Toxic Differential says Bears will miss the playoffs

Hey, I love statistics as much as anyone, and likely more than most – but this looks suspiciously like a case of someone looking at the results, and then creating a statistic to fit those results.

The theory here is that the “Toxic Differential” will result in a particular won-loss record. It’s a nice simple idea, but a bit too simple in my opinion.

First, I don’t like the ‘mix’ of numbers. There are only two plus/minus numbers, added together, and the variation of the two columns is quite intriguing. Just looking at the numbers, it would seem that the Steelers, Eagles, Chargers, Giants and Bucs (38-22, .633) should have BETTER records than the Packers, Patriots, Falcons, Chiefs and Jets (45-15, .750). That’s a difference in final record between 10-6 and 12-4 between the top 5 and next 5, when projected out.

Whereas, it looks like turnover differential is still a more powerful predictor of, or has greater effect on, the result (W-L). The teams with the top two records (Pats and Falcons) are #2 and #4 in turnover differential (+24), but only breaking even in big plays (+2), while numbers 1 and 3 (Eagles and Steelers) also are in a favorable position. Meanwhile, the Chargers and Giants have negative turnover differentials, and very well could wind up on the outside looking in.

The second issue with these simple numbers are that not all big plays are created equal. There are the factors of the position on the field and the score at the time. A 20-yard run from your own 8 yard line is not of the same quality as a 20-yard run from the opponent’s 28. Furthermore, a 30-yard pass completion by a team that is losing by 17 points isn’t as meaningful as a 30-yard pass by a team in a 3-point game (either winning or losing).

Turnovers in a game with a large current point spread may not be as meaningful as in a tight game, but they are equally as meaningful on all parts of the field. If your defense gets a takeaway on your own 10-yard line, that saves 3 or 7 points; if it gets a takeaway on the opponent’s 10-yard line, then it generates 3 or 7 points (usually, of course).

The third issue has been touched on repeatedly – the lack of big plays by special teams being taken into account. Certainly field position will have a huge influence over scoring, and the quality of any resulting big plays on offense (or given up on defense), and this could easily be calculated by the proponents of this statistic (why they didn’t include this in the first place is beyond me.)

And those are only my cursory issues with the numbers, as stated. The linked article has more holes than a prevent defense. Some quick examples:

The Colts lost Sunday in large part because they couldn’t stop the Cowboys’ ground game and couldn’t run the football themselves. But, the biggest reason they’re 6-6 is their toxic differential.

Once again, everyone repeat: “Correlation does not imply causation.” The biggest reason is NOT their toxic differential, but the fact that Peyton Manning has been doing his best impression of Rex Grossman over the past few weeks. A statistic is not the reason the Colts have been sucking of late. That’s a very poorly-written hyphothesis.

“In my experience in coaching, there’s not a better indicator or playoff predictor than the toxic differential,” Billick said. “You can say, ‘You gotta run the football, you gotta stop the run.’ But at the end of the day, maybe five out of the top 10 rushing teams will make the playoffs.”

Did everyone catch that? Billick himself mentions the perception of run offense AND run defense generating wins, and then mentions only the top rushing offenses in their relationship to the playoffs!! Talk about moving the goalposts, and he didn’t even wait two sentences to do it. Brian, what happened to the teams stopping the run, and whether or not they make the playoffs?

Billick is spot on. What’s more interesting is where the top rushing teams that didn’t make the playoffs in 2009 finished in toxic differential. The Titans ended up 27th at minus-17, with the Dolphins right behind them at 29th (minus-33). How about poor Cleveland? The Browns were dead last with an awful minus-52.

Or, we can look at the part of the equation that Brian conveniently forgot in the next friggin’ sentence, rush defense. The Titans were tied at #10 (respectable), the Dolphins were #18, and the Browns were #28. The other top rushing teams on the list that didn’t make the playoffs were the Panthers and Jaguars, #22 and #19 in rush defense, respectively. The only team that doesn’t really make sense is the Titans, who were #2 in rush offense and #10 in rush defense. However, they played two different QBs, their top receiver was also their top rusher (CJ2K), and they finished 8-8, well behind the 14-2 Colts.

But we don’t really have any evidence that the teams with the best balance of both rushing offense AND rushing defense fared any worse than those with the best toxic differentials.

Here, they aggregate Toxic Differential for the past ten years, and make some conclusions:

The top 12 teams are a veritable who’s who of the NFL over the past decade. Eight of the 10 Super Bowl winners in the 2000s are represented, as well as both teams that won multiple Lombardi Trophies (Patriots and Steelers). The regular-season team of the decade is listed (Colts), as well as the NFC’s best club over the same span (Eagles).

Misleading, at best, once again. Of those 12 teams they mention, only five have won the Super Bowl (Steelers, Colts, Patriots, Bucs, and Ravens), and only one more has participated (Eagles), while the other six (Chargers, Packers, Broncos, Jets, Cowboys, and Jaguars) haven’t even been in a Super Bowl during this period.

My hypothesis is that turnover differential is more important than toxic differential or big-play differential. Let’s look at the six teams on the list that participated in the Super Bowl versus the six that didn’t.

Super Bowl 6: Toxic differential +806
Non-Super 6: Toxic differential +405

Wow! There you have it, right? Not so fast….let’s break out these numbers further.

Super Bowl 6: Big-play differential +483
Non-Super 6: Big-play differential +354

Hmmm….what’s going on here?

Super Bowl 6: Turnover differential +323
Non-Super 6: Turnover differential +51

There you have it. Game, set, match. If you were only analyzing turnover differential, those non-Super 6 might not even be ON the list (certainly the Cowboys would be far down).

And my personal favorite:

“Teams that win the explosive-gains differential win 65 percent of their games,” he said. “That’s just a plus-1. Teams that have accomplished both a positive turnover and explosive gains differential win over 92 percent of their games.”

Really, the team that has more takeaways than turnovers and more big plays than the other team will win? Geez, I had no idea.

I haven’t been able to find any stats for the ten-year period between 2000-2009 (regular seasons), and I don’t want to compile it myself, but I did find this:

http://www.usatoday.com/sports/football/nfl/2006-02-04-turnovers_x.htm

In the 201 games during the 2005 regular season where there was a difference in turnover margin, the team that won the turnover battle was 140-61.

In the 57 playoff games since the 2000 season where Team A has an edge in turnover margin over Team B, Team A is 46-11, including 7-2 in this postseason.

In the 31 Super Bowls where one team wins the turnover battle, that team is – get ready for this – 29-2.

Again, I realize it’s not the same time period and is a smaller sample, but by the numbers listed, in 2005 teams that won the turnover battle won 70% of their games (better than the 65% for big plays). This increased to 81% in the playoffs over a 5-year period, and 94% in the 31 Super Bowls where a team won the turnover battle.

After further analysis, I think this is much ado about nothing. Turnovers are a fine metric of winning and losing in the short (game), medium (season) and long (decade) runs. Adding big plays to turnovers is like adding bad weather to drinking alcohol while driving – of course it’s going to seal the deal.

But I propose that turnover differential is a much better indicator than big-play differential, and equally as good as,if not better than, toxic differential when it’s all said and done.