Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
sweet thursday
Sep 16, 2012



Let's talk Advanced Stats and eventually make our own so we can sell it to Disney and make a lot of money.

Examples of our competition:
DVOA
DVOA is a method of evaluating teams, units, or players. It takes every single play during the NFL season and compares each one to a league-average baseline based on situation. DVOA measures not just yardage, but yardage towards a first down: Five yards on third-and-4 are worth more than five yards on first-and-10 and much more than five yards on third-and-12. Red zone plays are worth more than other plays. Performance is also adjusted for the quality of the opponent. DVOA is a percentage, so a team with a DVOA of 10.0% is 10 percent better than the average team, and a quarterback with a DVOA of -20.0% is 20 percent worse than the average quarterback. Because DVOA measures scoring, defenses are better when they are negative

PRO FOOTBALL FOCUS:
I have no idea what their secret sauce is but the good news is after reading this, neither do they
https://www.pff.com/grades
PFF employs over 600 full or part-time analysts, but less than 10% of analysts are trained to the level that they can grade plays. Only the top two to three percent of analysts are on the team of “senior analysts” in charge of finalizing each grade after review. Our graders have been training for months, and sometimes years, in order to learn, understand and show mastery of our process that includes our 300-page training manual and video playbook. We have analysts from all walks of life, including former players, coaches and scouts. We don’t care if you played.

ESPN QBR WTFBBQ
ESPN’s Total Quarterback Rating (Total QBR), which was released in 2011, has never claimed to be perfect, but unlike other measures of quarterback performance, it incorporates all of a quarterback’s contributions to winning, including how he impacts the game on passes, rushes, turnovers and penalties. Also, since QBR is built from the play level, it accounts for a team’s level of success or failure on every play to provide the proper context and then allocates credit to the quarterback and his teammate to produce a clearer measure of quarterback efficiency.
--‐----

There's some others too but the main point is the market is ready for a Grand Unified Theory of Football: a way to rank QBs based on Defenses, Defenses based on Special Teams, and Special Teams based on the sales of alcohol in the third quarter. We can do this.

To do list:
- Quantify the Eye Test
- Qualify the Gut Check
- Put the Science back in Who's Hot and Who's Not
- invent the Kirk Cousins Modifier that explains the inverse relationship between Skill and When The Lights Getting Too Bright on a Monday Night

sweet thursday fucked around with this message at 15:54 on Nov 30, 2021

Adbot
ADBOT LOVES YOU

Impossibly Perfect Sphere
Nov 6, 2002


sweet thursday posted:

- invent the Kirk Cousins Modifier that explains the inverse relationship between Skill and When The Lights Getting Too Bright on a Monday Night

It's called Clutch OP.

And there's only one stat that matters: Just Wins Games

sweet thursday
Sep 16, 2012



Impossibly Perfect Sphere posted:

It's called Clutch OP.

And there's only one stat that matters: Just Wins Games

Meaningless words and phrases. Where is the numbers, the hard data, the /'s and ='s and %'s?
You are now in charge of our Clutch Division. I want some fluff on my desk by Friday.

Impossibly Perfect Sphere
Nov 6, 2002


Didn't a statistician in TFF - within the past few years - do a long effort post proving that fantasy football points are a more accurate predictor of QB success than any other advanced stats? Just copy whatever Yahoo does.

Yates
Jan 29, 2010

He was just 17...



sweet thursday posted:

Meaningless words and phrases. Where is the numbers, the hard data, the /'s and ='s and %'s?
You are now in charge of our Clutch Division. I want some fluff on my desk by Friday.

Kawalimus
Jan 17, 2008

Better Living Through Birding And Pessimism


Stats in football can never be as accurate as stats in baseball. In baseball performances are way more isolated. But in football your success way more heavily depends on your teammates. Add to that football has so few games played relative to the other sports. And its just harder to get an accurate statistical measure. But its still ok to try. But you just have to acknowledge that its tougher for football.

sweet thursday
Sep 16, 2012



That's a good point.

