Welcome to the YTS Leaderboard, Tez Walker!

Ravens Pass-Catchers Through 14 Games

College basketball teams in the Power 5 conferences schedule early-season “cupcake” games.  The reasons are numerous.  Sometimes teams want to increase their home revenue, so they pay teams to come play, hopefully earning more in ticket sales than they pay.  These are “buy” games or “guarantee” games – not guaranteed wins, but the visiting school is guaranteed a certain amount of money for making the trip.  Other times, a coach has a young team, or a new team with a lot of turnover, so he wants some “tune up” games before the rigors of conference play, to help his team jell.

Here is Maryland coach Kevin Willard talking about some of the considerations that go into scheduling. First he talks about the easy games helping his team jell, then he talks about “tune up” the games before & after conference games so his guys are ready, then have time to recover.

On Sunday the Ravens played a tune-up game against a Division 2 opponent.  It wasn’t a “challenging” game in the least – God, the Giants are SO bad – the only adversity the Ravens faced was self-inflicted (Lamar Jackson’s early fumble; the zillion penalties).  And they dominated, exactly as they should.  We didn’t “learn” much of anything about the Ravens in this game. There was no chance we would.

However, the game was perfectly situated to serve as a way to ease out of the bye week and into the stretch run.  The schedule-makers didn’t do the Ravens any favors with the current run; but this lead-in to it was everything they could have asked for.

Now they need to take advantage.

Rating the Rating

With not much to talk about in the game, let’s spend a long time dwelling on the history and details of the NFL’s Passer Rating statistic.  Lamar’s stat line against the Giants was this:

21 of 25 (84%) for 290 yards (11.6 yards-per) with 5 TDs and 0 INTs

Fabulously efficient.  Yet that was NOT good enough for Lamar to bag his 5th career game with a Perfect Passer Rating of 158.3.  His rating for the game was “only” 154.6.  How the hell is that possible?  What goes into the Passer Rating calculation, anyway??  And where the hall does it come from???

So. Passer Rating was invented in 1973.  Before that, from 1962 to 1972 the NFL used “ordinal” rankings, or an inverse ranking system, across 4 categories to rank the best passers.  What is an “ordinal” or inverse ranking system?  You’re familiar with this, I’m just using the fancy term for it.  The 4 categories were

• Completion Pctg
• TD Passes
• INT Pctg
• Yards-per-Attempt

To make an “inverse” or “ordinal” ranking system, you do this:

Data from PFR

Same kind of thing you might do for college rankings.  Bart Starr ranked 1st in completion%, 6th in TD passes, 1st in INT% and 3rd in Yds-per-Att.  Add up those ranks, you get 1 + 6 + 1 + 3 = 11. Do that for everybody; the player with the highest composite rankings (“ordinal” numbers) has the lowest sum (“inverse”).  Declare him the best passer.  In 1972 they swapped TD percentage for TD count, but still used the ordinal/inverse ranking.

Then-commissioner Pete Rozelle tasked a committee to come up with a better way to rank passers.  The committee was led by Don R. Smith, who was the VP of Public Relations at the Hall of Fame.  Smith worked with Seymour Siwoff of Elias Sports Bureau, official statisticians of the NFL, and Don Weiss, executive director of the NFL.  They studied the last dozen years of passing statistics, back to 1960 – not all passing stats, but the four components of the 1972 ordinal/inverse rankings: Completion%, TD%, INT% and Yards-per-Att.

For each component stat, Smith & Siwoff & Weiss used a linear scale based on three target points of reference: 0 for a horrible performance, 1 for an “average” performance, and 2 for a record-setting performance.  They capped each component at just under 2½ to leave a little room for records to go up while at the same time not letting one component completely dominate the other three.  And by “just under 2½” I mean exactly that: they capped each component at 2 and 3 eighths, which is 2.375.  Random!

In detail:

Completion Pctg

In the 1960s to early 70s, a completion% of about 50% was around average, and the single-season record was Sammy Baugh’s 70.3% in 1945.  So the three stat wizards Smith & Siwoff & Weiss used a basic linear formula, the same y = mx + b you learned in grade-school algebra, to plot a line.  50% gives 1 and 70% gives 2, so y = 1/20 • x + (-1.5).    A completion pctg of 30 (x = 30) gives a zero in that formula, so that’s the floor.  Each component is capped a 2.375, so if you plug in 2.375 for y in that formula, and do some algebra, you get x = 77.5.  So a Quarterback score in this component is maxed out when he hits 77.5%.

