PDA

View Full Version : Defying FO: Can The Texans Buck Advanced Statistical Analysis


CloakNNNdagger
07-26-2010, 09:39 PM
.................And Make The Playoffs?

Evidently, Football Outsiders 2010 Almanac has given the Texans 5.6 projected mean wins.

This writer summarizes each on of the five reasons that FO thinks the Texans will be worse than they were last year.

[Clink LINK (http://houston.sbnation.com/2010/7/26/1585842/texans-football-outsiders-nfl-2010-predictions))

DiehardChris
07-26-2010, 10:03 PM
Rivers McCown wrote it for SBNation. He's a newish voice on the Houston Texans blogger scene, and he's KILLING it. You guys should bookmark that site and keep reading his stuff for sure.

JB
07-26-2010, 10:05 PM
Rivers McCown wrote it for SBNation. He's a newish voice on the Houston Texans blogger scene, and he's KILLING it. You guys should bookmark that site and keep reading his stuff for sure.

Yep! I follow him on twitter so I can read all his stuff. Classic!

m5kwatts
07-26-2010, 10:07 PM
These are some flaws in the formula I've detected:

-Does strength of schedule really mean anything? We all know the Steelers had the toughest schedule when they won the Super Bowl a few years ago. And so often we see teams (including our Texans) play down to their competition.

-Weather is impossible to factor in this formula. On any given weekend the weather conditions can weaken an opponents' assumed advantages over another. (Hate to use the Steelers again as an example but) Did FO factor in the weather when the Steelers beat the winless Dolphins 3-0 on MNF in the mud? Of course not.

-Injuries are a part of the formula. One thing I thought was interesting was when the article asserted that "the Texans WILL have more injuries in 2010 compared to 2009." Thats an outrageous claim with no tangible evidence other than league trends to back it up. I don't care what stats you use, injuries to an NFL team cannot be predicted with calculated numbers.

-Numbers don't always tell the truth in the NFL. Clearly this formula relies on statistical evidence. But how often does statistical evidence tell the truth about what's happening on the field? Guys get tackle stats for merely being in the area of a play. Another example, say player A goes unblocked on a snap and gets a sack. And lets say Player B was the one tying up the blockers who allowed Player A to get the sack. Well lets say Player A signs with another team in the offseason. He carries all the credit for that sack with him to the next team when it really belong with player B. But Football Outsiders says "Wow player A got a sack that makes his new team that much better in our formula." Thats a glaring flaw.

CloakNNNdagger
07-26-2010, 10:21 PM
-Injuries are a part of the formula. One thing I thought was interesting was when the article asserted that "the Texans WILL have more injuries in 2010 compared to 2009." Thats an outrageous claim with no tangible evidence other than league trends to back it up. I don't care what stats you use, injuries to an NFL team cannot be predicted with calculated numbers.


Certainly, I don't know for sure. But one thing they may be looking at is how many of the base roster is coming back off of 2009 injuries.........making them higher risk for reinjury in 2010.

76Texan
07-26-2010, 10:34 PM
It's utterly useless to use data from the previous year to predict the future.
Teams don't have the same players they did.
They are coaching changes that will affect the level of play of some team

TheRealJoker
07-26-2010, 10:36 PM
"-Injuries are a part of the formula. One thing I thought was interesting was when the article asserted that "the Texans WILL have more injuries in 2010 compared to 2009." Thats an outrageous claim with no tangible evidence other than league trends to back it up. I don't care what stats you use, injuries to an NFL team cannot be predicted with calculated numbers."

I somewhat disagree with this part of your post. Sure, freak injuries are unpredictable. But depending on how tough a coach runs a training camp, and how good the S&C/training staff are, also plays a role in the team's injuries.

gtexan02
07-26-2010, 10:39 PM
How many wins did they project for us last season?

steelbtexan
07-26-2010, 10:44 PM
7-9

I called it.

Make fun of me later, but I dont want to hear the usual excuse why the Texans didn't make the playoffs Uncle BoB.

m5kwatts
07-26-2010, 11:01 PM
Certainly, I don't know for sure. But one thing they may be looking at is how many of the base roster is coming back off of 2009 injuries.........making them higher risk for reinjury in 2010.


