## NYT: Game Probabilities - Week 8

Weekly game probabilities are available now at the nytimes.com Fifth Down. This week I also look at how the Cowboys' downgrade at quarterback might affect their chances against Jacksonville.

### 7 Responses to “NYT: Game Probabilities - Week 8”

1. anthony says:

Brian,

I'm intrigued by the Romo calculation.

Especially because it seems to be at odds with one of the findings from Stubling On Wins

"For pitchers in baseball, quarterbacks in football, and goalies in hockey; wins are directly assigned to the individual. Essentially, if you play at one of these positions and your team wins, then you are given a win. If your team loses, you are given a loss.

The analysis of consistency, though, suggests that this practice is misguided. Remember, the less consistent a player, the less his performance is about his own abilities. So who are the least consistent players in North American sports? It’s the very players who are traditionally assigned wins and losses. Pitchers, quarterbacks, and goalies are believed to be the players most responsible for winning. But what these players do is heavily dependent on their teammates...Quarterbacks can’t block for themselves and generally don’t catch their own passes....Every play in baseball and football begins with the pitcher and the quarterback ... Given how frequently these players are the focus of the action, it’s natural to assume these players are the primary determinants of outcomes. The study of consistency, though, tells us that what people assume about pitchers, quarterbacks, and goalies is incorrect. These players are not solely responsible for team success. These players are not even solely responsible for many of the numbers people attach to their performance."

Do you have any insight into whether Kitna's performance is likely to be independent of the team?

2. Buzz says:

Kitna is also playing with an all star cast of wr's and a solid running game (although a very poor o-line). My guess is that this is the best supporting cast that he has had which would indicate that his stats should be higher than his career average. However, he is also older which would suggest a penalty.

It will be very interesting to see what his efficiency stats end up at but my first guess is that they will be on the high end of his career stats - maybe not so differently than when Cassel took over for Brady with so many talented players around him.

3. TBD says:

ditto what I said last week: if you have GB as a 59% favorite to win this week @ the NYJ, you have systematic error. A pure quant model can't possible be correct, you need qualitative inputs.

4. Dan says:

Brian, you have Denver ranked higher than San Fran in the efficiency rankings, yet you have the Niners as favorites this week in their neutral-site game. Why is this?

5. Brian Burke says:

Dan-Thanks for the catch. In the 4 years I've been doing this, I don't think I have remembered the neutral-site game once!

Denver would likely have a 54-46 advantage or so.

6. Anonymous says:

Next week (week 9) is another neutral site game. The Bears and Bills will play in Toronto.

7. Jim Glass says:

Fifth down column:

"No single position affects a team’s fortunes more than quarterback."

Anthony said:

"I'm intrigued by the Romo calculation. Especially because it seems to be at odds with one of the findings from Stubling On Wins...

"Remember, the less consistent a player, the less his performance is about his own abilities. So who are the least consistent players in North American sports? ... quarterbacks"

When a team has 53 players there's no logical contradicion in saying both: the QB is by far the most important single player on the team, and he is a lot less important than most fans and commentators believe.

The inconsistency issue is spot on. Stark examples abound. Start with Earl Morrall, career journeyman backup who got promoted via starter's injury to QB the '68 Colts and became NFL MVP by leading them to a 15-2 championship season ... then got cut out of the league, was picked up for \$100 by the Fins, got promoted by starter's injury again, and became AFC Player of the Year throwing for a great 9 yards per attempt leading them through most of their 17-0 season.

Skip ahead to Matt Cassell, after not starting a game since high school while getting thrown in for the Pats he was a top 10 QB by rating, good enough to get a frachise QB contract in KC, where he finished 30th in y/a the next year. That was after PFR.com did an empirical study finding starting QBs are worth only a bit over 2 pts/game more than backups, and controversially predicting that if Brady went down it would cost the Pats only one or two games. Then Brady went down and the Pats won one game less than the Vegas line on them before his injury.

But QBs still get all the credit, even from stat mavens who know better. PFR.com itself just did a post: 'QBs who have won the Super Bowl. Disclaimer: teams win Super Bowls, not QBs. Now back to QBs who have won the Super Bowl...'

"Do you have any insight into whether Kitna's performance is likely to be independent of the team? "

It can't be, but how dependent will it be? Nobody's been able to quantify it yet which is why QBs still get all the credit. But a plausible guestimate might come from using the baseball concept of win shares.

Say the difference between teams is in their best 36 players (the rest are replacement level). Give the 22 starters each one "share" of team results and the next 14 situation players/subs each a half share, 29 shares total. Say the offense has 4 of those half-share players plus its 11 starters.

If the average starting QB is worth 3 other average starting players (rather more than NFL payroll data indicate) then the QB produces 3/31 or 9.7% of team results, and 3/15 or 20% of the O's results. If the QB is worth 4 other players he produces 4/32 or 12.5% of team results and 4/16 or 25% of the O's results. If he is worth 5 other starting players, he produces 5/33 or 15% of team results and 5/17 or 29% of the O's results.

Even *that* QB's 15% of team results is probably only about as much as that of the #3 player on a basketball team. ("Let's go out and get a new 3rd-best player to lead us to the NBA Championship!") Yet he is worth *five* other starting players! The most important player on the team by far -- yet far less important than most think. And most of "his" stats are produced by his teammates. Maybe 2/3rds.

With this caveat, that may apply to Kitna as I remember him. The above applies to competent QBs. Really *bad* QBs, because the ball runs through them on every play, can destroy an O singlehandedly, all by themselves.