This guy has spent endless hours pouring over numbers confirming what I've been complaining about about the Man Who Will Be Nameless. It is far better to NOT throw an interception than to throw a TD pass.
Turnovers, via pass or fumble, are perhaps the single worst thing that can happen to your team, even with a good defense. Woody Hayes' comment on why he ran the ball so much: "There's only three things that can happen when you pass, and two of them are bad". Long read, but worth it. Two of our three principal running backs also have what politely could be termed slippery fingers.
------
The winning significance of touchdown passes vs. avoiding interceptions
February 7, 2006
Rick Cina
Watching a quarterback throw a touchdown pass is certainly more exciting than watching a quarterback who carefully avoids interceptions. It’s much easier to overlook many of the interceptions that a quarterback throws if he also slings a substantial number of touchdown passes. After all, when it comes to winning games, it’s more important for a quarterback to throw for touchdowns than it is to keep the ball out of the hands of the opponent. That seems to be the conventional wisdom, anyway.
But, because I'm curious, I decided to analyze all 256 football games in 2005 to put that conventional wisdom to the test. I wanted to learn more about how important (or unimportant) avoiding interceptions can be to a team’s chances of winning games. Since a touchdown pass is an automatic 7 points, and an interception (or turnover) may or may not eventually lead to points, I have usually assumed that touchdown passes are much more important than avoiding interceptions are in the winning equation. So I was a little surprised by what I found.
What happened when Brett Favre didn’t throw any interceptions?
It all started when I decided to examine the record of the Green Bay Packers when Brett Favre didn’t throw any interceptions. I decided to look at 6 years worth of Packers regular season games (2000-05) to obtain a relatively large sample size. Of those 96 games, I was able to find 32 in which Favre finished the game without throwing an interception. The Packers won 28 of those no interception games (28-4). I also identified 35 games in which Favre threw 1 interception. The Packers won 21 of them (21-14). So, since the 2000 season, the Packers had a record of 49-18 (.731) when Favre threw 1 or no interceptions. But then the tables turned when Favre threw multiple interceptions. The Packers only won 8 out of 29 games (8-21, .276) in that 6 year span when Favre threw 2 or more interceptions.
Now, I had assumed that a similar winning pattern would occur with touchdown passes thrown---the more touchdown passes thrown in a game, the more wins. But while there was a correlation between touchdown passes thrown and wins, it didn’t appear to be nearly as strong as the correlation between avoiding interceptions and wins. For I found that when Favre threw 1 or no touchdown passes in a game, Green Bay’s record was a respectable 24-20 (.545). But in the 52 games in which Favre threw 2 or more touchdown passes, the Packers owned a solid, yet unspectacular 33-19 (.635) record.
I expected there to be a much higher concentration of wins when Favre threw multiple touchdown passes in a game, and I expected there to be a lower concentration of wins in the games he didn’t throw any or just 1 touchdown pass. Instead, the Green Bay winning percentage wasn’t substantially higher (.635) when Favre threw multiple touchdown passes than when he threw just one or no touchdown passes (.545), especially when comparing those winning percentages to the disparity in winning percentages between the 0 or 1 interception games (.731) and the 2 or more interception games (.276).
Is this pattern of winning duplicated in the rest of the NFL?
That’s when I decided to probe further. I wanted to see if a similar pattern of winning existed in the rest of the NFL in terms of the impact of avoiding interceptions versus throwing touchdown passes.
The problem I ran into, though, was that there were just too many games in which both quarterbacks threw 0 interceptions, for example, or when both threw 2 touchdown passes, and thus I still had to count it as both a loss and a win in those situations in which a tie existed. In other words, each just canceled out the other. That’s exactly what happened in the October 23, 2005, Packers-Vikings game in Minnesota, for example. Both Culpepper and Favre threw 2 touchdown passes and no interceptions that day, which meant that both teams were credited with a loss as well as a win for throwing multiple touchdown passes and no interceptions.
Still, when I examined each of the 256 games this way, I did manage to find a modestly similar pattern to the one I found with Brett Favre and the Packers. For instance, teams that had quarterbacks who threw 0 interceptions had a 131-68 record (.658), and teams had a 22-61 record (.265) when their quarterback threw 2 interceptions in a game. On the other hand, teams that didn’t get any touchdown passes from their quarterbacks had a 53-83 record (.390), while teams that had quarterbacks who threw 2 touchdown passes had a 74-53 record (.583). So, once again, the correlation didn’t seem to be as strong between throwing multiple touchdown passes and winning as it was between avoiding interceptions and winning. But I still wasn’t satisfied with the informative quality those numbers, as I thought that the significant number of ties might undermine what I was trying to gauge.
