PDA

View Full Version : New Angle on Analyzing the Steelers


thethinkingguy
01-21-2013, 08:56 PM
All, my partner and I feel there is a dearth of quantitative analysis being applied to the Steelers. Thus, we've founded the site: http://www.steelcitystats.com.

I encourage you to check out our first article, entitled "Why did the Steelers Miss the 2013 Playoffs?"

We use basic statistical analysis and some intermediate methods to obtain which variables (e.g., passer rating) are important to predicting why teams have won or lost games. No data analysis can provide a 100% accurate conclusion, but I think you'll see that the data speaks for itself.

If you like it, please forward it to friends and family. If you don't like it, please let us know what we can do better.

Hawaii 5-0
01-21-2013, 09:20 PM
great stuff, thank you! :drink:

ZoneBlitzer
01-22-2013, 04:22 AM
Ew. Sorry I took one look at the dude with naked beer gut and tats of Steelers on it and lost my appetite for the article. I'm a fan of the team and all but there's no way I'm going to emblazon their faces on me like Mt. Rushmore. That's a little weird. It's like aging sports fan meets Tiger Beat Teen magazine.

pete74
01-22-2013, 04:30 AM
Good read

VaDave
01-22-2013, 05:14 AM
YEE GADS!! Thank you for a classic example of why I flunked statistics in college.

BTW, That sure took all the fun out of wailing and gnashing of teeth......

NateDog07V
01-22-2013, 12:03 PM
Awesome read.... keep it up !

JVip
01-22-2013, 12:36 PM
good article...

Riddle_Of_Steel
01-22-2013, 08:29 PM
So....being the usual skeptic that I am....I am wondering where the 2011 Chargers, who finished with the #1 offense and #1 defense, yet finished 8-8 fit into this analysis.

How do those two losses to JAX in 2007 happen? David Gerrard was terrible in both games, if memory serves, and we lost because MJD and FT ran for a combned 250+ yards on us in the first game, and ran well against us in the wildcard game as well. According to the statistical analysis, the running game is NOT a factor in wins-- only passer ratng and turnovers.

NSMaster56
01-22-2013, 08:30 PM
GREAT article.

Only nit to pick is this:

To reiterate, Pittsburgh lost the division because of poor QB play and poor passing defense in the three division games they lost.

While your analytics 'prove' your statement, it's more apt that Pittsburgh lost the division because of turnovers and Special Teams play in the three division games they lost (Punt Return TD vs. BAL, 7 TO's vs. CLE and pick-six vs. CIN).

It would be interesting to see the stats based on your model for all 32 teams.

GREAT stuff.

Riddle_Of_Steel
01-22-2013, 08:31 PM
Great article anyhow-- I just like to poke holes in statistical studies like this, because I am convinced that you cannot determine outcomes from simple statistical correlation.

If you could, don't you think the owners would be more interested? Imagine how much the "winning algorithmn" would be worth to the right people...?

desertsteel
01-22-2013, 08:36 PM
Good stuff!

I once prepped the Professor of Statistics from Northwestern University for a trial on credit card fraud. I COULD NOT get him to understand how overlimit fees were billed LOL. I guess we each have our gifts.

harrison'samonster
01-22-2013, 08:44 PM
Great article anyhow-- I just like to poke holes in statistical studies like this, because I am convinced that you cannot determine outcomes from simple statistical correlation.

If you could, don't you think the owners would be more interested? Imagine how much the "winning algorithmn" would be worth to the right people...?

great point. Statistics are very usefull, but they can't account for things like John Elway's little helicopter move in the SB. Or Charlie Batch moving the offense against the Ravens to beat them at home this year.

thethinkingguy
01-22-2013, 09:03 PM
Great comments, all, and thank you for the compliments. And Riddle of Steel, you can never go wrong with a Conan reference in my book...

The article is not trying, and the model is not predicting, that rushing the ball is useless. What the model notes is that passing the ball well is more conducive to winning than is running the ball well over the last three years. I think you all know this intuitively to be true, but I'm just trying to show it statistically.

Surely running the ball is valuable, it just doesn't predict wins as well as passing does, that's the bottom line. The new NFL analytical way of thinking is as follows: Winning teams that run the ball well do so mainly because they're passing well. They're running once the pass is established as a lethal means of attack.

If you're into this kind of stuff, I encourage you to follow me and my partner on Twitter at @the_thinkingguy and @the_drinkingguy. We'll tweet out (and I'll try to remember to come back here and post) when a new article is posted to www.steelcitystats.com

As far as the question on San Diego...I think you meant the 2010 Chargers, right? The model predicts they should have won 10.5 games. They went 9 and 7. The standard deviation for wins is 2.5 in the league, so the model accurately predicts this.

