Wildcard Game Analysis

NO-SEA

Everyone is going to remember the incredible Marshawn Lynch TD run, but the play that tilted the game outcome most was thanks to the Seahawks defense. With 9:59 to play, down by a TD, NO threw the ball on 3rd and goal from the 3. The SEA defense stopped Devery Henderson on the 1-yard line, forcing a NO FG. That play dropped the Saints' WP from 0.36 to 0.23, the biggest swing of the 4th quarter.

It seems to be the biggest reasons for the upset were some inventive plays run by SEA. The fake block, fall down, spit out left wide open for the TD play by John Carlson was clever. But more importantly, all game long the SEA offense invited the NO defense to jump short routes then double-move deep. The NO defense gambled and lost. They were heavy favorites and should have been doing everything except gambling. Low variance is the friend of the favorite.

The NO decision to go for the conversion on 4th and less than 1 on their own 37 late in the 3rd quarter was smart. By this point they were no longer the favorite and needed to take some risks, but it's probably a good decision in almost any situation. It's the percentage play for the typical drive, but it's especially smart when down by two TDs well into the 2nd half.

A lot of people are shocked by this upset, a 7-9 division winner defeating the reigning champs. But the efficiency prediction model saw this as a much closer game (59% to 41%) than most (10.5 point favorites, equivalent to about 85% to 15%), but not because it thought Seattle was any good. It's actually because NO has been ranked ranked near 20th in the league for most of the season.

NYJ-IND

Why didn't Rex Ryan call a timeout with 1:42 left in the 4th quarter? The Colts had just completed a 12-yard pass for a 1st down to the NYJ 36-yard line. The Jets had all 3 TOs left. The next snap was with 1:10 remaining. At wost for the Jets, IND was in range for a 53-yard FG attempt, so the Jets should have been thinking about getting the ball back with as much time as they could. Huge mistake.

Some critics don't like the call to pass the ball on 3rd down. But IND was not content with a 50-yard FG attempt. A completion and conversion would have made the FG much easier and taken more time off the clock. It was a gamble, yes, but probably a good one with the ball in Manning's hands.

Caldwell is getting criticism for his timeout with 29 sec left. The Jets had a 2nd and 8 at the IND 32 and had just run for a 2-yard gain. The clock was running. Perhaps Caldwell was thinking it was possible to stop the drive there, and if the Jets made a FG, have enough seconds for a desperation FG drive. But the way things worked out, it didn't make much of a difference. The Jets ran only one more play, and still had a timeout of their own remaining. 29 seconds was more than enough time to call a play, run it, and have at least a second left to call a timeout before a FG.

That one play happened to be a huge one. The 18-yard pass to Braylon Edwards meant the difference between a 50-yard FG attempt and a 32-yard attempt. The baseline percentages for each kick are about 55% and 85%, respectively, effectively making that pass a +0.30 WPA play.

BAL-KC

Can someone explain to me why Jamaal Charles only got 9 carries in this game, and only 1 in the 2nd half? BAL came out of halftime with only a 3-point lead. Was KC already in desperation mode? Charles was gashing the BAL defense at 9.1 YPC. And while we're at it, why did Charles only get 13 carries per game all season when he was hammering defenses at 6.2 YPC?

The one Charles carry in the 2nd half did not end well for KC. On 4th and 1 from the BAL 33, Charles was stuffed for a 4-yd loss. On the outer edge of FG range in cold weather, it was a good decision to go for it, and would have been in any game situation. It was a turning point in the game, and the Chiefs would never threaten again.

I think offenses need to go deep periodically on these 3rd and short and 4th and short situations. These stack-the-line-with-extra-linemen-and-plunge plays are too predictable. Just run a normal set of plays you would on 2nd and 5. The average play in the NFL gains 5 yards. The median is 3. Force the defense to defend the entire field and teams will see their success rates improve, overall, in short yardage situations. I think they have this goal line mindset, where there is no deep pass option. With 40 yards of field ahead of you, use it every now and then.

Baltimore was up 23-7 with under 5 minutes to play and faced a 4th and 1 on the KC 25. They smartly went for it. A simple 1-yard gain ends the game. Willis McGahee cut through the KC defense for a TD. It wasn't terribly critical to the game outcome, but it shows that the strategy of relying on an offense to pick up an easy yard or two to seal a win, rather than kick and cross your fingers your defense can hold up, does work.