"-Injuries are a part of the formula. One thing I thought was interesting was when the article asserted that "the Texans WILL have more injuries in 2010 compared to 2009." Thats an outrageous claim with no tangible evidence other than league trends to back it up. I don't care what stats you use, injuries to an NFL team cannot be predicted with calculated numbers."

I somewhat disagree with this part of your post. Sure, freak injuries are unpredictable. But depending on how tough a coach runs a training camp, and how good the S&C/training staff are, also plays a role in the team's injuries.

Neither of these were mentioned as factors in the formula but I agree they would make sense. This is what the article says about injuries:

Although I certainly believe that Houston has a lot of young talent on defense, the parts haven't added up to anything valuable yet -- even with the eighth-healthiest unit in football last year, the Texans' defense was 18th in the league in DVOA. They're going to suffer more injuries this year (even without including the Cushing suspension), and our projection system thinks the increase in injuries will be more important than the growth of Houston's young players

The bold part is what I'm objecting to. So because we were the 8th healthiest team in football last year, what goes up must come down and we'll be an injury-nagged team this year? Thats quite a reach.

And I have a discrepancy with their "healthiest football team" ranking. That doesn't measure the importance of the injured players. A team could be the healthiest team in the league for a year but lost their starting QB week 1 and lose 12 games that season because of it.

DiehardChris
07-26-2010, 11:01 PM
Yep! I follow him on twitter so I can read all his stuff. Classic!

I assume that you also follow me on Twitter, else I'll get my feelings hurt.

@HoustonDiehards :)

JB
07-27-2010, 06:20 AM
I assume that you also follow me on Twitter, else I'll get my feelings hurt.

@HoustonDiehards :)

Of course I do!

badboy
07-27-2010, 09:39 AM
I am willing to read any thought process concerning the Texans as long as it appears logical, research was done and it does not appear someone is being a homer or attacker. Most of us realise the team could go 5-11 or 11-5. I think this is the best roster we have had and if our draft picks are what front office thinks they are and Smith does well on Oline, we could be very good. Sure there are question marks but that can be said every off season at this time. Right now I am not concerned about injuries that might occur or other teams. If the players can do what the game plan says and Kubes improves on penalty flag decisions and game clock management (and he should), we will go beyond round one of play offs. Our offense should be just as good if not better and the D should improve quite a bit imo. I think we can have a 6-0 or 5-1 start. I fear no opponent this season.

Lucky
07-27-2010, 06:38 PM
I like stats. A lot. They're good for telling you what happened. Not as good for telling you what will happen.

One reason NFL football is so difficult to predict is the sheer lack of statistically significant events. There are only 256 regular season games in a NFL season. As opposed to 2430 in a MLB season. There's not a lot of data to make predictions from.

Not to mention determining what data to use when running your models. Or if one set of data strongly affects another set. I had to laugh when I read this:

Since 2000, 15 first-round cornerbacks have started 10 or more games in their rookie seasons, and only 4 of them have made the playoffs. Cornerbacks usually take a few years to develop, and early in their careers they make easy targets.Really? Starting a 1st round rookie CB was a major determining factor in these teams not making the playoffs? Only 12 of the leagues 32 teams make the playoff in a given year (37.5%). 4 of the 15 teams starting 1st round rookie CB made the playoffs (26.7%) So the 10.8% difference is because of the 1st round CBs? Hell, just start a 2nd round or later CB, then.

There's a difference between losing Cushing for 4 games due to injury, and 4 games due to suspension. The Texans have the entire preseason to formulate a plan for this occurrence. The Texans know that they will be without Chester Pitts for the bulk of the season. Unlike last season. Does FO's modeling take that into account?

An equally valid model could be concocted that shows the Texans are more likely than not to become a playoff team. Just by selecting and/or discarding data that does or does not support the model. Attempting to use numbers only in determining what has happened is a daunting task. Attempting to use numbers only in determining what will happen is folly. What do your eyes see? Well, nothing yet. There hasn't even been a preseason game played. But, from what I've seen in the past, and what maneuvers I've witnessed in the offseason, this does not look like a team more likely to lose more games than they win.

I trust my eyes.

michaelm
07-27-2010, 11:32 PM
I trust my eyes.

Agree with your entire post, but this sums it up nicely.

The Pencil Neck
07-28-2010, 10:14 AM
How many wins did they project for us last season?

Last year, they projected us to get 6 wins.