The 2011 Chargers weren't the dominant ones in scoring offense and scoring defense. But, if you're curious, the model predicts they won 7.75 games. They won 8.

Fire Arians
01-22-2013, 10:55 PM
good read, for the most part can you say turnovers killed the season? ranking 28th in fumbles definitely doesn't help the winning cause

VaDave
01-23-2013, 07:14 AM
So....being the usual skeptic that I am....I am wondering where the 2011 Chargers, who finished with the #1 offense and #1 defense, yet finished 8-8 fit into this analysis.

How do those two losses to JAX in 2007 happen? David Gerrard was terrible in both games, if memory serves, and we lost because MJD and FT ran for a combned 250+ yards on us in the first game, and ran well against us in the wildcard game as well. According to the statistical analysis, the running game is NOT a factor in wins-- only passer ratng and turnovers.

That anomoly is called Zone Blocking. We had no answer for it. We still don't defend it well. That may be the only glaring deficiency of the LeBeau defensive system. In all fairness Jacksonville ran it to perfection that year.

You may remember in that playoff game, the Jaguar's victory was greatly added by a missed holding call.

Riddle_Of_Steel
01-23-2013, 05:34 PM
That anomoly is called Zone Blocking. We had no answer for it. We still don't defend it well. That may be the only glaring deficiency of the LeBeau defensive system. In all fairness Jacksonville ran it to perfection that year.

Unfortunately true. We can't seem to nail down a ZBL on our own Oline (ala Kemoeatu), but other teams have done so with success against us at times.

You may remember in that playoff game, the Jaguar's victory was greatly added by a missed holding call.

That was a bad memory I was trying to forget...the 4th down that sunk our 2007 season.

Riddle_Of_Steel
01-23-2013, 05:41 PM
Great comments, all, and thank you for the compliments. And Riddle of Steel, you can never go wrong with a Conan reference in my book...

If the "Crush your enemies, see them driven before you, and to hear the lamentation of their women" part were shorter-- I would have taken that as a pseudonym LOL.

The article is not trying, and the model is not predicting, that rushing the ball is useless. What the model notes is that passing the ball well is more conducive to winning than is running the ball well over the last three years. I think you all know this intuitively to be true, but I'm just trying to show it statistically.

Surely running the ball is valuable, it just doesn't predict wins as well as passing does, that's the bottom line. The new NFL analytical way of thinking is as follows: Winning teams that run the ball well do so mainly because they're passing well. They're running once the pass is established as a lethal means of attack.

Fair enough. I have to keep in mind too that statistics show overall trends, and that exceptions, while they do occur, get flattened out over time and their impact is minimalized when you factor everything in.

If you're into this kind of stuff, I encourage you to follow me and my partner on Twitter at @the_thinkingguy and @the_drinkingguy. We'll tweet out (and I'll try to remember to come back here and post) when a new article is posted to www.steelcitystats.com

I was into statistics when I was majoring in psych, but have since moved on to IP network engineering. (my trade today). I forgot nearly everything I learned in college statistics-- although when someone lays it out as well as you did-- it is not hard to follow, even for a laymen.

As far as the question on San Diego...I think you meant the 2010 Chargers, right? The model predicts they should have won 10.5 games. They went 9 and 7. The standard deviation for wins is 2.5 in the league, so the model accurately predicts this.

Not bad at all. Keep the good information flowing....

NSMaster56
01-23-2013, 06:45 PM
You may remember in that playoff game, the Jaguar's victory was greatly added by a missed holding call.

Big Ben pick six (in addition to a couple other porous picks) and KR TD doomed the game more than any amount of MJD or bad calls.

thethinkingguy
01-23-2013, 09:16 PM
If the "Crush your enemies, see them driven before you, and to hear the lamentation of their women" part were shorter-- I would have taken that as a pseudonym LOL.

I love it. "Crom laughs at your four winds. He laughs from his mountain."


Fair enough. I have to keep in mind too that statistics show overall trends, and that exceptions, while they do occur, get flattened out over time and their impact is minimalized when you factor everything in.


Yes, exactly. Remember, statistics can only DISPROVE theories or myths. Statistics prove nothing. But, that doesn't preclude them from being useful. I find them to be incredibly helpful.

I'm amazed by the lucidity in this forum. If anybody would like some data ran, please let me know and I'm happy to help. I'll be tweeting out our next article when it's ready and I'll post here as well.

austinfrench76
01-24-2013, 01:05 PM
Good read buddy! Thanks.

NSMaster56
01-24-2013, 06:46 PM
If anybody would like some data ran, please let me know and I'm happy to help.


For (game-by-game) comparison purposes it would be interesting to see the other 31 teams QB stats.

I'll be tweeting out our next article when it's ready and I'll post here as well.

Added yous guys to my faves.