GB-PHI

Andy Reid is the Advanced NFL Stats Coach of the Week thanks to his call on 4th and goal from the 1. Down by 11 with about 4 minutes to play, Vick dove forward for the TD. It was smart for a couple reasons. First it's the percentage play. Second, the Eagles were down by 11 and needed two scores. They could have relied on the FG plus TD plus 2-point conversion to tie the game. But remember that only ties it. Further, the TD allows you to go for the 2-point conversion now rather than at the end of the game. It pulls the outcome of the conversion forward in time. So on the subsequent drive, you know whether or not you need a FG or TD.

It's hard to explain, and this is a very tricky concept. But think of it this way: Either the 2-point conversion is going to be good or not, whether you try it now or at the end of the regulation. You might as well know as soon as possible in the game if it will be successful. With that knowledge, you can adjust your strategy going forward. This knowledge has tremendous value, and makes the option to go for it on 4th and goal from the 1 that much more lucrative because it makes the TD now more lucrative.

Aaron Rodgers' 8-yd sack on 3rd down with 2:16 to play may have been a blessing in disguise for GB. Had Rodgers thrown the ball away for an incomplete pass, PHI, who had no timeouts remaining, would get the ball back ahead of the 2-min warning. As it happened, the sack allowed the clock to run, and PHI didn't get their first snap until 1:45 left.

Vick's interception to end the game was especially egregious because it was so unnecessary. It was 1st down on the GB 27, and there were 44 seconds remaining. There was absolutely no reason to force a jump ball into the end zone. They had enough time left to leave the entire field in play. In other words GB didn't only have to worry about the sidelines and end zone. Vick should have had options underneath the coverage.

Top offensive players of the weekend were Hasselbeck, Tomlinson, Garcon, and Heap.

Top defensive players of the weekend were Tramon Williams, Raheem Brock, Brodney Pool, and Eric Berry. In terms of EPA, you can throw in Tamba Hali, David Hawthorne, and Dawan Landry.

  • 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

40 Responses to “Wildcard Game Analysis”

  1. Anonymous says:

    "These stack-the-line-extra-linemen-and-plunge plays are too predictable. Just run a normal set of plays you would on 2nd and 5. The average play in the NFL gains 5 yards. The median is 3."

    This is not good logic. On 4th and 1, you want the play that has the highest percentage gaining at least one yard. A deep pass is like 50-50 odds in the best of times. If you make 10 yards half the time and incomplete half the time, that's much worse on 4th and 1 than a dumb run that gets 1 yard 70% of the time and loses 1 30% of the time, even though the latter play might have lower average gain.

  2. Brett says:

    I disagree, Anonymous. You are overvaluing 1st downs. 1st downs are great, but the ultimate goal is to score more points than the other team. A play-action pass in this situation may have a lower conversion rate like you say, but the possibility of scoring a TD is so much greater that it more than makes up for the lower success rate.

  3. Anonymous says:

    But the reward of a deep pass is much greater than a dumb run that gains 1 yard + first down.

    The fact that the defense has to prepare for both makes the chances of the success greater when you do run it. A game theory based analysis of this situation would be fairly easy to base your tendencies on if you can quantify the value of each play. The % of time a team should go deep is certainly nonzero.

  4. Ian Simcox says:

    First anon, have to say I agree with Brian on this one.

    The average NFL pass goes for 5-6 yards, yet on 4th and 1 the average pass gains 4 yards. Same with rushing, normal plays average 4 yards, 4th and 1 rushes go for about 2.

    This has to be due to offenses running overly cautious plays and defenses adjusting to that. Offenses must be changing, otherwise you'd have to ask why defenses don't play like it's 4th and 1 on every play.

    Green Bay had great success on goaline plays this season by going 5 wide, with Aaron Rodgers having the option to run it straight up the gut. It's easier to run the ball when there are no defensive players in the area. Get them out the way by stretching the field.

  5. Anonymous says:

    "The clock was running. Perhaps Caldwell was thinking it was possible to stop the drive there, and if the Jets made a FG, have enough seconds for a desperation FG drive."

    This is a terrible rationale because it's not possible. Let's say for arguments sake the Colts do just about the best they can hope for and force 2 incomplete passes. Now it's 4th and 8 and the average passing play takes about 6-7 seconds so they have about 14 seconds left. A field goal attempt takes another 4 seconds leaving them with 10 seconds left and a kickoff will take another 5 seconds. So they are somewhere around their own 20 yard like with 5 seconds left. 5 seconds is not enough time for a desperation field goal drive.