Winning percentages with more touchdown passes vs. fewer interceptions
So I went through all 256 games in the 2005 season yet again to compare how well throwing touchdown passes might predict wins compared to how well avoiding interceptions might predict wins, but this time I excluded the ties, only counting games in which one team had more interceptions or touchdown passes than the other. What I found was a little unexpected.
Considering there were more passing touchdowns (644) than both rushing touchdowns (431) and return (punt, kickoff, interception, fumble) touchdowns (97) combined in 2005, I would have expected that the team with more passing touchdowns than the other team would have almost always won the game. While they did win most of the time, I expected a higher winning percentage than I found.
In my analysis there were 173 games (out of 256) in which one team had more touchdown passes than the other team (the rest were ties). And in those games, the team with more touchdown passes had a 121-52 record, a .699 winning percentage. That translates to 11.2 wins per 16 game season. Not bad at all.
But then I looked at the record of teams that threw fewer interceptions than the other team. Out of 171 such games, the record for the fewer interception team was 133-38, a .778 winning percentage. That translates to about 12.5 wins per 16 game season, which is an even better winning rate than the more touchdown pass games.
What these statistics seem to indicate, then, is that throwing fewer interceptions was a better predictor of wins in 2005 than throwing more touchdown passes was. Now, it might be overly ambitious to say that avoiding interceptions is actually more important in the winning equation than throwing touchdown passes. But such a conclusion might be reasonable nonetheless.
And, by the way, avoiding turnovers in general (fumbles lost also) can also be considered a very important ingredient in the winning equation. There were a straight 200 games in which one team had more or fewer turnovers than the other. Teams that had fewer turnovers had a 160-40 record, an .800 winning percentage. That’s about a 13-3 regular season record. Of those 40 relatively rare instances in which the winning team actually lost more turnovers than the losing team, 26 had just one more turnover than the other team. So, out of 256 games in 205, there were only 14 instances in which the winning team had either 2 (11) or 3 (3) more turnovers than the losing team.
But interceptions occur because a quarterback has to play from behind, right?
One of the probable responses to the suggestion that avoiding interceptions might be just as, or possibly more important to winning than throwing touchdown passes is that the losing team usually has a quarterback who has to play from behind for a longer portion of the game than the winning team’s quarterback does, and, as we all know, playing from behind means that a quarterback usually has to throw quite a few more passes and thus risk more interceptions. Therefore, throwing more interceptions can largely be a consequence of playing while behind.
While that explanation seems very reasonable on the surface, I decided to test it out by comparing the interception rates of quarterbacks when playing from behind to interception rates when they’re ahead. And I found that, with a few exceptions, most quarterbacks avoided interceptions rather well when their team was trailing, and there really weren’t substantial differences between their interception rates while ahead compared to when their team was trailing.
I first decided to examine the passing statistics for the primary starting quarterbacks from the 17 teams that finished the season with a winning record. Later I compared them to starting quarterbacks on the 15 non-winning (the Falcons finished 8-8) teams in 2005.
The 17 primary quarterbacks from winning teams ranged from Kyle Orton and Gus Frerotte of the Bears and Dolphins to Peyton Manning and Tom Brady of the Colts and Patriots. After adding their passing numbers together, I was able to determine that all 17 winning quarterbacks had a combined 2.8 interception percentage (about 1 interception every 36 passes) while passing from behind, and a 2.4 interception percentage (about 1 every 42 passes) for these winning quarterbacks when passing while ahead. A 2.8 interception percentage for quarterbacks who are trailing is quite good. For the sake of comparison, Joe Montana and Steve Young had career interception percentages around 2.6 (1 interception every 38 passes). And the overall NFL average for interception percentage in 2005 was 3.1 (about 1 every 32).
There were 15 teams that had non-winning records in 2005, but because Buffalo had two quarterbacks (Losman and Holcomb) who started 8 games each and attempted almost exactly the same number of passes (228 vs. 230), I decided to include both of them. So I had 16 quarterbacks from non-winning teams to examine. And I found that they all combined for a 3.1 interception percentage (1 every 32) while ahead, but a 3.6 interception percentage (1 every 28) while behind.
It should be noted, however, that these combined interception percentages for non-winning quarterbacks were quite skewed by one man, Brett Favre. Because the Packers played with a lead only about 25% of the time in 2005, Favre amassed 401 pass attempts while behind, the most in the NFL. Only Kerry Collins of Oakland came close, with 376 pass attempts while behind. But while Collins threw 7 interceptions when passing while behind, Favre threw 23, a 5.7 interception percentage (1 every 17 or 18 passes). If we were able to remove Favre’s aberrational ahead-behind statistics from the non-winning quarterback group altogether, we find that those other 15 quarterbacks still combined for a 3.1 interception percentage when passing while ahead, but they now had a 3.1 interception percentage when passing while behind too. In other words, except for Favre, quarterbacks from non-winning teams were largely able to avoid interceptions at the same rate when they were ahead as when they were behind.
The explanation that quarterbacks who trail in games almost inevitably throw more interceptions than quarterbacks who don’t seems to imply that expectations should be lowered for quarterbacks who are "stuck" throwing passes when their team is trailing. I would think that avoiding mistakes and coming through when the team needs him most would be an appropriate way to evaluate the performance of a quarterback. Although it may be more difficult, it certainly isn’t impossible for a quarterback to avoid throwing a lot of interceptions when his team is trailing. The 2005 passing statistics for each team’s top quarterback seem to support that.
Theoretically, turnovers result in a 4 to 5 point swing
But back to interceptions and their role in the winning equation. Why is it that interceptions seem to have such an impact on which team wins and which team loses? After all, most interceptions aren’t returned. There were 507 interceptions thrown in 2005, with 47 of them (9.3%) returned for touchdowns. So only a small number of them might be said to have had an immediate impact.
I think it’s important, though, to look at the bigger picture of what happens when a turnover occurs. We shouldn’t just think of a turnover as a better chance to score points with the better field position that can often ensue. Instead, a turnover could also be counted as a time when the offense was prevented from scoring points. And, when compounded with the increased chances of scoring points off the turnovers, a turnover can have a significant impact on the final score.
I decided to look into this statistically. I found that the average scoring differential every time a turnover occurred in 2005 was 4 to 5 points. What I mean is that (1) when a team turns the ball over, it first loses nearly 2 points on average by missing out on a scoring opportunity on that drive, and (2) the team that collects the turnover scores nearly 3 points off each turnover on average. So, as a hypothetical example, if Team A turns the ball over 1 time, and Team B turns the ball over 3 times, the point differential averages out to about a 9 point advantage for Team A because that team had 2 fewer turnovers.
I determined this theoretical 4 to 5 point differential by using the average points scored per game per team (20.6) and dividing that by the number of offensive drives each team averaged per game (11.8) to find a points scored per offensive drive average of 1.75. In other words, on each drive foiled by a turnover, that’s an average of 1.75 points left off the scoreboard. I then counted up the total number of lost fumbles (388) and interceptions (507) in 2005 (895), and the number of points scored off of both interceptions and lost fumbles (2,550) either via return or on the subsequent offensive drive, to ultimately find a points scored off turnovers average of 2.85. Those two averages were added together to determine a 4.6 points scoring differential each time a team turns the ball over.
If we think about giving up a turnover as a potential 4 to 5 point swing in favor of the other team, avoiding turnovers can seem like a more critical component in the winning equation. Teams that turn the ball over a lot may be able to compensate if they have an excellent defense or a quick-strike offense, but most of the time a turnover will prove costly, especially when the game is otherwise close.
Touchdown passes, interceptions, winning, and quarterbacks
Conventional wisdom says that throwing touchdown passes is more important than avoiding interceptions in the winning equation. After all, a touchdown pass means an automatic 7 points. And of the 644 touchdown passes thrown in 2005, 435 of them (68.2%) went for 8 yards or more, which means that touchdown passes can often be big-time plays that can have a big-time impact.
On the other hand, an interception may or may not lead to points. An interception can sometimes function as a punt, especially if it’s 3rd-and-long and the pass attempted is long enough. Or, an interception can be harmless and even worth the risk if it occurs at the end of the half or at the end of the game on a so-called "Hail Mary" pass into the end zone.
And yet despite the readily apparent significance of throwing touchdown passes, or the sometimes inconsequential nature of throwing interceptions, it may be the case that avoiding interceptions may be just as, and possibly more important to winning than throwing touchdown passes. That’s what the passing statistics from the 2005 season seem to indicate.
So while quarterbacks who throw for a lot of touchdown passes can be considered quite valuable, quarterbacks who take care of the ball and avoid interceptions may be just as, or maybe more valuable. And quarterbacks who both throw a lot of touchdown passes and avoid interceptions may be the most valuable of all.
