Pro Football Reference Blog

Chase, for the last few years, you ought to use the exact data. I understand that, to keep things "equal", you used approx. time. But with the 99 Rams, for example, you could and should use the exact time. With the last 7 SB's on your list, for 5 of them you can give the EXACT time (and the '88 one prob. too.). To me, for these teams, this overstates their "knockout" time. Because if you were right, all of the losing teams would have had time to comeback. In reality, only the Titans did--and they almost made it. The other 5 never had a realistic chance to get a game-tying/game-winning score--that's where the "knockout" occurred.
Also, I think there should be a factor for losses in the first table. Nobody would say that the '54 Browns were as dominant as the '51 version--for the difference in record, if nothing else.
3rd, I don't think you can say that there was a "knockout" if the margin of the final score was 7 points or less (unless the losing team scored "garbage" points with the game out of reach). To stay with boxing terms, you don't get credit for a knockout, even if you win by unanimous decision--it's just a win. Whereas the Bucs don't get enough credit for the SB win against the Raiders--IIRC, they were ahead by a comfortable margin most of the game, even though the Raiders scored a decent number of total points.
Maybe you could use "exact time" for anything recent for which you have the time of scores, and "approximate time" for anything before that.

A quick question about this, how was overtime handled? Would that be a negative knockout time, or just at 0?

Next week, we'll look at the 2010 results.

Will that be a guest post by the Psychic Friends Network? hehe sorry

The main flaw in AKO is that it doesn't take any account of margin of victory. As a result, it overvalues great defensive performances, especially shutouts. Imagine that two teams, A and B, both score a field goal in the first quarter. Both defenses will go on to record shutouts, but Team A adds two TDs before halftime, while Team B nervously holds onto a 3-point lead. Those teams will have the same AKO, but the game almost certainly did not unfold in remotely the same way.

If you could devise a stat that combined AKO with, say, final margin of victory, then you'd be closer to "measuring and ranking dominance".

I second the recommendation of using 7 pts as a minimum. I'd also like to see the negative time subtracted off.

Excellent plan, though. I like this.

If one wanted to incorporate margin into this scheme, one could simply calculate the margin for every moment in the game (using the fudged time of score), then roll that into a single number for the game via some sort of average. For instance, take three games that end 28-14 with these quarter-by-quarter scores...
.
1) 28-0, 28-0, 28-0, 28-14
Avg margin: 22.75. Clearly a blowout with some garbage points
.
2) 7-14, 14-14, 21-14, 28-14
Avg margin: 1.75. Winner spent 1 quarter losing and only took a commanding lead with 7.5 minutes left. It was anybody's game for the other 52.5 minutes
.
3) 0-7, 0-14, 14-14, 28-14
Avg margin: -3.5. The winning team spent over 30 minutes behind and was only up for the last 10 minutes. Sure it's a win, but...

One could get fancier by weighting the time or weighting the time or differential, or both.

A possible way to use the Win Probability graphs would be to find the average Win % for each team during the game (for each point on the graph if possible). The further away the average win % was from 50% the more dominant (or dominated) the team was during that game.

Example:

Game 1: NE vs TEN 10/18/2009 59-0
Game 2: CIN vs CLV 10/4/2009 - Final Score: 23-20 OT

Game 1:
TEN Average Win % per 1/5 Qrt - 9%
NE Average Win % per 1/5 Qrt - 91%

Game 2:
CIN Average Win % per 1/5 Qrt - 59%
CLV Average Win % per 1/5 Qrt - 41%