    Additionally, in post game comments Caldwell talked about how his real strategy was he wanted them to get extra snaps because they could fumble the snap like Kerry Collins did the week before. Basically he was betting on a low probability play with the risk of giving up a 30% swing in FG%. Not sure how anyone can like those payoffs.

  6. Kulko says:

    I don't think anon is completely wrong.

    Due to football rules, there is a discrete jump in expected Points when you reach the new first down. So naturally the optimal strategy for 4th and 2 is different then 2nd and 5, as you should choose plays which give you better probability of making it. So getting lower average yardage is to be expected.

    OTOH game theory also explicitely prooves, that you still should mix in different strategies, as long as they are not strictly inferior, to avoid your opoonent stacking the line.

    Of course the only proove, that offenses are doing wrong, would be if conversion rates declined or payoofs of different strategies where to grow significantly apart.

    That all said, I remember another very nice example from one of the Pats games (I think it was Cleveland)

    When the opposing offense faced 4th and 1 they broke huddle, in one of the usual runheavy formations. When NE staked the middle of the line too, the QB audibled into a 5 wide set. NE was forced to at least cover all eligble receivers and spread the filed, giving up the easy QB sneak, once the middle where cleared.

  7. jimlynch9999 says:

    You guys are focusing on the five yard average play but need to focus on the three yard median. That means that half the time you gain more than three yards, including all those incomplete passes. The real comparison would be the percentage of times a dive play on third and short makes two yards vs the percentage of times a normal play makes two yards.
    On a related note, I've never understood why teams don't script all plays in advance. You sometimes hear about teams that script their first 15, but I would think that scripting all plays would keep defenses surprised, as they wouldn't gain any knowledge from down and distance.

  8. Adam Davis says:

    "Low variance is the friend of the favorite."

    I understand this point, and I agree with it. But what to do when a team's dominance is defined by their attacking/gambling style? Gregg Williams' defense is predicated on gambling both at the line of scrimmage (blitzing) and in the secondary (anticipating/jumping routes). So if you have become the favorite *because* of your gambling style, does it make sense to abandon that style when faced with a (supposedly) inferior opponent??

  9. Brian Burke says:

    You DO NOT want to run the single play that gives you the best chance to gain a first down.

    You DO want a mixed strategy of plays that maximizes your overall probability of success.

    Even if you ignore the extra value of a deeper gain, you do not want to allow the defense the luxury of only having to defend a single play type.

  10. Anonymous says:

    It's the ol' counterintuitive idea (and fact) that running a play that has a LOWER expected rate of success actually increases your long-term success rate. While a deep ball might be 50/50, and a run up the middle 70/30, using the deep ball (theoretically) increaes the success rate of your run up the middle.

  11. jamerchant says:

    Any comment on the Colts decision to kick a field goal from the Jets 14 yard-line with 4:37 left (and go down 14-13) rather than go for the touchdown? I was playing around with the WP calculator and I think kicking the field goal dropped the Colts WP by 16% or so (I did the calculation quickly so I may be wrong on the details).

    The really interesting thing (to me) was that there is very little difference in WP between kicking off to your opponent down 1 with 4 minutes to go and giving your opponent the ball at the 14 down 4 with 4 minutes to go. In fact, the down 4 scenario has a WP of .39, and the down 1 scenario (assuming a kickoff to the 25) has a WP of .29.

  12. Pat says:

    Brian, what did you think of the decision to kick (Aker's 2nd miss)?

  13. William says:

    you criticized the Saints for gambling on coverages and jumping routes because they were the favorite, saying that they should have played more conservatively in coverage. In theory I think this a good idea but the Saints have been playing that way for a long time, it is not something that they implemented only for Seattle and I think trying to alter your basic playing style for one game might be more detrimental than the risky coverage.

  14. Brian Burke says:

    Good point, William. But they were up by 10 at one point. Can't they throttle that back just a little?

    Pat-That's definitely a mistake to try the FG. 4th and 1 on the GB 16, down by 11, 13 min to play? Should be a no-brainer if you have any confidence in your team's ability to get 1 yd, which I guess Reid did not have.

  15. Ed says:

    Brian,

    This is a questions not a critique. Revis held Wayne to 1 yard on 1 catch the whole game. How does EPA and WPA account for a player that is so dominant that that a HOF QB refuses to throws his way?