TD%

In the 60s to early 70s time frame, the league-avg TD% hovered between 5% and 5.6%, so the stat bros took 5% as the “1” score.  The best single-season TD% of that era was George Blanda’s 9.9% in 1961, so they took 10% as the “2” score.  They missed something here: the actual record was Sid Luckman’s 1943 season, 28 TDs on 208 attempts, 13.5%.  But for the data they used, 5% gives 1 and 10% gives 2, so the y = mx + b formula that fits those two points is just to divide TD% by 5.  Since each component is capped-out at 2.375, the max TD% a passer can receive credit for in the rating system is 2.375 • 5 = 11.875.  That’s less than the actual record was at the time!  Oops.

INT%

Obviously the best INT rate is zero, and INT rates get worse as they get higher.  Their interception component gets subtracted from the “cap” of 2.375.  The average INT rate of the time was taken to be about 5.5%, so that’s the “1” value.  Notice how that’s almost the same as the “1” score for TD%.  It was a different era.  The best practical INT rate was Bart Starr’s 1966 season, 3 INTs in 251 attempts → 1.2%.  They used 1.5%, which is close, as the value that would give a “2”.  So if you remember your y = mx + b, slope is “rise over run”.  The “rise” is from 1 to 2, and the “run” is from 5.5 to 1.5, so the slope is -¼.  The formula for the INT% component is the cap 2.375 minus INT% • ¼.  An INT% of 9.5 would give a zero score in this component, and scores don’t go negative, so any worse doesn’t affect the component score.

Yards-per-Att

The average yards-per-attempt for qualifying passers hovered around 7 to 7.5 in the years 1960 to 1972.  Even today it’s not much different: this season the value is 7.2.  So 7 was taken as the “1” value for this component.  Tommy O’Connell had 11.17 yards-per-attempt in 1957, which is outside the 1960 to ’72 time frame, but maybe the guys remembered it.  Sid Luckman had 10.86 in 1952.  The wizards took 11 for the “2” value.  If you use  y = mx + b to find the line passing through (7,1) and (11,2) you get y = ¼ (x – 3).  That’s the formula for this component: yds-per-att minus 3, divided by 4.  The cap of 2.375 equates to a yards-per-attempt of 12.5.

Putting it together

So now they have scores for each of the four components.  They wanted a “1” score in each component to represent average performance, and a “2” score in each component to represent record-setting performance.  So they summed the 4 scores, and then multiplied that sum to put the stat into an intuitive range.  Now, you or I might figure that we’d want a score of 4 ones to give a 50 and a score of 4 twos to give 100, so we’d multiply by 12.5.  That seems to make sense.  Smith & Siwoff & Weiss figured different: they multiplied by 100/6, which is 16⅔.  I have no idea what the hell they were thinking.  With that weird multiplier, an “average” performance where you get 1s across the board gets a passer rating of (1+1+1+1) • 100/6 = 66.67.  A “record-setting” performance where you get 2s across the board gets a passer rating of  (2+2+2+2) • 100/6 = 133.33.

The only thing I can figure is, they wanted to roughly map to letter grades.  A 50% is not a “passing” grade, it’s a fail, and scoring an “average” performance around a “fail” grade might have felt wrong to them.  A 66.7 is not failing.  It sort of feels like it – that’s a D, not something I would want to bring home to Dad on my year-end report card – but it passes, and it’s close to 70 which is a C.  Still seems a weird choice.  If they’d used a multiplier of 17.5, rather than 16.67, then an “average” score across all four components would give a Passer Rating of 70.

At the high end, they probably figured that a “record-setting” performance in every category across the board was not achievable, so the score of 133.3 wasn’t going to come into play.  And they weren’t wrong!  That 133 number has never been hit over a full season.  There have been super-high single-game Passer Rating numbers.  But the stat was designed to work full-season totals, and it still “works” for that – I mean, works as much as it ever did.  The single-season record for Passer Rating is 122.5 by Aaron Rodgers in 2011 (more on that single-season record below).

