Turnover Analysis Updated Thru ND

Submitted by Enjoy Life on September 14th, 2011 at 5:17 PM

First a few notes:

(1) Turnover Statistics for the WMU Game: Although the NCAA is not including any statistics from the WMU game, I will include TOs from the game. Excluding TOs for any specific game would significantly distort the analysis for the year since the NCAA National Rankings are based on total TO data and not per game TO data! Therefore, my data will reflect U-M with +3 TOM better than the NCAA data and the National Ranking for U-M will reflect this +3 TOM also.

(2) Situational Analysis: Instead of using an average value for each TO, Expected Points are used to enhance the analysis.

Now, on to the good stuff.

Whoooo Baby!: Oh what a game. Mid September Twenty Eleven. What a very special game for me. Cause I remember, what a night……

Synopsis for Turnovers: WTF! The game had a total of 8 turnovers. Michigan lost 3 but gained 5 from ND, leaving M with a TOM of +2 for the game and +5 for the year. (This does include the meaningless TO on the last play of the game. Of course, no impact was included for this TO.)

For the second week in a row, TOs were a significant reason M won the game. Eliminating the meaningless TO at the end of the game, M had a TOM of just +1. But, the net result was a whopping advantage of 7.5 Expected Points. There should be no doubt – without the TOs, M does not win this game.

Actual Score 35 31
Adjust for TOs -7.5 0.0
Score Without TOs 27.5 31.0

(See the Section on Gory Details below for how the adjustment for Expected Points (EP) is calculated.)

National Rankings: Remember the chart and table below include the WMU game and will NOT be the same as the (incorrect) NCAA Rankings. It should come as no surprise that DRob is not throwing the ball consistently. Interceptions are ranked #100 Nationally. Overall, M is Numero Uno in turnover margin (be still my heart!).

  TO Lost TO Gained
Total 1 0 3 3 7 5 3 8 5
M Natl Rank N/A 1 100 50 N/A 2 11 2 1

Synopsis for "Un-Official" Turnovers: Missed field goals, blocked punts, being stopped on fourth down, etc. are not officially counted as turnovers even though the impact is often the same as a turnover. None occurred in the game.

The Gory Details

Expected Point (EP) Analysis: Last year I used an average of 5 points for each TO (based on sites such as Football Outsiders). This always seemed a little weird to me since TOs are not created equal. I am now using the concept of Expected Points (EP) the the Mathlete and others have explained in detail. Basically, the probability of scoring depends on the yard line that the offense is at (which seems fairly obvious in retrospect). Therefore, the impact of a TO also depends on the yard line where the TO is lost and the yard line where the TO is gained.

Here are the details for the ND game.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Qtr Time Down LOS TO Lost By EP Lost TO Gain By Spot EP Gain EP Gain W/O TO Net EP Gain Total EP Diff
1 0:15 2-8 M35 M -2.0 ND M39 -3.4 -1.8 -1.6 -3.6
2 11:06 3-9 V36 ND 2.1 M V45 3.1 1.8 1.3 3.4
2 5:30 1-10 M18 ND 4.6 M M18 1.3 1.8 -0.5 4.1
3 7:36 1-10 M30 ND 3.8 M M29 1.8 1.0 0.8 4.6
3 6:08 1-10 M43 M -2.4 ND V29 -1.8 -1.5 -0.3 -2.7
4 6:08 1-G M07 ND 5.7 M M09 1.0 1.8 -0.8 4.9
4 4:23 1-10 V30 M -3.8 ND TB -1.4 -2.0 0.6 -3.2
4 0:02 KO V24 ND 0.0 M V14 0.0 0.0 0.0 0.0
TOT 7.5

EP Differential: + is M advantage, – is M disadvantage

The analysis is a bit tricky because: (A) the TO directly results in lost EP for the offense but (B) only modifies the EP for the team gaining the TO because the team gaining the TO would have gotten another possession even without the TO (due to a punt, KO after a TD, KO after a field goal, etc.). The Net EP Gain must take into account the potential EP gain without the TO. The EP gain without the turnover is based on where the field position would have been for the next possession if the TO had not occurred.

For example, in Line 1 of the Table above.

(1) The EP for a team at their own 30 yard line = 2.0. As soon as DRob threw the interception, M lost these EP (Columns 4, 5, & 6 in yellow above).

(2) ND took over at the M39, which has an EP of 3.4 (Columns 7, 8, & 9 in green Cells above).

(3) However, if the TO had not occurred, ND would have eventually had another possession. What yard line would the possession have started? Uh, no one knows! For this analysis, I calculate the yard line for the next possession based on a net punt of 36 yards from either: (A) the original line of scrimmage (LOS) for interceptions or (B) the spot of the fumble.

If M had not thrown the interception and had punted the ball a net 36 yards from the M35, ND would have taken possession at their own 29 with an EP = 1.8 (Column 10 in purple Cell above).

(4) The net gain for intercepting the ball is 3.4 EP minus 1.8 EP = 1.6 EP (Column 11 in white Cell above).

(5) The total EP for this TO = EP Lost by M + Net EP Gained by ND = 2.0 + 1.6 = 3.6 EP.

Therefore, the interception resulted in 3.6 fewer expected points for M.

Details for Turnovers: Here is overall summary for all games by player (data in yellow was affected by this week's game).

  TO Lost TO Gained
Player FMB FL INT Player Force Recover INT
Denard 3 Herron 1 1
Hopkins 1 0 Kovacs 1 1
Gallon 1 0 RVB 2
Campbell 1
Floyd, JT 1
Robinson, M 1
Robinson, T 1
Total 2 0 3 2 5 3

In Case You Were Wondering About WMU: Last week I made a mistake that I have made many times before – I trusted the "experts". I decided not to include the WMU statistics since the NCAA did not consider the game to be "official". As stated above, the NCAA is simply and obviously wrong in their decision. The distortion caused by excluding the WMU game is far greater than the distortion (if, in fact, there is any) of including the WMU game.

Actual Score 34 10
Adjust for TOs -22.7 0
Score W/O TOs 11.3 10.0

After adjusting for the TOs, M had a slim 1.3 EP lead when the game was called due to weather. This is close enough to conclude that TOs were a significant reason that M  also won the WMU game.


Enjoy Life

September 19th, 2011 at 2:48 PM ^

I've included a summary chart in this week's diary on Turnovers.

The chart is in 10 yard increments, the expected points between the increments is linear so I didn't put in the entire chart.

Here is a similar chart from the folks at Football Outsiders (note it is in the opposite order of yard line).