- Petition the NFL to add more games to the season so we have a bigger sample size.

Kalli
Jun 2, 2001




Also advanced analytics have mostly made coaches go for it on 4th down more often and throw the ball more instead of never ever ever as opposed to baseball where it's turned the game into an unwatchable slog.

Magicpants
Sep 15, 2011


Certified Poster

sweet thursday posted:

That's a good point.

- Petition the NFL to add more games to the season so we have a bigger sample size.

Sadly, this may not be enough. The big problem with football is they let a bunch of guys run around and do different things on every play, which is just terrible game design. Makes quantifying individual outcomes very difficult. Every part of the play should happen independently. First have the the pass rushers go against the oline, if the defense gets past them, its a sack; or if not the qb comes on the field and throws the ball as far as he can. Then the wr and cb stand on the spot where it lands and a jugs machine fires a ball at them. And for christsake we absolutely cannot let coaches call different plays, everyone does the same thing from here on. Actually now that I think about it, get rid of coaches, why does this sport even have coaches if they're not putting them on the field? Literally no coaches have ever gained more yards than any player in a game, look it up. So dumb.

Nodoze
Aug 17, 2006

If it's only for a night I can live without you

Kirk Cousins is good

Kalli posted:

Also advanced analytics have mostly made coaches go for it on 4th down more often and throw the ball more instead of never ever ever as opposed to baseball where it's turned the game into an unwatchable slog.

There is a lot more to it than this. It's not the advanced analytics so much as the strategy teams have started employing with relief pitchers, and as it turns out you don't and won't ever have to pay those guys real money. You can just burn them out before free agency and let them go like a running back and then the next dude throwing high 90s takes his place

Nodoze fucked around with this message at 19:07 on Nov 30, 2021

Impossibly Perfect Sphere
Nov 6, 2002


Nodoze posted:

Kirk Cousins is good

Close thread. It's already hosed.

fartknocker
Oct 28, 2012

Damn it, this always happens. I think I'm gonna score, and then I never score. It's not fair.




Wedge Regret

Stats cannot comprehend the glory of Super Bowl Champion Jared Lorenzen.

Ornery and Hornery
Oct 22, 2020



OP, thank you for doing the needful.

About time we broke ground on this deliverable.

Management has been on my rear end.

Ornery and Hornery
Oct 22, 2020



Nodoze posted:

Kirk Cousins is good

There is a lot more to it than this. It's not the advanced analytics so much as the strategy teams have started employing with relief pitchers, and as it turns out you don't and won't ever have to pay those guys real money. You can just burn them out before free agency and let them go like a running back and then the next dude throwing high 90s takes his place

Please ELI5 how advanced stats new new strategies have made baseball unwatchable. More unwatchable, I guess.

I have not seen a MLB game in years.

Spoeank
Jul 16, 2003

That's a nice set of 11 dynasty points there, it would be a shame if 3 rings were to happen with it


Ornery and Hornery posted:

Please ELI5 how advanced stats new new strategies have made baseball unwatchable. More unwatchable, I guess.

I have not seen a MLB game in years.

The long and the short of it is it's all dingers, walks and strikeouts now and coaches don't trust starting pitchers to go more than 5 or 6 innings anymore and then it's a cavalcade of RP.

Spoeank fucked around with this message at 23:03 on Nov 30, 2021

Ornery and Hornery
Oct 22, 2020



Spoeank posted:

The long and the short of it is it's all dingers, walks and strikeouts now and coaches don't trust starting pitchers to go more than 5 or 6 innings anymore and then it's a cavalcade of RP.

Three true outcomes. Advanced Statistical Perfection.

Cavauro
Jan 9, 2008




Joe Burrow

Spoeank
Jul 16, 2003

That's a nice set of 11 dynasty points there, it would be a shame if 3 rings were to happen with it


Hello here is how you can grade offensive lines, it is my proprietary advanced analytic:

(Sacks + Hits + Hurries)/Yards per Throw

I call it SHHYT. Bad lines are bigger piles of SHHYT. Call me PFF_Spoeank


Edit:

Ornery and Hornery posted:

Three true outcomes. Advanced Statistical Perfection.

