Has anyone spent any time on injury probability? (1 Viewer)

[chris1969]

I'm thinking about working up some equations to help predict injuries. I know starting out that you can't predict an injury in any particular season, but I think a long term projection may be valuable for a few reasons.

1) Possible becomes more probable over time. A person that is rated as more of an injury risk and doesn't get injured in a particular season, should have a slightly higher injury risk the next season.

2) Having an understanding of the major factors that predict injury will help in a tough decision between 2 players.

3) It will give me something to do until training camp starts.

Some info I'm considering...

First I figure that there should be some overall base for injuries just for playing in the NFL. Even if you sit the bench, you still practice so your base risk would be equal to a starter. Your risk would increase with playing time in games. Your risk would also increase or decrease by position. A kicker or punter would probably have less risk than a RB. Your size could increase your injury risk if it is not optimal. Certain injuries increase your risk of having more injuries, like concussions or knee injuries. I'm not sure, but I would suspect that your team record could effect your injury risk. I would assume if the players around you are not as good it would increase your injury risk. Speed for your position? maybe. Age? probably. Touches? yep.

If you guys can think of more let me know.

Also what's really important is if any of you know a source of injury stats. I would assume I'd need at least 5 years of data to get anything close to accurate and probably closer to 10 to 20 years.

Thanks for any help you can give.

 

[Jeff Pasquino]

Think of what makes sense as far as logical relationships, like:

Proximity to the ball should correlate directly with injuries (if you have the ball, people are coming at you, and if you're close by, you're in the line of fire).

Historical likelihood of injury (if you got hurt before, likelihood goes up).

Age would also be a factor - older would likely get injured easier (but again may know how to avoid injury as well, but more likely the odds catch up with you).

Do you get hit a lot? That matters.

Actually, think about this like a car accident - is it because you're a bad driver, are you around bad drivers, or are you just on the road a lot?

You need an actuary.

 

[chris1969]

Think of what makes sense as far as logical relationships, like:

Proximity to the ball should correlate directly with injuries (if you have the ball, people are coming at you, and if you're close by, you're in the line of fire).

Historical likelihood of injury (if you got hurt before, likelihood goes up).

Age would also be a factor - older would likely get injured easier (but again may know how to avoid injury as well, but more likely the odds catch up with you).

Do you get hit a lot? That matters.

Actually, think about this like a car accident - is it because you're a bad driver, are you around bad drivers, or are you just on the road a lot?

You need an actuary.

Funny you said that, I was thinking about using some risk analysis equations.

 

[FantasyTrader]

"equations to help predict injuries"

I don't believe what you are attempting is possible with any degree of accuracy. There are far too any factors and NONE of them point to direct causation.

You could look historically and develop correlated factors but not causation (predicting). You're probably better off developing a profile of factors leading to least % of injuries to the profile leading to the highest %.

The other problem is you have no idea how to weight the importance of factors in relationship to each other.

I admire you for taking a crack at it though - interested to read what you come up with.

 

[chris1969]

Actually any risk can be measured by probability. While I can't predict if you'll get injured, I can predict your probability that you'll get injured. Your insurance company can't predict an accident, (They'd cancel your policy if the could), but they can figure out your probability of having one.

 

[The Shizzle]

You'd have to use carry #'s as a major factor.

The year before Shaun Alexander hurt his foot he received a ton of carries.

Just as an instant guess, Larry Johnson is a guy that makes me nervous. He racked up a record amount of carries in 2006. With Croyle/Huard as the QB and no big Willie at the line, he will see smaller holes and more guys in the box ready to take him down. The combo of those two makes LJ a guy I could see as getting banged up this year.

 

[chris1969]

I already mentioned touches. One thing on the RB carry thing though. Both LT and LJ seem to do a good job of avoiding big hits by running out of bounds when needed. I'm sure some receivers do the same thing. I might have to look at that as well.

 

[Jeff Pasquino]

Think of what makes sense as far as logical relationships, like:

Proximity to the ball should correlate directly with injuries (if you have the ball, people are coming at you, and if you're close by, you're in the line of fire).