  16. Brian Burke says:

    They can't, right now. EPA or WPA as a concept can do it, but using the NFL play-by-play as the record of what happened on the play is severely limiting. It's not the concepts of WPA or EPA, it's the data.

  17. Anonymous says:

    How about all those passes in 3rd-Down-and-short by Indy early in the game? I really think that they should have run on these plays. Your analysis supports running in these situations. The fact that they did not killed a number of drives and might have been the difference in the game.

  18. Jim Glass says:

    Brian: The Colts are up on Jets 16-14, 3:10 left in the game. The Jets have just gone three and apparently out. To quote the play by play...

    Weatherford punts 46 yards to IND 29, Center-46-T.Purdum, fair catch by 15-B.White. PENALTY on IND-10-T.Smith, Running Into the Kicker, 5 yards, enforced at NYJ 25 - No Play.

    At 3:02 the Jets get a first down on their 30 instead of the Colts having a first down on their 29. How much was that swing worth?

  19. Anonymous says:

    Jim, Colts were down 14-13 at that point.

  20. Bigmouth says:

    Brian, I'll bet Bill Walsh would have agreed with you. I believe he preached that 3rd and short is a great situation to try for a TD because it's the rare situation where you know what the coverage will be.

  21. Bigmouth says:

    Ed, I just wanted to raise the possibility that teams are NOT staying away from exceptional CBs like Revis so much as targeting far less capable CBs on the same team. I say this because I think the concept of a "shutdown" CB is something of a myth in today's NFL. The rules prohibiting contact after 5 yards are too restrictive, and the ubiquity of zone defenses means 1-1 situations are rare.

    Of course, this is sheer conjecture on my part. Only empirical research can answer the question.

  22. Anonymous says:

    Is there much of a difference between "staying away from Revis" and "targeting far less capable CBs"?

  23. Bigmouth says:

    Anonymous, yes, at least possibly. Let's say you have a league-average CB and a CB who's well-below average. Wouldn't it still make sense to target the well-below average CB, at least until the other team shows it can stop you? Again, though, this is sheer conjecture. But since individual defensive-metrics are scarce, I think it's a possibility we have to consider.

  24. Bigmouth says:

    PS: Just to clarify, I'm not saying Revis is a league-average CB. Just that there might be factors other than his talent that influence any disproportionate playcalling against him

  25. Ed says:

    Bigmouth,

    As you yourself say, your claim is not falsifiable as well as being highly subjective. There is no difference in targeting lesser capable CB vs avoiding a CB. It's semantics at it's best.

    Additionally, there is a limit to your argument, as manning should be targeting lesser CB's ALL THE TIME, yet Wayne got 100+ receptions. Additionally, NBC tracked the times he was lined up 1-1, and his targets(1).

    There is a whole other matter to consider. I would think that the drop of from an above avg CB, to a below avg CB is not significant enough in pro sports, that a HOF QB would avoid "one" CB the entire game.

  26. Anonymous says:

    "How does EPA and WPA account for a player that is so dominant that that a HOF QB refuses to throws his way?"

    Look at the receiver's average EPA and WPA for the season, then compare it to his EPA/WPA for the game. That's the EPA/WPA that the good DB is responsible for.

  27. Brian Burke says:

    If every DB was assigned full time to each receiver man-to-man, that might work, but that's not the case.

    If we knew which DBs were on the field for each play, which ones were in coverage, which coverage was called, and which DBs were targeted, we could do a lot statistically. Only the teams themselves have that info.

  28. Ed says:

    Bigmouth,

    I understand why statistics would be hard pressed in assessing a CB's worth as well as individual O-Lineman's pass protection value(what about individual LB coverage?).

    Inherently, the above player's job is defensive in nature, and it is hard prove that a player avoided something from happening, as it is proving a negative. Cutting out near-infinite possibilities(in the context of a football game, it is much less) is harder then proving a near 1:1 causal event like a WR catch.

    A system like UZR in baseball would be a start. However, there is different coverages. So you not only need to know the coverage, but also the result.

  29. Jim Glass says:

    Jim, Colts were down 14-13 at that point.

    My bad -- but the same question.

    The Jets players are telling the papers that penalty saved the game for them.

  30. Anonymous says:

    Am I crazy or did Andy Reid kick twice in managable 4th down situations (both kicks in question were of course, missed). I was expecting to see an analysis of both of those missed kicks.