That's right. There's more in there (swing plane optimization) but MLB is basically just maximizing TTO now

Kalli
Jun 2, 2001




Spoeank posted:

Hello here is how you can grade offensive lines, it is my proprietary advanced analytic:

(Sacks + Hits + Hurries)/Yards per Throw

I call it SHHYT. Bad lines are bigger piles of SHHYT. Call me PFF_Spoeank

Ignores TTT, QB's who hold onto the ball too long are forcing their line into worst situations and eat more SHHYT

syzpid
Aug 9, 2014


according to QBR, David Garrard has the 10th best single season by a QB. A season in which Fred Taylor ran for just short of 5 and a half yards an attempt, and Maurice Jones-Drew had 4.6 yards an attempt. Quinn Gray started the only 4 games of his career that year with the Jags and his 73.6 QBR, if it qualified would put him between last years Russel Wilson & Lamar Jackson.

Despera
Jun 6, 2011


Nodoze posted:

Kirk Cousins is good


So you are saying thay we wont need eyes where were going?

CannonFodder
Jan 26, 2001



Passion’s Wrench

Impossibly Perfect Sphere posted:

It's called Clutch OP.

And there's only one stat that matters: Just Wins Games

Ahh, the Eli Manning 4th Quarter stat.

blarzgh
Apr 14, 2009

SNITCHIN' RANDY


Grimey Drawer

Some of the best recent advanced analytics to come publicly available are EPA and CPOE; they have the advantage of large sample sizes, being per play statistics, and have shown a very strong correlation with game outcome and Future performance.

blarzgh
Apr 14, 2009

SNITCHIN' RANDY


Grimey Drawer

EPA stands for expected points added; it uses gobs and gobs of historical play data to give you the fractional amount of points any given play result adds.

For example, a team that gets 7 yards on first down, from their own 20, gains a small amount of expected points (.03 or something). A team that gets 3 yards and a first down inside the opposing 20 yard line gets lots of expected points (think 2.5).

It calculates this based on the change in probability of a drive resulting in points, caused by the preceding play.

This helps give weighted values to yardage gained on particular plays at various places on the field, which is incredibly useful for measuring the impact of a player's contributions, ignoring meaningless yardage totals.

Similar to win percentage added, but more granular.

blarzgh
Apr 14, 2009

SNITCHIN' RANDY


Grimey Drawer

CPOE is completion percentage over expectation. It's an excellent yardstick for sifting out all the screen passes, and other high percentage throws, from a quarterback's overall completion percentage.

Using play-by-play and tracking data, we can find the average completion percentage at every depth of target. The league might complete passes that are thrown to receivers 15 yards down the field only 50% of the time, whereas swing passes behind the line of scrimmage are completed 95% of the time.

CPOE looks at an individual passer, and compares their completion percentage at every depth of Target, and tells us whether they're completing those passes and a higher, or lower rate than league average.

So if Dak Prescott completes passes 15 yards down the field 60% of the time, he's got a cpoe of plus 10

pasaluki
Feb 27, 2008

THIS WHAGON HAS NO BREAKS! I HAVE THE HEART OF THE BUUFALO the strength OF THE MOUNTAIN, THE FURY OF THE THUNDER AND MY WILL IS UNBREAKABLE! I will not surrender to KNOW ONE


The Ravens being kind of deceased but still effective made me think it would be cool if there was an advanced stat that interpreted a team's resiliency

Is there a way to quantify teams in terms of how good they are at surviving/enduring injuries to critical players.

Teams like the Titans and Ravens and even the Cardinals in some games seem to be able to keep up a level of effectiveness regardless of whose injured. I think maybe teams that are higher in resilience skew towards that run, teams with more depth vs top end talent, teams that run simpler defensive and offensive packages. Maybe there's a way to quantify that using this stupid science crap.