Historical likelihood of injury (if you got hurt before, likelihood goes up).

Age would also be a factor - older would likely get injured easier (but again may know how to avoid injury as well, but more likely the odds catch up with you).

Do you get hit a lot? That matters.

Actually, think about this like a car accident - is it because you're a bad driver, are you around bad drivers, or are you just on the road a lot?

You need an actuary.

Funny you said that, I was thinking about using some risk analysis equations.

Mrs. Pasquino used to be one.

 

[chris1969]

Think of what makes sense as far as logical relationships, like:

Proximity to the ball should correlate directly with injuries (if you have the ball, people are coming at you, and if you're close by, you're in the line of fire).

Historical likelihood of injury (if you got hurt before, likelihood goes up).

Age would also be a factor - older would likely get injured easier (but again may know how to avoid injury as well, but more likely the odds catch up with you).

Do you get hit a lot? That matters.

Actually, think about this like a car accident - is it because you're a bad driver, are you around bad drivers, or are you just on the road a lot?

You need an actuary.

Funny you said that, I was thinking about using some risk analysis equations.

Mrs. Pasquino used to be one.

Maybe Mrs. P could hook me up with a few pointers? It may be a moot point though, I'm not finding great luck looking for injury statistics.

 

[Jeff Pasquino]

Think of what makes sense as far as logical relationships, like:

Proximity to the ball should correlate directly with injuries (if you have the ball, people are coming at you, and if you're close by, you're in the line of fire).

Historical likelihood of injury (if you got hurt before, likelihood goes up).

Age would also be a factor - older would likely get injured easier (but again may know how to avoid injury as well, but more likely the odds catch up with you).

Do you get hit a lot? That matters.

Actually, think about this like a car accident - is it because you're a bad driver, are you around bad drivers, or are you just on the road a lot?

You need an actuary.

Funny you said that, I was thinking about using some risk analysis equations.

Mrs. Pasquino used to be one.

Maybe Mrs. P could hook me up with a few pointers? It may be a moot point though, I'm not finding great luck looking for injury statistics.

Just because it is hard doesn't mean it isn't worthwhile.

 

[Dirty Weasel]

You may be able to find an injury probability guru if you have a friend in the insurance business. All insurances companies seem to use some sort of mathematical formula to determine risk. You'd just have to get that formula and tailor it to each NFL position. Good luck in your quest.

 

[chris1969]

Jeff maybe you could ask Mrs P this...

I first thought that I could take 32 teams x 53 man rosters and divide that into the number of injured players in a year and get a very simple baseline to start with. The problem is that NFL rosters are fluid. You don't have the same 53 guys the entire year. Then I thought, well if I look at it weekly during the season with the rosters as they were at game time vs the number of injuries that week then that might be more accurate. A second thought on that is that including the 1 or 2 week wonders may artificially lower the base risk because your odds of getting hurt in a 1 week period are much less than in a 1 year period. Since the only guys of interest are guys that should be on the roster all year unless they get hurt, this is important. Then I thought why not do both and average the 2! Take my 32 x 53 number and divide by 17 and add that to the one week wonder numbers, (making sure to include the roster before the bye week as the roster during the bye week), then divide by 2. The first should be artificially high because it doesn't have enough players. The second should be artificially low because it includes low risk players. The average should be just about right though. Ask her if I'm missing something on this?

 

[chris1969]

You may be able to find an injury probability guru if you have a friend in the insurance business. All insurances companies seem to use some sort of mathematical formula to determine risk. You'd just have to get that formula and tailor it to each NFL position. Good luck in your quest.

I actually work for a car insurance company, so I plan on getting some help, but there a people that specialize in this specific field that make big bucks from big corporations. As far as all insurance companies using the same formula, there wouldn't be a need for actuaries if that were the case. Most insurance companies use different formulas based on the same basic formulas. That's why it pays to shop for your insurance. For instance if you have bad credit, (yes your credit does effect your insurance), but a good driving history, you'll rate really good with one company and really bad with another.