  31. Buzz says:

    Jim,

    Just using the WP calculator I come up with the following. If there was no penalty the Jets would have had a 46% chance to win at that point. After the penalty they had an 81% chance to win for a change of 35% due to the penalty. Bigger play than the one Brian mentions in the write-up to Edwards for a 30% increase.

  32. Ben says:

    Brian,

    Can you explain how you converted Win Probability into a point spread (as you did with NO-SEA in your post)?

  33. Jim Glass says:

    Thanks, Buzz. I'm wondering if after the game Peyton gave Smith a consoling pat on the head, "That's OK kid, we learn from experience", or...

    This is the fourth time in the playoffs that Peyton's played better than the average winning playoff QB (by AYA) and been eliminated. Over the last 15 years no other QB has lost more than one of these.

  34. Brian Burke says:

    Here is where I did the conversion from pt spread to win probability:
    http://www.advancednflstats.com/2008/03/office-pool-3.html

  35. Anonymous says:

    Brian, do you have a chart where it shows the expected points scored on the drive from the yard line the drive starts (EX: 1.5 points/drive from the 20 yard line). I'm not talking about the one where it has negative values for drives starting beyond the 15 yard line. I just want to know the total points scored from drives at that yard line/possessions.

  36. Brian Burke says:

    It's not exactly what you're looking for, but this might be helpful.

  37. Jonathan says:

    Every year I laugh my head off at talking heads who think that Manning doesn't show up in playoff games. I've heard people not even bother to argue that Manning was sub-par in the NYJ game, they have actually presupposed that he didn't bring the stuff. i.e. "Why do you think Manning couldn't bring his A-game for yet another playoff game, was he really that bad, or was the Jets defense that good?"

    Laughable. Horribly laughable. 8.7 YPA, 0 INT, one sack for six yards. Against THAT defense, with Revis as Reggie Wayne's personal shadow and a pedestrian 3.5 ypc out of his RBs.

    What else is he supposed to do? Play special teams so his team doesn't take penalties? Slice off his coach's hands so he doesn't call the dumbest timeout imaginable? Pay Bill Belichek out of pocket to be his personal assistant?

  38. Ian Simcox says:

    Jonathon - I did wonder whether although Peyton's salary is justified for him, does it affect the rest of his team because it takes up too much cap space.

    According to the USAToday.com stats, in 2009 Manning was, by $4.5m, the most expensive quarterback in the league. That $4.5m is more than the Colts spend on any LB, DT or RB, and is the equivalent of a Vilma, Fletcher or Merriman (at LB) or McFadden, W Parker or Ryan Grant (at RB).

    There is very little more I think Peyton can do on the field, but one option for him, if he wants to improve his chances of another ring, would be to take a pay cut and let the Colts bring a bit more depth and quality to their squad.

  39. Jim Glass says:

    Ian, I think you are on the money with this. It makes perfect sense to build a team around a great QB like Peyton, spending the cap disproportionately on him and the few players needed for him to be most productive -- but the Colts may have pushed this to a fault and beyond.

    The seven highest-paid Indianapolis Colts in 2009 will have a combined $81.3-million cap cost -- which leaves the bottom 46 players on the active roster, eight practice-squad players and, say, estimated injured-reserve players to split the remaining estimated $40.7 million of the cap. [a href="http://sportsillustrated.cnn.com/2008/writers/peter_king/09/01/mmqb/2.html#">Peter King]

    That's basically 2/3rds of the cap on seven players who, IIRC, were Peyton, five more O players and Freeny. It's no wonder they've always had D problems.

    I saw Polian say the Colts use the Tampa 2 defense because it is the cheapest, leaving more money for Peyton's O. On D they have to pay real money only for a pass rusher -- the zone DBs they can plug out as soon as they reach their free agency year and eligible for a raise, plugging in cheap draft pick replacements.

    "Defense wins championships" may be exaggerated, but geeze. It's one thing to build an O around a QB -- but how many QBs have had their team's defensive schemes catered to them?

    A big advantage the Pats had with Brady for a bunch of years was that as a sixth-round draft pick he was *cheap*, the greatest bargain in the NFL against the cap. When the Pats took him with that pick, Belichick and Brady both won the lottery.

  40. Jim Glass says:

    Sorry about messing up that link. :-(

Leave a Reply