I love me some maths and I love me some statistics but I don't care how advanced the techniques are that you use, you gotta play the games. The biggest problem from projecting from last year's stats is that these are no longer last year's teams.

I love Football Outsiders but I'm not worried about their predictions.

sometexansfan
07-29-2010, 05:29 PM
We had better or else... :bat:

CloakNNNdagger
07-29-2010, 08:40 PM
Here's additional material further addressing the 5.6 wins projected by FO.

It is a very worthwhile "chat" between Rivers McCown and Vince Verhei of FO.

RiverSide Chat: Vince Verhei Of Football Outsiders (http://www.battleredblog.com/2010/7/29/1593366/riverside-chat-vince-verhei-of)

CloakNNNdagger
07-30-2010, 08:25 AM
These are some flaws in the formula I've detected:

-Injuries are a part of the formula. One thing I thought was interesting was when the article asserted that "the Texans WILL have more injuries in 2010 compared to 2009." Thats an outrageous claim with no tangible evidence other than league trends to back it up. I don't care what stats you use, injuries to an NFL team cannot be predicted with calculated numbers.

Certainly, I don't know for sure. But one thing they may be looking at is how many of the base roster is coming back off of 2009 injuries.........making them higher risk for reinjury in 2010.

It's utterly useless to use data from the previous year to predict the future.

"-Injuries are a part of the formula. One thing I thought was interesting was when the article asserted that "the Texans WILL have more injuries in 2010 compared to 2009." Thats an outrageous claim with no tangible evidence other than league trends to back it up. I don't care what stats you use, injuries to an NFL team cannot be predicted with calculated numbers."

I somewhat disagree with this part of your post. Sure, freak injuries are unpredictable. But depending on how tough a coach runs a training camp, and how good the S&C/training staff are, also plays a role in the team's injuries.

Neither of these were mentioned as factors in the formula but I agree they would make sense. This is what the article says about injuries:

Originally Posted by CloakNNNdagger
Certainly, I don't know for sure. But one thing they may be looking at is how many of the base roster is coming back off of 2009 injuries.........making them higher risk for reinjury in 2010.

Originally Posted by TheRealJoker
"-Injuries are a part of the formula. One thing I thought was interesting was when the article asserted that "the Texans WILL have more injuries in 2010 compared to 2009." Thats an outrageous claim with no tangible evidence other than league trends to back it up. I don't care what stats you use, injuries to an NFL team cannot be predicted with calculated numbers."

I somewhat disagree with this part of your post. Sure, freak injuries are unpredictable. But depending on how tough a coach runs a training camp, and how good the S&C/training staff are, also plays a role in the team's injuries.

The bold part is what I'm objecting to. So because we were the 8th healthiest team in football last year, what goes up must come down and we'll be an injury-nagged team this year? Thats quite a reach.

And I have a discrepancy with their "healthiest football team" ranking. That doesn't measure the importance of the injured players. A team could be the healthiest team in the league for a year but lost their starting QB week 1 and lose 12 games that season because of it.

I communicated with Football Outsiders concerning the basis of their "Team Health" analysis.

Dear Doc,

A variety of factors, including the team's health over the past few
years and the nature of the injuries (e.g. if the injury record is
mostly traumatic injuries like a torn ACL as opposed to nagging ones
like a sprained ankle).

-Bill

thunderkyss
07-30-2010, 08:32 AM
It's utterly useless to use data from the previous year to predict the future.
Teams don't have the same players they did.
They are coaching changes that will affect the level of play of some team

So, what do we use to make us believe (those of us that do believe) that our team will finish better than they did last year?

thunderkyss
07-30-2010, 08:58 AM
I will say this for all the nay-sayers.

This will be the most important season in Texans history...
:splits:

This is our first opportunity to have back to back winning seasons. Sure, it wouldn't warrant the fanfare of last years first winning season, but if they don't finish 9-7 or better, for what ever reason, I think that will set the team, the franchise, back several years.

It wouldn't matter what happens to the coaching situation after that, IMO, the team would be devastated, and it will take years to recover.

But I will say this, for the stats. Sure it is worthless to predict future outcomes based on previous data. But the idea should be to look for trends. FO numbers are especially useful, I think, because the numbers are about as normalized as you can get them. Which, I think is better for head to head comparisons.

