Football 101: QB Rating System

[JD_KaPow]

The great book "Hidden Game of Football" looked at a lot of these things in depth. Basically, the passer rating (not, as others have said, QB rating) overvalues TD passes and undervalues INTs (meaning the formula doesn't subtract enough for them), among other problems. It's not a great measure.

 

[TheCount]

Teacher, I have a question. Who in the blazes came up with this system?

 

[Gemini Dolly]

Okay, who else is turned on?

 

[SkinsFan28]

Thanks for the breakdown on that, I always wondered. And thanks to the explanation of how each variable got it's percentage figured out. I wonder why they capped it, and set 0 as base, seems like if a person has a negative or extremely high number that should be reflected also.

Again, interesting stuff.

 

[joseephuss]

I have always hope that if someone developed an improved system they would include components for yards per completion and TDs per completion.

In the older days QBs had lower completion percentages. When they did complete passes they made them count. They threw the ball down field more. The yards per completion was around 14.0 in the 50s and 60s and is around 11.5 today. Meanwhile the YPA has hardly changed going from 7.0 in the 50s and 60s to around 6.8 today.

Touchdowns per pass attempt has gone from about 5.0% in the 50s and 60s to around 4.0% today. TDs per pass completions has gone from about 10.0% in the 50s and 60s to around 6.8% today.

I like the current passer rating system. I would like to an improved system. Even if it were improved it would still remain only a tool to help in evaluating a QB. It would not evaluate a QB on its own. You still have to watch them play in order to properly do that. And if you are a coach, scout or GM you have to much more than that.

I think the DVOR or whatever acronym it is bases its evaluations on a lot more statistical information. I don't know that much about those systems.

 

[daschoo]

Okay, who else is turned on?

thats not a normal reaction!

 

[Hostile]

Okay, who else is turned on?

When a woman says turned on, all men immediately are. So thanks to you saying that, about 98% of this forum is.

I hope you're happy.

 

[CrazyCowboy]

great job Hos.....I believe I could do this in a pinch ;-)

 

[Hostile]

BTW, I got a PM from someone who reminded me that the four primary calculations are called CITY.

Completion %
Interception %
Touchdown %
Yardage %

So you could change the order and have the final formula be C+ I +T + Y, instead of A + B + C + D.

Just change the order if you can remember the default #s.

 

[YosemiteSam]

when I signed up for this site I was under impression there would be no math

you sir have shattered a dream

While it is very interesting to know how it's done. There are hundreds of websites that will to the qb rating math for you. So, there is no need to cry about it!

 

[Hostile]

While it is very interesting to know how it's done. There are hundreds of websites that will to the qb rating math for you. So, there is no need to cry about it!

I actually did not know that. Send me the links and I will add them to the OP. Or post them and hopefully those who are interested will bookmark the links.

 

[theogt]

Aikman had some ways to rank passing attacks in Offense that to me makes sense for QBs.

Adamknite posted a link, but I haven't gone to look yet.

All I know is Aikman's systems for Offense and Defense as a whole made better sense than current rankings. As kmd24 correctly pointed out the benchmark for those ratings are from a different generation and they don't work any longer with the advanced passing Offenses today.

The link refers to Aikman's team efficiency in terms of offense and defense. He doesn't have a passer rating system.

I agree that a team-oriented system is a much better gauge at determining overall team performance, but in terms of individual performance of QBs, Aikman's system certainly does a poorer job than just the QB rating system, simply because it takes into account non-QB factors, such as running back performance, etc.

It's really hard to think of a better gauge of a QB's performance than QB rating. You can make arguments for adding other statistics such as fumbles and rushing yardage by QBs, but I think there are very convincing arguments that those would distort the rating system as to make it more inaccurate, rather than accurate.

 

[YosemiteSam]

I actually did not know that. Send me the links and I will add them to the OP. Or post them and hopefully those who are interested will bookmark the links.

Google QB Rating Calculator

I check Romo's rating every week

 

[TwentyOne]

When i read the thread title i thought you'd gave us some reason why the equation looks like it does (means: why do comute the QB rating like this and not else) ?

I was a little bit disappointed by your post just because this equation is basic math - i think there is no no need to explain how to use the 4 basic mathmatic operations.

 

[Hostile]

Google QB Rating Calculator

I check Romo's rating every week

Thank you.

 

[Hostile]

When i read the thread title i thought you'd gave us some reason why the equation looks like it does (means: why do comute the QB rating like this and not else) ?

I was a little bit disappointed by your post just because this equation is basic math - no need to explain how to simply add some numbers.

I would say "I'm sowwy, fohgive me?" but it has already failed twice.

 

[theogt]

Those arbitrary "subtract 100 then divide by 20" definitions are a bit silly. It's totally unnecessary.

And I love how a QB goes 31-35 for 400 yards and 3TD's and it's a perfect QB rating. He missed 4 passes. Obviously not "perfect"

Thanks for sharing though.

Actually, the divisors are calculated such that each category (completion %, TDs per attempt, etc.) has a roughly equal variance among quarterbacks for any given year. I've ran the variance tests before and they're consistently very very close. This is what makes the formula work in that you can simply combine the variables into a single number.

 

[Spectre]

What a dumb formula this is.
I mean, it works okay as a reflection of good/poor play but are all those steps really needed to arrive at this conclusion?
What, did someone just decide to "divide by 20" or "multiply by 100" for no apparent reason?
What does those figures represent, if anything?

 

[the kid 05]

Now, if you will excuse me, my head hurts.

how does this differ from the Aikman scale you've mentioned before?

 

[theogt]

What a dumb formula this is.
I mean, it works okay as a reflection of good/poor play but are all those steps really needed to arrive at this conclusion?
What, did someone just decide to "divide by 20" or "multiply by 100" for no apparent reason?
What does those figures represent, if anything?

Check my post directly before yours.

I'll expound on this a bit.

Let's say that you have 20 linebackers that you're trying to compare. You want to come up with a single number that is derived from a formula that compares both their height and weight.

One option is to simply add their height in inches to their weight in pounds, without using any divisors. But this distorts the results to a point where the formula is meaningless. For example, you could have a linebacker that is 70 inches and 240 pounds (with a rating of 310). He would have the same rating as a player that is 74 inches and 236 pounds (rating of 310). Very clearly this does not make sense. One inch does not equal one pound when comparing football players.

In the same way, you can't compare completion percentage, touchdown percentage, interception percentage and yards per attempt by just adding them together. You have to come up with a formula that adjusts the numbers such that they are comparable and can be added.

One way to do this is to take those 20 linebackers and run a variance test (a statistical test) to see what the variance is for height and weight of the group. In very simple terms, this tells you how much the players deviate from one another in their height and weight. In order to directly compare height and weight, you would want to have the variance for each to be exactly the same. In other words, you want them to be as heavy/light compared to each other as they are tall/short. You do this by adding a divisor to each variable (e.g., divide by 20 or multiply by 100). You can adjust the divisor until you come up with one that gives you equal variance for each category. Once you have equal variance, you can simply add the numbers together to come up with the total index number.