USC Head Coach Pete Carroll was on the DKLA Show on ESPN 710 radio. During the segment he talked about one of his favorite topics and the most important statistic to him in terms of winning or losing a football game - turnovers. I'll post part of the interview below. For review however I'll also put up up some trend data related to turnovers and its effect on the Trojans. Simply stated 2006 was a very poor year for USC with regards to forcing turnovers compared to previous years. In his interview this deficit was highly "frustrating" for Pete Carroll to say the least.
In 2006 USC forced 22 turnovers or 20 fewer than in 2003 for example. In addition but not shown above USC committed 18 turnovers last season to tie for 15th in the nation in terms of fewest allowed. This made USC a net +4 in turnover margin for the entire year which was far below the levels established in 2002-2005 for example when they forced between 36 to 42 per season or around 3 per game and lead the nation in this category.
The effects of turnovers can not be ignored as Pete comments below in the interview. A forced turnover not only stops the opponent from driving down the field it also provides the USC offense prime field position. If there is no turnover then the opposition punts and USC starts 40 yards or so further away from the goal line. Over the course of a season these longer drives starting deep in your own territory are a big difference compared to starting inside the opponents 40 yard line for example. (Note: I calculated the average field position when USC forced a turnover for the past several years and it averaged right at the opponents 40 yard line each year.)
This field position is a big deal to USC football or any program. Starting from inside their own 20 yard not many teams march the length of the field during the season. USC went more than 80 yards for a score only 9 times in 2003 and again just 9 times in 2006 for example. From inside the opponents 40 yard line however USC suddenly jumped up to about a 60-70% probability of coming away with a score. Like the old saying in real estate goes, "location, location, location" or in this case in football "field position, field position, field position".
Head Coach Pete Carroll mentioned some trends in a couple of speeches over the off season to alumni groups and at coaching clinics. He wants most of his comments kept private that involve matters of team or personnel but I see no harm in analyzing some of his more general comments for interested fans. In particular Pete noted in many instances that the offense in 2006 struggled due to fewer turnovers and worse field position. However the probability of scoring as well as the average length traveled per drive in 2006 was not all that different than in previous years excluding the 2005 season. I decided to take a closer look and see if I could measure this effect in more precise terms.
For starters it is amazing that over the course of the season the 2006 Trojans were as likely to score as the 2003 edition at an aggregate level. Any individual game varies of course. Here is what I found breaking down the total number of drives for each of these two years. Note that 2003 had far more drives than 2006.
As noted this chart breaks down the entire season of drives by USC for both 2003 and 2006. Interestingly in both seasons the Trojans scored at just over 41.0% of the time they had the ball. 75% of those scores were touchdowns and 25% field goals. The other 59% of the time USC was either not scoring (and usually punting) or committing a turnover. The 2003 drive summary also highlights that the defense and special teams directly scored 10 times that year on fumble returns for TD and Int's returned for TD as well as a kick return. The 2006 defense and special teams only accomplished this twice. So as strange as it may sound to many fans Pete is right when he says that the offense was on average as likely to score in 2006 as it was in 2003 for example. Keep in mind however that the 2003 squad had 35 more drives than the 2006 squad to work with so of course it still scored more total points. The 2004 squad scored on roughly 45% of its drives for comparison and the 2005 team just over 57%.
I looked at some other data to get a different perspectives as well. I took the past few seasons of drive data and checked how many points per game the USC Trojans scored when starting on their own side of the field. Amazingly it was very close for the years 2002, 2003, 2004, and 2006. The following chart summarizes that overall trend.
In other words USC has only scored between 22.2 to 24.1 points the past few seasons when starting on its own side of the field. 2005 was of course an exception with the skilled players on offense.
I checked a couple of other dimensions as well out of curiosity about the offense such as 3rd down conversion efficiency, and average drive length. The results were again fairly similar. There was not a huge difference except for 2005 of course. 2006 it turns out had a higher 3rd down conversion percentage over the course of the season than the 2002-2004 years.
The 4th down conversion percentage for USC was even better in 2006 than in previous seasons. USC had an amazing 71.4% conversion on 4th downs in 2006. Most fans can only recall the attempts that failed I suspect due to the nature of the game and the way the human mind works.
I created an average drive length metric as well taking the seasons total yardage and dividing it by the total number of drives. The result was again very consistent with the exception of 2005. (Note: 2002-2004 had more yards per game of course averaging just over 445 yards per game versus 392 in 2006 due to a longer game and 9 more plays on average. If you adjust 2006 for 9 more plays it too averages about 440 yards per game).
These averages painted a strange picture and of course conceal the variation within the totals. Good games and bad games are contained with the data sets for each year. So I thought it would be more useful to show a histogram of yards gained in 2003 versus 2006 overall to see some more of the variation. This next chart breaks the entire seasons worth of drives down into categories of length e.g. less than 20 yards, 20-39 yards, 40-59 yards, etc.
.
As you can see the shape of the curve is roughly the same. The height of the curve is different since 2003 had more drives to work with. For example the teams moved the ball less than 20 yards the majority of the time in both years (45-47% of the time). They also both only went 80+ yards for scores 9 times in those two seasons (4.8-5.9% of the time). The rest of the drives are in the middle range. 2003 has more total drives due to the longer games in 2003 and the higher number of turnovers forced by the defense. But surprisingly the variation is not all that different between these two years at least.
So the million dollar question is if the 2006 offense was not as bad as often criticized by fans and the media where are all the missing points? Well I had a clue from Pete Carroll's comments regarding turnovers, field position, and the probability of scoring. None of the above charts reflects field position and turnovers in conjunction. I decided to plot scoring points in 2003 and 2006 when USC started at or inside the opponents 40 yard line after forcing a turnover. Here is what that chart looked like. Suddenly I could see where all the missing points were coming from...
There was a huge drop off in short field scores for USC after turnovers in 2006. The only game where USC forced a high number of turnovers in 2006 was the first game against Arkansas as Pete notes in his interview. Not too surprisingly the offense put up 50 total points that day. In 2003 the offense scored a total of 184 points on short field drives starting inside the opponents 40 yard line. In 2006 that statistic declined to 90 points. The difference of 184-90 = 94 points or 94 / 13 games = 7.2 points per game. The USC offense has averaged 37 points per game during the Pete Carroll era. The 2006 squad put up only 30.46 points per game. Tack on another 7.2 points on short field drives and suddenly the 2006 team is right back in the general ball park of where other teams have been (excluding 2005 of course).
On average this explains a lot about the perceived weakness of the offense in 2006. The offense did indeed sputter at times due to a weak running game that only ranked 68th in the NCAA. The loss of the fullback hurt the run game as well as pass protection and the short passing game. Just as equally harmful however was the shortfall in turnovers that concerned Pete Carroll so much last season. Due to the file size I could not post the entire Pete Carroll interview on the DKLA show but here is the two minute segment where Pete describes the concerning drop off in turnovers in 2006. Click here to access the 20MB 2 minute file.
In 2007 this will be something that I will track even closer than normal. If the defense produces as expected and the offense suddenly starts putting up tons of points though expect to see the usual media and fan board chatter about the play calling suddenly looking much improved. It may in reality however just be Pete's favorite statistic - turnovers and short field scoring opportunities.