I haven't done the work, of studying our trends from game to game. But I have to believe (based on what my eyes tell me) that we were a better team, a totally different team really, at the end of 2009, than we were at the beginning.

No doubt in my mind, if we had started 2009 with that team, we would have won 11 games.

So there is no doubt in my mind, that even though we will play the "toughest" schedule in 2010, some of those teams on our schedule are going to look "not as tough" at the end of the year as they do now, and we will be part of the reason.

drs23
07-30-2010, 09:48 AM
I will say this for all the nay-sayers.

This will be the most important season in Texans history...
:splits:

This is our first opportunity to have back to back winning seasons. Sure, it wouldn't warrant the fanfare of last years first winning season, but if they don't finish 9-7 or better, for what ever reason, I think that will set the team, the franchise, back several years.

It wouldn't matter what happens to the coaching situation after that, IMO, the team would be devastated, and it will take years to recover.

But I will say this, for the stats. Sure it is worthless to predict future outcomes based on previous data. But the idea should be to look for trends. FO numbers are especially useful, I think, because the numbers are about as normalized as you can get them. Which, I think is better for head to head comparisons.

I haven't done the work, of studying our trends from game to game. But I have to believe (based on what my eyes tell me) that we were a better team, a totally different team really, at the end of 2009, than we were at the beginning.

No doubt in my mind, if we had started 2009 with that team, we would have won 11 games.

So there is no doubt in my mind, that even though we will play the "toughest" schedule in 2010, some of those teams on our schedule are going to look "not as tough" at the end of the year as they do now, and we will be part of the reason.

:goodpost: I've thought along the same lines but could not have penned them as well as you have here, TK.

Runner
07-30-2010, 12:39 PM
Isn't this based on the same DVOAs that were widely proclaimed by this board last year as the gold standard of statistics (when, coincidentally, they predicted great improvement for the Texans)? As I recall, they were frequently cited as concrete proof of the direction the Texans were headed. They also were touted as a better indicator of the team than the much maligned win/loss record.

I believe that cacophony I detect is the sound of tunes changing.

The Pencil Neck
07-30-2010, 12:45 PM
Isn't this based on the same DVOAs that were widely proclaimed by this board last year as the gold standard of statistics (when, coincidentally, they predicted great improvement for the Texans)? As I recall, they were frequently cited as concrete proof of the direction the Texans were headed. They also were touted as a better indicator of the team than the much maligned win/loss record.

I believe that cacophony I detect is the sound of tunes changing.

Statistics tell you what is and what was.

We don't have any statistics for this team or any of the current teams in the NFL. When we get into the season and start piling up statistics for this team, then I'll be back on FO's bandwagon. They've got some great stats and those stats can help us look for what's going right/wrong with the team.

But when it comes to prediction? No. There's no statistic out there that can predict the future. They predicted we'd win 6 games last season. They were wrong. Now they're using statistics from last season to try to predict what's going to happen this season and they've got us winning 5 games. When it comes to prediction, they're the same as any "expert" out there putting together power rankings and trying to predict what's going to happen in the coming season.

Their prediction is largely based on how tough our schedule looks. But their definition of the "toughness" of the schedule is based on how strong these teams were last year. That's as good a method as any to predict the future but it's also not something you should bet on, either.

Runner
07-30-2010, 12:54 PM
Statistics tell you what is and what was.

We don't have any statistics for this team or any of the current teams in the NFL. When we get into the season and start piling up statistics for this team, then I'll be back on FO's bandwagon. They've got some great stats and those stats can help us look for what's going right/wrong with the team.

But when it comes to prediction? No. There's no statistic out there that can predict the future. They predicted we'd win 6 games last season. They were wrong. Now they're using statistics from last season to try to predict what's going to happen this season and they've got us winning 5 games. When it comes to prediction, they're the same as any "expert" out there putting together power rankings and trying to predict what's going to happen in the coming season.

Their prediction is largely based on how tough our schedule looks. But their definition of the "toughness" of the schedule is based on how strong these teams were last year. That's as good a method as any to predict the future but it's also not something you should bet on, either.

I don't give these stats any more or less weight than I did last year - which wasn't very much. I do think FO have an interesting approach and can give some insight into some aspects of the team. I also think that insight varies from team to team and is surrounded by a lot of noise, so it's predictive use is limited.