(Most of the foregoing comes from [1] a 2009 academic paper written for the Computational Science & Engineering Dept at San Diego State U, available here; [2] a 2005 piece on the Passer Rating stat from the Hall of Fame, available here; [3] a 2004 NYT article available here; and [4] the famous 1988 analytics book The Hidden Game of Football by Carroll & Thorn & Palmer, available here.)

The “perfect passer rating” is a maxed-out performance, where you hit the cap in every component, giving a passer rating of (2.375 + 2.375 + 2.375 + 2.375) • 100/6 = 158.33.

And that, FINALLY, brings us back to Lamar.  From the above, the components of the Passer Rating formula are capped.  Comp% caps out at 77.5%; TD% caps out at 11.875%; INT% caps out (floors out?) at zero; yards-per-att caps out at 12.5.

Lamar was short of another Perfect Passer Rating game by 0.9 yards-per-attempt.  He gets his maxed-out 2.375 score in 3 of the 4 categories, but in yards-per he gets “only” a 2.150.  Remember the designers of the system wanted a “2” score to represent a record-setting performance, so this is still great.  But not maxed-out.  Lamar’s final Passer Rating for the game is (2.375 + 2.375 + 2.375 + 2.150) • 100/6 = 154.58.

So that’s Passer Rating.

There are so, so, SO many criticisms of this stat. I hardly know where to start: everyone hates it. I actually don’t hate it, which makes me an oddity in football stats discussion circles. I’m kinda comfortable with it.  It was the first “sophisticated” football stat I was ever exposed to, and they were showing it to me on TV when I was still a teenager, so it snuck into my brain before I had my full stat-nerd shields up.  But I know I can only use it in very limited circumstances: to highlight a good game or a good stretch, and to compare within similar circumstances/seasons.

Let me see if I can sort the criticisms of this stat into some categories:

• Clunky, multi-part, non-intuitive calculation
• Artificial caps
• Historical drift
• Game-relevance

The clunky calculation speaks for itself.  No one can do this in their head.  I set up a spreadsheet for it, YEARS AGO – during the Kyle Boller years, in fact! – where I put in the completions, attempts, yards, TDs & INTs, and then copied down the formulas in the other columns from the rows above.  I’ve copied that same spreadsheet from computer to computer over the last 20 years, adding rows to it.  Maybe that’s one reason I don’t hate the stat, because I can quickly & easily do things like, “Lamar’s Passer Rating if you count the drops as receptions would be …”  For normal people, there’s no way to calculate a Passer Rating quickly.

The caps are weird.  They were weird at the time – seriously, a 0-1-2 system?  Capped at 2⅜?

The two biggest issues with Passer Rating are historical drift and game relevance.  The historical drift is easy to see.  I posted this graph of the growth in league-average Passer Rating back in Week 9:

A Passer Rating from 1977 can’t compare straight-up with one from 2000, and one from 2000 can’t compare with 2024.  It’s like dead-ball pre-Babe Ruth era baseball home run records versus modern times.  The stat has the same name, but things are totally different.  There are ways to try to adjust for the historical drift.  One way we looked at in Week 9 is to divide a player’s rating by that year’s league average, and list the ratio.  Another way is to create a “multiplier” by using the ratio (this year’s league avg) ÷ (that year’s league avg), then multiply the older Rating by it to “adjust” it to current standards.

For an example of that second approach, take Roger Staubach’s 1973 rating of 94.6.  The league average that year was 64.9.  This year’s league average 93.7.  The multiplier would be (93.7 ÷ 64.9) = 1.44.  Staubach’s rating times the multiplier would be 94.6 • 1.44 = 136.6.  So we’d say Staubach’s 94.6 would be equivalent to a 136.6 rating today.  I like that approach better than stacking ratios, because it preserves the look and “flavor” of the Passer Rating.  It helps us be duly impressed by the great passing performances of yesteryear.

“Game relevance” comes down to the facts that, first, Passer Rating does not include sacks; and second, changes in scheme & tactics gave rise to ways of “gaming the system.”  Especially after Bill Walsh’s West Coast Offense proliferated around the league, a Quarterback (or an Offensive Coordinator) could pad the Passer Rating without really helping the team win.  The small-ball dink-&-dunk offenses could generate high completion percentages and keep the yards-per-attempt high, without really moving the ball.  Think of a checkdown pass on 3rd-&-10, leading to a punt.  The completion percentage goes up, but the offense goes off the field.

Everyone’s got their own tweak to “fix” passer rating for game relevance.  My own is to leverage the ideas in YTS: replace completion pctg with Success Rate, and replace Yards-per-Attempt with Net-Yards-per-Attempt.  That brings sacks into it, and rewards keeping-the-chains-moving rather than meaningless completions (like an 8yd gain on 3rd-&-10).  It also would cut everyone’s rating by around 20 points, which would bring the number back into the range its designers were envisioning back in 1973.  If you did that with this year’s Passer Rating, the top 3 would remain the same, but Josh Allen & Jordan Love would move up, while Sam Darnold & Jalen Hurts would move down.  But PFR only has Success Rate back to 1994, so this version of Passer Rating can’t be retro-calculated to seasons before that.

It’s probably true that Passer Rating just isn’t worth trying to “fix.”  It’s not ever going to be as good as DVOA or EPA or the other advanced stats we have nowadays.  DVOA normalizes to the era (0 is average every season), so it would have to be first choice for comparing historical seasons to modern ones.  It and EPA and this column’s YTS all incorporate sacks and “keeping the chains moving” (successful plays).  PFF is based on per-play film grading, and Completion Pctg Over Expectation (CPOE) is also based on individual play analysis, so they give something that the full-season stats can’t.  Most of the sophisticated stat people in the NFL space don’t even notice Passer Rating.

I myself am not quite ready to chuck it.  We’re all so used to seeing it!  Passer Rating has been part of the NFL-scape for 50 years.  It incorporates TDs & INTs.  It aligns nicely on a “100 = very good” scale, so even if the calculation isn’t intuitive, the output usually is.  We know a very good Passer Rating when we see one.  It’s still reported on sometimes; and a Perfect game makes the news.

Also, Lamar leads the league in it, so for at least a few more weeks I’m going to keep paying attention to it. 

Stats

Here are your stats for the game:

(Data from NFL)

Hello, Tez Walker!  I guess I can retire the #FreeTezWalker hashtag.  One thing we can say about this dude is that he has earned his spot on the gameday-active roster.  This was his first catch of the season, but it was his 6th game.  What does a raw, unpolished 4th-rd rookie have to do to get on the field?  Play special teams, of course.  Walker played 15 snaps on Teams vs the Giants; made the tackle on the 1st-quarter kickoff after the TD to Andrews.  He also had double-digit spec-teams snaps vs the Chargers and Bengals.

There’s an old-school philosophy of player development at work here.  It’s easy for us to be impatient with this slow-poke stuff, but there is a wisdom to it.  Learn to be a professional, get your body right to be able to play at this level (running, blocking, tackling), and you will get your opportunities.  First-rounders don’t need to go thru this process, but less-developed players often do.  Justice Hill played 95 snaps on special teams as a rookie, and almost 200 the following year. Now he’s an essential piece of the offense.  If Walker is taking a similar path – and responding to it! – then we could see him in the WR3 role full-time next season.  Note that Nelson Agholor and Tylan Wallace are both pending free agents this year.  Walker’s development over the next few weeks might be important.

Oh; and hello to Rashod Bateman too.  Bate left last week’s game with a knee injury and no targets.  I guess he’s okay!

Zay Flowers’ day is interesting.  He catches 6 of 7, notches a QS – and is the worst of the Ravens pass-catchers by efficiency!  That’s a testament to how Lamar was dealing it.

Justice Hill with a perfect day receiving.  He has turned into SUCH a quality player.  He’s not going to make anyone’s Pro Bowl list; but he’s productive as hell and a crucial cog to this offense.

The Mark Andrews watch continues, with his hot streak running to 10 games:

His yardage totals aren’t eye-catching, just 35 yards a game, but the consistency is a difference-maker. He’s the team leader in QS. Oh, and the Red Zone production is also a difference-maker.

Season Stats & Leaderboard

Here is a pop quiz, before we look at the full-season stats to date. We have a new player atop the Ravens YTS leaderboard.  Given what you know about how stats work, who do you think it is?

(Data from PFR except last 3 cols)

Yep: when your only target for the season is a 21-yd Touchdown catch, your efficiency numbers are going to look damn good.  It’s a mirage, of course; Small Sample-size Theater.  But the TD has to be a great confidence-builder for the young man.  It’s also going to make the coaches look positively at him.  Something to build on.

Tylan Wallace: still perfect.  My low-tech take on Wallace’s stat line is: when a guy catches everything and makes big plays, hey maybe he should be more involved in the offense, huh?  But, only so many eligible receivers can take the field at a time.  Who are you going to take snaps away from to use Wallace?  Bateman, who’s 10th in the league in yards-per-target and just caught two Touchdowns?  Isaiah Likely?  Mark Andrews?  Zay Flowers?  This offense is one of the best in the league.  It ain’t broke, so no one’s in any hurry to “fix” it.  Wallace might have to be content for now with being the best WR4 in the league.

11 Ravens have caught Touchdowns this season.  Last year the number was 9.  Same in 2019; the other Greg Roman years the number was usually 7.  In the Super Bowl season of 2012, only 6 Ravens caught TD passes, which seems surprising.  Remember, that team lost 4 of their last 5 in the regular season and fired the Offensive Coordinator after game 13.  The low in team history was 5 players catching TDs.  That happened in 1997, 2011 and 2003; but in 2003 one of the players was Jonathan Ogden on a trick play.  So to me 2003 was the “real” low.

11 is not the team record.  In the inaugural season of 1996, 12 players caught TD passes – one was Ogden, again.  Could this year’s squad tie that record?  Well, Diontae Johnson seems a poor bet to catch a TD for the Ravens.  So I think they’d need a trick play to a lineman – or to Lamar!  Derrick Henry has thrown 5 TD passes in his career (including playoffs), some to his Quarterback.  Lamar catching a TD could be a real possibility.  However, I imagine Todd Monken is holding that in his pocket for a desperate situation.  I doubt we see it in the regular season.

Is there value in having so many different players catch Touchdowns?  I think there is.  It signals diversity – different ways to challenge defenses – and depth.  The Ravens don’t have that one dominant receiver like Justin Jefferson or Ja’Marr Chase.  So instead they’re using a bunch of guys, which creates a “culture” of scoring and productivity.  This reflects a Todd Monken philosophy: every player is a threat, every player gets used.  A fun way to play (when it’s working).

Flowers is 14th in the league in yards, 23rd in receptions.  Bateman is 10th in yards-per-target; Andrews is 23rd, but 5th among Tight Ends.  Andrews has climbed to 6th overall in receiving Success Rate, 3rd among TEs.  Likely is 23rd in Success Rate.

Lamar is back on top of the league in Passer Rating, at 120.7.  That’s the 4th-highest single-season passer rating leader all-time.  This list is crazy, so I’m going to post it at length:

MVPs and Hall of Famers.  Steve Young!  Joe Montana!  Dan Marino!  This is an amazing list to be near the top of.  (Note that list is just the leaders year by year.  There are also 2nd- and 3rd-place finishers in some years who would place on this list if they were included.  For example, in 2019 Ryan Tannehill finished first in Passer Rating with 117.5; but Drew Brees posted 116.3 that year, and Lamar posted 113.3.)

So, before we get too excited – Lamar’s rating probably isn’t going to stay this historically high.  He’s got tough games remaining, starting with Pittsburgh; also on the road at Houston, who have, hello, the #1 defense by DVOA; and finishing up with a divisional game, which of course will be a dogfight.  So Lamar’s efficiency numbers will inevitably decline a bit.  To me, that’s the reason to post it now and admire it.  He’s having an amazing season.

Lamar’s passer rating could decline by 10 pts and he’d still lead the league.  He’s 1st in TD%, Yards per Attempt and Net Yards per Attempt.  He’s 2nd in TD passes (2 behind Joe Burrow) and 5th in yardage.  Lamar is first in YTS (NetYdsPerAtt times SuccessRate), decimal points ahead of Jared Goff.  His TD to Interception ratio is – well it’s ridiculous, let me put the top ten on a table so we can properly appreciate it:

League average is 2-to-1.  Baker Mayfield is right behind Lamar with 32 TD passes, but he also has 14 INTs for a ratio of 2.3 to one.  Patrick Mahomes has 22 TD to 11 INTs, a perfect 2-to-1 ratio.  Lamar’s ratio is 5 times higher than that!  It’s double Josh Allen’s, who’s getting a lot of MVP buzz.

Your Ravens are still the #1 offense by DVOA.  They’re #1 in Yards-per-play, #1 in Red Zone TD%, $1 in Passing YTS.  They’re 2nd in 3rd-down conversions,3rd in points-per-drive, and 4th in Scoring%.  Most interestingly, they have moved into the #1 overall spot in DVOA:

Aaron Schatz of DVOA writes:

We’ve got a new No. 1 team atop the DVOA table this week, but it’s not the team you’re expecting. For most advanced EPA-based metrics across the Internet, the No. 1 team in the NFL is now the Buffalo Bills. DVOA has things a bit differently. Our new No. 1 team is the 9-5 Baltimore Ravens.

Baltimore knocks Detroit down to second place after the Lions lost to the Bills. Green Bay is a surprising third, followed by Buffalo, Philadelphia, and then Minnesota. Kansas City is up one spot to seventh, San Francisco down one spot to eighth.

The Ravens have been floating around with one of the best offensive DVOA ratings we’ve ever measured for most of the season. Their defense has now improved to the point where the whole team comes out at No. 1.

Because he’s a quant, he investigates in his write-up this week why his method (DVOA) has the ratings different from the other method (EPA).  The details are interesting to us stat nerds; but his executive summary is that turnovers and 4th-down plays have greater weight in EPA than they do in DVOA.  That makes sense.  Turnovers and 4th-downs are enormously high leverage plays: they are going to have an outsize effect on “expected points.”  But those plays are a little random.  You have Small Sample Size issues. DVOA tries to be “predictive,” and turnovers are not very predictable.  Likewise, 4th-down plays can be coin flips.  So while they have enormous impact, DVOA down-weights them in order to be more “predictive.”

Who’s right?  Well, we’ll see, won’t we?

With the win and the performance the Ravens playoff odds by DVOA have ticked up to 99.8%, and their odds to win the Super Bowl have ticked up from #4 to #3.

Next Up

Steelers Week, pt 2.  The game is on Saturday, not Sunday.

Baltimore needs this game in order to win the division.  I think we all agree that they probably need to win it for their psyches, too.  As of Thursday the Ravens are 6-point favorites.  That’s a pretty big line for a 10-4 team to be an underdog by.

As you may know, Lamar has never started against the Steelers at home with the crowd present.  He did start the 2020 home game vs the Steelers; but that was the COVID year, and only a few thousand fans were permitted.  Lamar had a poor completion% that day, 13 of 28.  But he was at 7.4 yards-per-attempt, which is solid, and he had 2 TDs to 2 INTs, plus rushed for 65 yards.  He lost a fumble on the potential game-winning drive in the 4th quarter, and that iced the game (a 4-point Steelers win).

And that was Lamar’s only home start vs Pittsburgh.  He’s missed all the other home games vs the Steelers. Here’s the full list:

• 2018 – both games played by week 9; Lamar became starter week 10
• 2019 – season finale, Ravens had clinched, Lamar sat
• 2020 – game 7, covid so few fans. Loss
• 2021 – season finale, Lamar out with injury
• 2022 – second-to-last game, Lamar out with injury
• 2023 – season finale, Ravens had clinched, Lamar sat

You think Lamar is hype for this game?  I do.  But if so he’s concealing it well.  He seemed very chill at Monday’s post-practice press conference.

I, on the other hand, am tense as hell. Alternately confident and fearful.  I said this to “Huddle It Up Films” on Xitter:

I’ll tell you what:
I think team “walk in yo trap take over yo trap / we own Lamar” is in for a reality adjustment on Saturday.

— jimzipcode.bsky.social (@zip_jim) December 19, 2024

Grrrrrr.

The post Welcome to the YTS Leaderboard, Tez Walker! appeared first on Russell Street Report.

Source: https://russellstreetreport.com/2024/12/19/street-talk/welcome-to-the-yts-leaderboard-tez-walker/