The Ravens are bizarre though. They are the only team that has an MVP QB that aren't particularly phased if he plays or is infected with disease, injured, or plays just part of the time and the other time shits himself violently in the locker room. They are a running based team that like is hemorrhaging running backs like just stacks of bodies but they are pretty much just fine and roll with it.

I think the Chargers would add so much data to this stat. I would need to dissect Bosa's corpse to really perfect it though.

blarzgh
Apr 14, 2009

SNITCHIN' RANDY


Grimey Drawer

pasaluki posted:

The Ravens being kind of deceased but still effective made me think it would be cool if there was an advanced stat that interpreted a team's resiliency

Is there a way to quantify teams in terms of how good they are at surviving/enduring injuries to critical players.

You can look at snaps lost to injury, and extrapolate record from there

Rooster Brooster
Mar 30, 2001

Maybe it doesn't really matter anymore.

I've been wanting to do an advanced stats thread for so long, thanks for putting one up! I hadn't heard of CPOE and that sounds pretty sweet.

Here are three topics I've been ruminating on and would want more people to know about : DVOA GOOD/DVOA BAD. Opponent Adjustments. Era-adjusted quarterback rating. I'll post about them if others don't, but just the first one now since I'm short on time.

DVOA GOOD/DVOA BAD

Don't worry, DVOA is simple! The explainer page is only 6672 words, after all! They at least have a headline explanation for everybody:

quote:

DVOA measures a team's efficiency by comparing success on every single play to a league average based on situation and opponent.

DVOA has some good ideas in it.

Situation-aware play weighting makes some sense. So, like, a 3 yard run on 1st and 10 is less valuable than a 3 yard run on 4th and 2. One issue it brings up is how much you weight those plays, how do you track how much better or worse they are? DVOA just weights them with some arbitrary, unknown value turning them into its DYAR stat.

It's a per-play stat, not a counting stat, meaning you can't make your DVOA better by just doing more, you have to do well on each play that you do run.

Opponent adjustments definitely make sense. Basically you play the "yeah but who have they beaten that's any good?" game but with math. This is a whole topic on its own.

They look at per-drive stats. This is something that always struck me as odd about football, per-possession outcomes never seem to be considered.

They at least try to include special teams. Advanced special team analytics, wow! Opponent-adjusted punts-over-average!!!!

They try to exclude "lucky" outcomes from being counted in a team's favor - fumble recoveries, deflected pass interceptions (as opposed to deflected pass incompletions), etc.

But there is a lot of bad:

The formulas and such used to calculate DVOA are completely opaque, we have to trust they make sense and are actually calculated correctly. Last I knew this was all in a big Excel spreadsheet, so god knows it'd be easy to make a mistake. They specifically say their latest 7.3 version fixes errors in prior calculations, so it has happened before.

Relatedly, DVOA has tons of tweakable parameters that the author can change at any time, and has. Like, okay, a 3 yard run on 1st and 10 should be worth a little less... how much less? It also weights passing offense as more valuable than run offense, but we don't really know how or by how much.

It's not clear to me why they condense their situation-aware offense and defense stats down into a single number, or even offense/defense totals. DVOA isn't even the raw stat, it's the percentile above league average. So I don't get to know that Atlanta got 5.6 defense-adjusted situational yards per pass, I know they got 89% of league average or whatever.

They point out DVOA is just a descriptive stat that describing how a team has done (at least they used, maybe they've changed stance on this), but then they proceed to use it predictively all over the place. Like how they insinuated Matt Flynn was destined to be the greatest quarterback since Marino because of the amazing DVOA number he posted in one week 17 game. They even now offer betting advice on their site based on DVOA, but last I heard it's predictive ability wasn't any better than ANY/A or fantasy points.

In fact, that should be its own point here. It's not at all clear DVOA is much better a descriptive or predictive stat than other, simpler ones, like quarterback rating, ANY/A, or heck, even Pythagorean points.

There are also issues with the site itself, like how they post the same "TEAM XXX has the NTH BEST/WORST OFFENSE/DEFENSE DVOA through M weeks EVER!" article many times every year, ignoring small sample sizes making such runs pretty meaningless. But hey, they got space to fill, I get it.