I'll still look at the win/loss record to see how the team is doing; I don't care what the DVOAs say.

The Pencil Neck
07-30-2010, 12:59 PM
I don't give these stats any more or less weight than I did last year - which wasn't very much. I do think FO have an interesting approach and can give some insight into some aspects of the team. I also think that insight varies from team to team and is surrounded by a lot of noise, so it's predictive use is limited.

I'll still look at the win/loss record to see how the team is doing; I don't care what the DVOAs say.

I think it helps to have everything at your fingertips because the W/L record can be just as deceptive as any other stat.

Goldensilence
07-30-2010, 01:02 PM
I will say this for all the nay-sayers.

This will be the most important season in Texans history...
:splits:

This is our first opportunity to have back to back winning seasons. Sure, it wouldn't warrant the fanfare of last years first winning season, but if they don't finish 9-7 or better, for what ever reason, I think that will set the team, the franchise, back several years.

It wouldn't matter what happens to the coaching situation after that, IMO, the team would be devastated, and it will take years to recover.

But I will say this, for the stats. Sure it is worthless to predict future outcomes based on previous data. But the idea should be to look for trends. FO numbers are especially useful, I think, because the numbers are about as normalized as you can get them. Which, I think is better for head to head comparisons.

I haven't done the work, of studying our trends from game to game. But I have to believe (based on what my eyes tell me) that we were a better team, a totally different team really, at the end of 2009, than we were at the beginning.

No doubt in my mind, if we had started 2009 with that team, we would have won 11 games.

So there is no doubt in my mind, that even though we will play the "toughest" schedule in 2010, some of those teams on our schedule are going to look "not as tough" at the end of the year as they do now, and we will be part of the reason.

I don't see that if they don't make the playoffs or increase their win total at least the team being so devastated that it takes years to recover.

I do think if that happens, we'll hopefully see a coaching change and with the right hire it could propel the team further. I still think had the job become available, Cowher would've taken it.

thunderkyss
07-30-2010, 01:03 PM
Isn't this based on the same DVOAs that were widely proclaimed by this board last year as the gold standard of statistics (when, coincidentally, they predicted great improvement for the Texans)? As I recall, they were frequently cited as concrete proof of the direction the Texans were headed. They also were touted as a better indicator of the team than the much maligned win/loss record.

I believe that cacophony I detect is the sound of tunes changing.

Yes and no.

Part of their 5.6 mean... is that they fully expect Schaub to miss at least two or three games, because of the flu.

My rose colored glasses don't see that.

They also see DanO as being a total flop when he becomes the man, simply because we had Grossman ahead of him on the depth chart.

I believe Rex was ahead for other reasons.

& I think they are making more out of missing Cushing for 4 games than I am. But they may be right on this one.

Runner
07-30-2010, 01:06 PM
I think it helps to have everything at your fingertips because the W/L record can be just as deceptive as any other stat.

I'd say the W/L record is less deceptive of a team's performance than most other stats, and even less deceptive than the opinion of a team's hard core fans.

Most sports put a premium on that W/L ratio when all is said and done.

Goldensilence
07-30-2010, 01:09 PM
I'd say the W/L record is less deceptive of a team's performance than most other stats, and even less deceptive than the opinion of a team's hard core fans.

Most sports put a premium on that W/L ratio when all is said and done.

At this point I would stop trying to argue. This one goes under:

"Best team to not make the playoffs or win their division!"

Runner
07-30-2010, 01:13 PM
At this point I would stop trying to argue. This one goes under:

"Best team to not make the playoffs or win their division!"

You're right. I'll attempt to remove myself from the fray.

Runner
07-30-2010, 01:15 PM
You're right. I'll attempt to remove myself from the fray.

The "Return of the DVOAs" was just too good to pass up...

The Pencil Neck
07-30-2010, 02:17 PM
I'd say the W/L record is less deceptive of a team's performance than most other stats, and even less deceptive than the opinion of a team's hard core fans.

Most sports put a premium on that W/L ratio when all is said and done.

There's no argument that W/L is the only important stat.

But if you're trying to figure out if your team is really any good and whether they're going to win or lose in the future, W/L is just as deceptive as other stats can be.

If you don't believe me, ask the Denver Broncos the past couple of years. Start off fast, get lucky and win some by fluke and then crash and burn.