[a4u2fear]

So last nights game between Baltimore and Miami presents. A good question

how to adjust Bal and Mia defense and offense for the three defensive TDS

a blind bat would give Baltimore offense 40 and D 0

Splitting it up you have Bal O 20 and Bal D 0-20=-20

the -20 is obviously wrong, but it will have to be negative in this scenario, but since D TDS are not common, how do you adjust the 20 pts?

i remember seeing an article once that fumbles were something like 80% random and when that capper calculated the fumbles expected he multiplied it by that probability.

Thoughts?

[Bsims]

That's a difficult question. Handicapping football by statistics is particularly difficult because of issues like the one you mentioned. As I've mentioned before, when I first did regression analysis 50 years ago on NFL data, I found the most significant factors in determining a team's score was the number of opponents turnovers, followed by their own turnovers. And this is probably the most difficult statistic to project.

Also consider that all turnovers are not created equally. For instance a hail mary pass at the end of the first half that is intercepted appears as a turnover in the statistics. But so does a fumble lost on the 5 yard line after a 6 minute 80 yard drive. Clearly these two turnovers are not equal in their impact on the final score.

One thing I've considered doing is to develop a regression equation with turnovers and other stats included. Then use some technique to project the other stats but not turnovers. For turnovers, I would use some other technique to assign a probability that each team has 0, 1, 2, up to 5 turnovers. Then I would use the regression equation with each possible combination of potential turnovers and the probability that this score would occur. Now I would have 36 possible scores with different probabilities for each. This range of projected scores could then be used with the lines and odds to compute expected returns for every possible wager.

Sounds like fun, and probably a good way of losing your ass.

[Waterstpub87]

So last nights game between Baltimore and Miami presents. A good question

how to adjust Bal and Mia defense and offense for the three defensive TDS

a blind bat would give Baltimore offense 40 and D 0

Splitting it up you have Bal O 20 and Bal D 0-20=-20

the -20 is obviously wrong, but it will have to be negative in this scenario, but since D TDS are not common, how do you adjust the 20 pts?

i remember seeing an article once that fumbles were something like 80% random and when that capper calculated the fumbles expected he multiplied it by that probability.

Thoughts?

You have the break the fumbles into play type in order to figure them out, these are fumble % based on play type I researched based on last 2 years data:

Sack: .145
Rush: .016
Completed Pass .009

Sacks dominate as the cause of fumbles. Therefore, if a team is going to get more sacks then their opponent, they should expect a higher fumble rate. Being that they are a small %, they won't occur often.

I wouldn't use touchdowns for rating anything. Yards/Interceptions/Completion are likely more predictive. Touchdowns are going to be pretty noisy.

[evo34]

You have the break the fumbles into play type in order to figure them out, these are fumble % based on play type I researched based on last 2 years data:

Sack: .145
Rush: .016
Completed Pass .009

Sacks dominate as the cause of fumbles. Therefore, if a team is going to get more sacks then their opponent, they should expect a higher fumble rate. Being that they are a small %, they won't occur often.

I wouldn't use touchdowns for rating anything. Yards/Interceptions/Completion are likely more predictive. Touchdowns are going to be pretty noisy.

Good info. Definitely sacks are a great predictor relative to other metrics. That's what some people fail to realize about Kaepernick. His sack rate was historically high (only David Carr, 1st overall pick bust from 2002, had a higher rate). That, more than anything else, is what scared away most NFL teams. If possible, they want low-volatility mediocrity out of a backup.

It's a side note, but an important and often overlooked consideration when evaluating the quality and riskiness of any QB.

[pokerdevil]

Books regularly run a prop "Will there be a D or ST touchdown?" and it's typically priced at roughly

Yes +180
No -220

Obviously you have to account for the remote but real possibility of 2+ D/ST touchdowns. But there's a starting point for you.

[semibluff]

Books regularly run a prop "Will there be a D or ST touchdown?" and it's typically priced at roughly

Yes +180
No -220

Obviously you have to account for the remote but real possibility of 2+ D/ST touchdowns. But there's a starting point for you.

Those numbers sound way off. I don't want to push a rival site but if you click on their comparison odds for NFL games and look at Anytime TD scorer D/ST normally shows at around +350

Later edit: Sorry. You're quoting for both teams combined and i'm looking at odds for 1 specific team.

[gdon44]

Good info. Definitely sacks are a great predictor relative to other metrics. That's what some people fail to realize about Kaepernick. His sack rate was historically high (only David Carr, 1st overall pick bust from 2002, had a higher rate). That, more than anything else, is what scared away most NFL teams. If possible, they want low-volatility mediocrity out of a backup.

It's a side note, but an important and often overlooked consideration when evaluating the quality and riskiness of any QB.

Sacks more a stat relating to OL than QB no? Sure QBs can hang on to the ball too long but QBs on bad teams tend to get sacked more because they have a bad OL in front of them. (see: Manning, Peyton)

[JacketFan81]

I've tried many times... practically impossible.

[Waterstpub87]

Sacks more a stat relating to OL than QB no? Sure QBs can hang on to the ball too long but QBs on bad teams tend to get sacked more because they have a bad OL in front of them. (see: Manning, Peyton)

Some places will tell you to consider sacks vs. team average. A theoretical example would be Kaepernick vs that other dude who played for minnesota before san fran that i forget the name of

Kaepernick was more likely to run, when he gets hit behind the line it is counted as a sack. Other dude doesnt run, and gets rid of the ball quicker, so doesn't get sacked as much.

The problem with this is: super small sample size for something that is low frequency. Additionally, most teams have the same QB most of the time.

Likely 80%+ of sacks are driven by OL

[Auto Donk]

every time I see the title to this thread I think "damn, how promiscuous do nfl players must be if gamblers have to take into account how many sexually transmitted diseases ("S T D's) they have" when deciding on which team to bet......

damn special teams....