This post is only 685 words long.

Impossibly Perfect Sphere
Nov 6, 2002


I like your post. Keep posting more about other stats.

Vincent Van Goatse
Nov 8, 2006

My "Not Vincent Price" shirt has people asking a lot of questions already answered by my shirt.


Smellrose

The biggest problem with stats is that they don't account for plays that add GLORY to professional football as a game. Seriously, what effect did Dan Connolly's beautiful kick return have on DVOA or QBR or whatever voodoo bullshit the statisticians use for their bloviating? None at all as far as I can tell. Which is pure bullshit IMO.

Now I hate the Pats as much as the rest of you, but even so we have to recognize the absolute beauty of fat linemen gaining yardage. It just isn't football if a fat lineman doesn't have an opportunity to do something incredible, and any stat calculation worth a poo poo should reflect this. gently caress clutch, there should be a metric that acknowledges linemen TDs or gaining yardage,

Vincent Van Goatse fucked around with this message at 06:31 on Dec 4, 2021

Chucktesla
Jul 13, 2014



Can any of these big brained stat guys tell the difference between split zone and duo

blarzgh
Apr 14, 2009

SNITCHIN' RANDY


Grimey Drawer

Chucktesla posted:

Can any of these big brained stat guys tell the difference between split zone and duo

Do you know the difference between Staley and Fangio's gap and a half assignments for defensive tackles?

Spoeank
Jul 16, 2003

That's a nice set of 11 dynasty points there, it would be a shame if 3 rings were to happen with it


Chucktesla posted:

Can any of these big brained stat guys tell the difference between split zone and duo

A split zone is the area between like the 7 pin and the 10 pin if you knock all the other ones down and a duo is a pairing of two people

Chucktesla
Jul 13, 2014



blarzgh posted:

Do you know the difference between Staley and Fangio's gap and a half assignments for defensive tackles?

No and I don't think the chargers d line does either

MrLogan
Feb 4, 2004



Maybe in another 20 years when the current NFL coaches all retire we'll see offenses do the smart thing and:

1) start going for 2 after TDs
2) start running play action for the majority of their snaps

It's mind boggling to me that NFL teams run PA as little as they do.

Hamhandler
Aug 9, 2008

[I want to] shit in your fucking mouth. [I'm going to] slap your fucking mouth. [I'm going to] slap your real mother across the face [laughter]. Fuck you, you're still a rookie. I'll kill you.

IMO the generation of some of these complicated and weighted composite statistics(like DVOA, QBR, etc.) are a fool's errand and it doesn't seem like they're that useful from a business standpoint. Are the people satisfied by "was this guy good" statistics really going to buy a subscription, or is it the guy who wants to know how often the DE lines up at DT without spending three hours on NFL gamepass? It might just be me but I am a lot more interested in their less convoluted stats and the stuff where my interest outweighs my guilt about them paying some poor Bangladeshi guy $5 a day to count how many times Christian Wilkins is in coverage,

Chucktesla posted:

Can any of these big brained stat guys tell the difference between split zone and duo

I think that's really kind of it- there aren't a lot of guys who can put the full picture together on both the statistics and football side, and I imagine those who can are probably guys who are largely employed and trained by actual NFL teams and not posting and selling subscriptions.

sweet thursday
Sep 16, 2012



Advanced Stats finally died today
[

A.o.D.
Jan 15, 2006

The Suffering of the Succotash.


I thought advanced stats had been saying that the Lions aren't as bad as their record all season long.

sweet thursday
Sep 16, 2012



A.o.D. posted:

I thought advanced stats had been saying that the Lions aren't as bad as their record all season long.
Interesting point. We need an advanced stat to take into account the competing advanced stats to see which one wins out

Adbot
ADBOT LOVES YOU

Mega64
May 23, 2008



Advanced stats say Kirk Cousins is the greatest quarterback of all time, so clearly it is the football game that is wrong.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply