Conservative decision-making is killing the Buccaneers' playoff hopes

Greg Schiano is a conservative coach. This shouldn't be a surprise to anyone, as he is a coach from the defensive side of the ball. Coaches from that side of the ball like offenses that protect the football, run the ball and don't take too many risks. Tony Dungy was much the same way. Yet that's a terrible way to win football games, and that attitude has cost the Tampa Bay Buccaneers repeatedly this season.

Given the opportunity to go for it on fourth down, Greg Schiano has passed it up time after time. Unless not going for it meant a certain loss, Schiano would always go for the conservative option. Essentially, he's been playing to keep the Bucs in the game as long as possible, rather than to give them the best chance to win a game. To understand this, we need to think of a game of football as a game of probabilities, much like a poker player would.

Around the league, NFL coaches are too conservative on fourth down and in their game plans. They have a tendency to stick to long-established wisdom based on feel, rather than solid analysis of probabilities. This may actually be rational: after all, they're less likely to be criticized for sticking to what "everyone knows" rather than going with the correct but unconventional call. There have been many, many articles on this simple fact and it is really rather undeniable. Advanced NFL Stats has an excellent, if slightly condescending article that explains a lot of the basic principles behind fourth-down decision making.

This is why I was a big advocate of hiring someone like Chip Kelly this offseason. While I've been very satisfied with Greg Schiano and he has done a remarkable job of turning around the Buccaneers, he could learn a lot from Kelly's application of probability and statistics. A large part of his offensive success is simply a result of being more aggressive than the opponent. And if hundreds of thousands of hands of poker have ever taught me anything, it's that being aggressive is almost always better than being conservative.

While the conservative playcalling has been evident on fourth downs, another issue has been third down. Specifically, the tendency of the coaching staff to be overly conservative on third-and-long. The nickname "Third-and-Ware" has become popular, as the team continues to run screens and draws to just get a few yards instead of actually attempting a pass down the field with one of the strongest arms in the NFL. We saw a prime example of this lunacy against the Denver Broncos. With the secondary clearly struggling and down four points, instead of attempting to get a first down, the Bucs called a draw with D.J. Ware on third-and-20 from their own 10-yard line. It's one thing to do so at the end of a half to end the half, but there's little point to gaining eight yards of field position against Peyton Manning.

This conservative playcalling has hurt the Bucs over and over again, and it may actually end up costing them the playoffs. It cost them a win last week, and at the very least hurt their already slim chances at a win this week. How can I be so sure of that? Because I checked the numbers. Let's go through a couple of these scenarios:

@Denver: Kicking a field goal down 18 in the fourth quarter (1)

With 15:00 left in the fourth quarter and down 18 points, the Tampa Bay Buccaneers lined up for a 50-yard field goal. Hitting the field goal would cut the Broncos' lead to 15 points, making it a two-score game. At first glance, a two-score game is a better than a three-score game. But it's not that simple.

First, we must realize that field goals are not automatic. In fact, based on league-wide averages, the Bucs should expect to make a 50-yard field goal about 57% of the time, according to Advanced NFL Stats' data. Another important fact: failing to get a first down would give the Broncos the ball at the Denver 32, while failing to make the field goal would give it to them at the 40-yard line. In addition, the odds of converting a fourth-and-six aren't bad, with a league average success rate of 45%

If we plug the data we have into Advanced NFL Stats' excellent fourth down probability calculator, the result is clear: if the Buccaneers thought they had a 39% chance of making the first down, they should have gone for it. This decision wasn't nearly the worst of them all, as the win probability graph suggests going for it or kicking the field goal were more or less equally valid decisions.

Unfortunately, that misses a few issues. The Bucs had a defense that had failed to stop Manning all game long. It was very likely that Manning would continue to do so and at least get a field goal on the next drive. The Broncos indeed did do that, and they turned what was a two-score game back into a three-score game. Schiano certainly kept the Bucs in the game a little longer with a field goal (that was far from automatic), but he didn't really improve their chances of winning the game.

@Denver: Kicking a field goal, down 18, in the fourth quarter (2)

Sound familiar? Yes, with 3:27 remaining in the game the Bucs faced a similar quandary: whether to go for it on fourth down or kick a field goal while down three scores. Had they been more aggressive earlier in the game, this could have been a two-score game, but instead they were facing the exact same situation they had faced earlier. It's important to note that the only reason this wasn't an automatic go-for-it situation was the fact that the Broncos had missed a 47-yard field goal.

All of the previous caveats apply, except there's less time now, which means you want to increase your variance. You want to go for bigger rewards to increase your chance of winning immediately. As a result, if the Bucs thought they had just a 30% chance of converting for a first down they should have gone for it rather than kicked a field goal. Kicking here was clearly the inferior option -- but it's conservative coaching that caused them to 'take the points'. Important to note: you're not 'taking the points', you're just attempting a field goal that fails over 50% of the time. So in terms of probability, you're taking 1.5 points.

Even though the Bucs eventually cut the deficit to one score, they never had enough time to come back. The field goal gave them the illusion that they were still in the game, while a touchdown could have actually put them in the game. They needed to score a touchdown eventually, and it's easier to score one when you have extra time and timeouts than when you have mere seconds at the end of a game.

vs. Atlanta: Kicking a field goal in the second quarter

Goal-line decision making has not been a real weakness of Greg Schiano this season. He's repeatedly gone for it on fourth-and-short at the goal line, and that should be pretty routine for coaches across the league. It's easy to see why: if you get a touchdown, you're happy. If you fail to convert, the opponent is backed up at their own goal-line and you have a good chance of forcing a safety (especially so with the Bucs' tendency for tackles for loss) or a punt. Essentially, the worst-case scenario is a touchback, which still isn't a horrible proposition. Volumes of articles have been written about this issue, and the conclusion is simple: if you're inside the opponent's five-yard-line and it's fourth down, you almost always have to go for it.

And that was certainly the case when the Bucs faced fourth-and-goal at the Atlanta four-yard line late in the second quarter. The fourth down calculator suggests that the Bucs would have to score a touchdown either 32% of the time (in expected points) or 35% of the time (in win probability)* to go for it. The Bucs have been outstanding in the red zone this season, and getting four yards on a fade pass should succeed well over 35% of the time.

Instead, the Bucs settled for a field goal, tying the game. They left a few expected points on the field, and that cost them when they eventually lost by one point. This decision would have looked even worse had Matt Bryant not missed a chip shot field goal for the Falcons a few minutes later.

*A quick note on win probability versus expected points: the former takes into account time remaining and current score, while the latter does not. Expected points speaks merely to the pure 'value' of a play, while win probability puts it in the context of the game.

vs. Atlanta: Kicking a field goal in the fourth quarter

Everyone remembers this one: Connor Barth couldn't hit a 57-yard field goal to put the Bucs up with 3:37 left in the fourth quarter of last week's game. The Bucs didn't get the ball back until there were just 20 seconds left on the clock and they stood no real chance of making up the deficit. It gave them a crucial loss in the playoff hunt.

And, of course, it was the wrong call. This one wasn't even close: on fourth-and-seven with that little time remaining in the game, you would need either a 30% chance (expected points) or a 25% chance (win probability) to go for it. Especially so when you realize that Connor Barth doesn't have the strongest leg, and you're facing a quality offense that had had little trouble slicing up a Tampa Bay secondary so far. The Bucs should have gone for it, going for the touchdown, or at least giving themselves a better chance of making the field goal, and taking more time off the clock.

The worst part: Schiano said after the game that he would have punted sooner than going for it. Going for it had a 42% chance of winning, kicking a field goal 34% and punting just 19%! A punt wouldn't just have been the worst of the three decision, it would have halved the team's chance of winning the game. That shows just how far off Schiano was in his decision making, and how he needs to improve.

Both of these past two decisions hurt the Bucs' chances to win a crucial game in the playoff race, and may end up being the reason why the team doesn't make the playoffs.

@New York Giants: Punting on fourth-and-two from midfield in the third quarter

Here's another game the Buccaneers lost by a single score, that they may have had a better shot of winning had they been more aggressive early on. Yes, this happened when the Bucs were up 27-13, but a good call is a good call regardless of the score. The Bucs had had little trouble going through the New York defense, and faced a fourth-and-two from their 49-yard line with 6:28 left in the third quarter.

The Bucs need to be able to convert fourth-and-two 44% of the time to make going for it the right call here based on expected points. Given the fact that Josh Freeman has been successful on every short quarterback sneak, and this should be a fairly easy call. In fact, this call should be routine almost regardless of the game situation. Here's one important note: the league average success rate at fourth-and-two is 60%.

One interesting point: win probability does suggest that punting is slightly better, with a 98% win probability versus a 97% win probability for going for it. This decision wasn't necessarily bad, but it's emblematic of conservative playcalling.

