Sunday, August 05, 2007

Less is MOE | by Michael

Don't tell Maryland's Ralph Friedgen that "Defense wins championships." The chalkboard Xs and Os guru might simply laugh. Over the span of his lengthy coaching career, Friedgen has invented a formula, based entirely on offensive statistics, that he and his staff argue directly correlates to winning football games. Last fall, as his team was making a run toward the ACC Championship game, he sat down to discuss this winning formula with the Washington Post.

Fridge calls his formula Major Offensive Errors (MOE). To calculate it, you count up:

  • sacks
  • offensive penalties
  • fumbles
  • interceptions
  • dropped passes
Then, count the number of total plays you had in a game. (Don't forget to add in the number of no-play penalties like false starts, illegal shifts, and delays of game to the total.) Finally, divide the errors by the total plays. What you end up with is a rating of your team's efficiency -- or inefficiency -- that Friedgen relies upon heavily.
"That's winning it for us; there's no doubt about it," Friedgen said. "It has at every level I've coached at. It's something I believe in very strongly, and it's one of the reasons we won our first year here."

Since his days as offensive coordinator at Georgia Tech, Friedgen has tinkered with various new statistics to measure errors. He came up with a similar stat while with the Yellow Jackets.

"It was kind of along the same lines as that, but we weren't smart enough to divide it into the number of plays," Friedgen said. "The year we won the national championship, we didn't make many errors either. It goes a long way to winning football games."
According to Friedgen, the team goal is to keep MOE under 12%. Throughout his career, Friedgen's teams have emerged victorious 95% of the time their MOE was under the target. (At Maryland, it's been closer to 90%.)

Whether or not Charlie Weis and his staff calculate MOE or a similar formula is unknown, but according to the data we collected over the 2005 and 2006 seasons, MOE appears to be a reliable predictor of success. In twenty-five games during the last two years, Notre Dame is 13-0 when the offense has kept its MOE under 12%.

The charts below capture the game-by-game MOE for the 2005 and 2006 seasons. (Key: Plays - number of plays; Int - interceptions; Fum - fumbles; Drp - Dropped passes; Sck - sacks given up; Pnl - Penalties. The dotted red line is the Fridge's 12% threshold.)

2005 Plays Int Fum Drp Sck Pnl MOE W-L
Pitt 68 1 0 3 0 3 10.3% W
UM 72 0 0 5 4 2 15.3% W
MSU 105 1 1 3 2 8 14.3% L
Wash 82 0 0 2 0 3 6.1% W
PU 72 1 0 1 0 1 4.2% W
USC 89 1 1 4 3 2 12.4% L
BYU 60 0 1 0 1 7 15.0% W
Tenn 64 0 1 0 4 3 12.5% W
Navy 70 1 0 2 0 3 8.6% W
Syr 70 0 0 1 1 1 4.3% W
Stan 82 2 0 2 1 1 7.3% W
OSU 74 0 0 5 5 1 14.9% L
TOTAL 908 5 5 28 21 35 10.4%

2006 Plays Int Fum Drp Sck Pnl MOE W-L
GT 81 0 0 2 2 8 14.8% W
PSU 71 0 0 2 3 3 11.3% W
UM 65 3 1 6 2 3 23.1% L
MSU 60 1 2 1 2 5 18.3% W
PU 80 0 0 2 5 1 10.0% W
Stan 73 0 0 2 1 0 4.1% W
UCLA 83 0 2 3 5 3 15.7% W
Navy 55 0 0 0 1 3 7.3% W
UNC 66 0 0 3 3 4 15.2% W
AFA 44 0 0 2 2 1 11.4% W
Army 64 0 0 1 1 2 6.3% W
USC 81 0 1 4 3 5 16.0% L
LSU 64 2 0 4 1 2 14.1% L
TOTAL 887 6 6 32 31 40 13.0%

One insight that immediately jumps out is that the highest MOE in 2005 (UM, 15.3%) would have placed fifth in 2006, and just about tied for sixth. It always seemed that the 2005 offense was more efficient than last year, and comparing MOEs for the last two seasons confirms that.

And yet, given all the returning talent for '06, why did the offense erode? As the chart illustrates, an increase in drops, penalties and sacks caused the MOE to vault from 10% to 13%, a serious decline in performance. Who were the culprits? Two talented offensive linemen subsequently drafted to the NFL were responsible for 37.5% of all offensive penalties. Veteran wide receiver Rhema McKnight held the unfortunate distinction of being accountable for 41% of the team's dropped passes. Despite returning four starters at OL, somehow Quinn was sacked at least twice in every game save Army, Navy, Stanford, and LSU. Also, keep in mind that MOE does not count poor throws to wide open receivers or route miscommunications between receiver and passer.

Can the Irish keep their MOE below 12% in 2007? With looming question marks at quarterback, running back, offensive line, and wide receiver, it is difficult to project a more efficient offense than last year's (as inefficient as it was at times). The key questions for a low MOE:
  • Will the quarterback throw as few interceptions as Quinn threw in '05 and '06?
  • Will the tailbacks fumble as rarely as Walker fumbled in '05 and '06?
  • Will a young and inexperienced offensive line commit fewer penalties?
  • Will that same line protect the quarterback better?
  • Will the receivers drop fewer passes?
If Maryland could pull it off with appreciably less talent than the Irish, then Charlie certainly can. Friedgen adjusted by placing more emphasis on ball security, asking his quarterback to make fewer deep throws, and making similar risk-averse offensive adjustments. Charlie can immediately put a dent in dropped passes and sacks with an emphasis on power running. This should open up the short and medium-range play action passes, which were non-existent last year due to the finesse running game. Meanwhile, greater competition for playing time along the offensive line should result in fewer penalties; those who are making mistakes will have to sit on the bench while others take their place. This luxury was inconceivable in '05 and '06, but back-to-back strong recruiting classes have improved the depth going forward.

Back to Maryland. The Terps were 8-2 overall and 5-1 within the ACC when the Friedgen article was published. They lost their next two games and failed to reach the ACC Championship. Did MOE play a role in that?

Against Boston College, to whom the Terps lost 31-16, they committed 9 mental errors out of 85 plays for a MOE of 10.5%. However, two of those mental errors, a fumble and an interception, were returned for touchdowns. The next week they traveled to North Carolina State, committed 19 errors out of 67 plays for a MOE of 25.6%, and lost 20-14. Keep in mind that box scores don't track dropped passes, so it's quite possible that Maryland's MOE against Boston College actually broke the 12% threshold.

A Friedgen quote from the Washington Post story best explains why Maryland lost these two games and sums up the concept of MOE.
"If you don't beat yourselves, you give the opportunity to the other team to beat themselves."