run by Gore followed by a 3 and out for a FG to make it 23 points.

    Finally he made a great pass to Davis, and that drive should have resulted in the game winning field goal, let's call that 14 points because it's really a TD drive but he takes the knee on the 1 yard line.

    I just don't see how that's good, when you are playing against the Saints' defense.

  32. Anonymous says:

    Sorry, in the previous post I forgot the pass to Davis on the previous drive, so maybe it's 34 or 35 passes for 200 or 210 yards. The point is that he gained 80 passing yards on the last drive that really shouldn't have happened if Harbaugh plays it right.

  33. Anonymous says:

    Brian--just wanted to see the WPA difference between the two game states, that's all, should have been more specific in the original comment.

  34. Boston Chris says:

    Rob F, why would I need to "KNOW". That's absurd. Should a QB not throw a pass unless he KNOWS it won't be intercepted? I had a gut feeling based on the training I've received (reading this site), which because it has been good training, turned out to be right. There's no reason that players can't be subjected to training for end of game situations so that their gut instincts are more likely to be right than wrong. I think its completely reasonable to place the blame on SF coaches, not for saying something before that specific play, but for not preparing A. Smith before this game for a situation like that.

  35. Rob F says:

    Yeah, so you had a gut feeling...just like Smith had a gut feeling that scoring the TD was the correct thing to do. If the point of these analyses is to determine which is actually correct, we have to go on more than gut feeling. Because invariably, our guts are going to be wrong some of the time.

    Again, what would your gut have said in all of the other what-if scenarios I listed? There are a lot of variables to take into account. I know I'd be a lot more frustrated if a player took a knee erroneously than if he scored a TD erroneously. So all I was saying was that in the heat of the battle, the default play has to be to score the TD. Only under predetermined circumstances should players be thinking about taking a knee instead of scoring a TD. Losing and late in the game, driving, clock ticking, formations to set up, defenses to read, tacklers to evade, etc., I just can't blame a player for scoring a TD.

  36. Rob F says:

    To use an analogy to another sport, scoring the TD here is like taking a shot on goal in hockey. It may not be the absolute best possible play, but it's never a bad play.

  37. Anonymous says:


    No additional analysis needed on the SF-NO situation. Simply put, the numbers say that a 2-point lead with 30 sec to play (and no timeouts) is safer than a 5/7-point lead with 2:00+ (and 1 timeout).

    Do those "numbers" reflect who was playing in that game? In other words, are they based upon a statistical analysis of the New Orleans offense versus the San Francisco defense, or an analysis of compiled data for all teams?

Leave a Reply

Note: Only a member of this blog may post a comment.