@New York Giants: Punting on fourth-and-two from midfield in the fourth quarter

The Bucs' horrible secondary came back to avenge itself there, as the Bucs found themselves up 27-19 at midfield midway through the fourth quarter. This time, they were on the Giants' 48-yard line and had a fourth-and-one -- and once again, they punted. The problem with punting there is that you generally get around 30-35 yards of field position, which isn't actually all that helpful, while you give up possession. In this case, the Bucs managed 36 yards of field position.

This call is a lot more interesting, though. In the expected points model, the Bucs would need just a 41% chance of getting the first down to go for it. However, according to win probability they would need a whopping 73% to go for it. Of course, there's one fact we're neglecting to mention: the league average rate of converting a fourth-and-one is a ridiculous 74%! I would guess the Bucs would actually have a slightly better chance on a quarterback sneak, especially so because they still had Carl Nicks in the lineup at the time.

Yes, the Buccaneers were up at the time, but it's a good illustration of the fact that you can't slow down just because you're up in a game. Being aggressive is generally a good thing, whether you're losing or winning.

@Redskins: Failing to go for it on fourth down on four different occasions.

This started early in the game, and had consequences throughout. In the first quarter, the Bucs passed up opportunities to go for it twice. Instead, they punted on fourth-and-five from midfield, and kicked a field goal on fourth-and-three from the 32-yard line. At this point, you should know the math intuitively here: the Bucs would need a conversion chance of respectively 40% and 41%(EP)/44%(WP) to go for it in those situations.

But other situations would crop up. In the second quarter, the Bucs trailed 21-3. On fourth down, at the two-minute warning, the Bucs kicked a 57-yard field goal instead of going for it on fourth-and-7. This one was especially egregious: 57-yarders often miss, and the Bucs needed to score a lot of points to win this game. Expected points suggest they needed just a 31% chance of conversion to make going for it the right call, while win probability says a 20% chance (!) would have been enough. Clearly the right call was going for it.

The worst part of this: I would venture a guess that Schiano never even pondered going for it here.

Are there more in this game? Yep. Here's one: third quarter, down 21-6, the Bucs punt on fourth-and-one from their own 25-yard line. Robert Griffin III had been slicing them up all game, but playing the field position game sounded smart. Well, it's not. The Bucs would need just a 67% chance(WP) to convert to make this the right decision. Again: the league average conversion rate on fourth-and-one is 74%, and the Bucs should be above that on quarterback sneaks.

Here's the worst one yet, by a mile, and it isn't at all intuitive. The Bucs faced a fourth-and-10 with 4:50 left in the fourth quarter from the Redskins' 48-yard line. They were down by just two points at that point. Win probability suggests they need just a 14% chance of making the first down to make going for it the right call. The league average of converting fourth-and-10? 37%.

@ Dallas: Four instances of failing to go for it on fourth down

Same old story. Tie game, second quarter, 12:08 left. On fourth-and-five, the Bucs punt. They needed a 36% chance of converting the fourth down to go for it, the league average success rate is 49%. Given Freeman's scattershot accuracy at the time, this call is fairly debatable, but being more aggressive surely wouldn't have hurt.

Later in the second quarter a more egregious decision: fourth-and-three from the 47. They would need a 36% rate of conversion, and the league average is 57%. Again: should have gone for it.

In the third quarter, down by three points at the Dallas 42-yard line on fourth-and-six. Would have needed a 35% chance of making it to make going for it the right decision. League average conversion rate: 45%.

And, the worst one: kicking a field goal, down 10 points with just 44 seconds remaining in the game from the Dallas 10-yard line. Think about this: what's more likely if you recover the ensuing onside kick: making it into field goal range, or making it back to the 10-yard line? I know which one, and you're extremely unlikely to get another chance of getting a touchdown in that game. This decision was really bad, relative to the chance of winning: they were twice as likely to win by going for it rather than kicking a field goal. That was a classic decision of keeping the hope alive, rather than maximizing the chance to win.

Conclusion

The Tampa Bay Buccaneers simply don't go for it often enough on fourth down, and they're leaving points on the field. In the six losses this season, I've identified a whopping 14 instances where the Bucs should have gone for it on fourth down. Keep in mind that I haven't even talked about the numerous bad decisions in wins this season, which certainly exist too.

Bottom line: Schiano's conservative decision making is hurting this team, and there is no reason for him to be this conservative. It is trivially easy to become acquainted with general win probabilities, and the situations of where to go for it and where not to go for it could be worked through in a few hours at the most. And those few hours could have meant the difference between a playoff berth and another January spent at home this year.