Better Box Score Metrics: AUDL Season Awards [AUDL Championship Weekend, 2023]

An EDGE-based taken on the AUDL Awards.

August 25, 2023 by Paul Würtztack in Analysis with 0 comments

Time for the annual “this is how we see the awards podium shaping up” review. Because +/- cannot be trusted for comparing players,¹ we offer this EDGE-based “statistical baseline” as a starting point for evaluating MVP candidates and other individual honors.

For the sake of a little extra data, I will include the two rounds of playoffs already played instead of limiting this to just the regular season as I did last year. However, so as to not give an unfair advantage to players on teams who got one or two extra games, all the cumulative stats that constitute EDGE will be calculated on a “per team-game” basis. If the team had 13 games, their EDGE total is divided by 13. If the player missed two of those games, they’re EDGE total is still divided by 13. I do not use per-games “played” because I also blend in E100, which adds a heavy dose of “productivity rate.” Players needed to play in at least seven games total to qualify.

One could imagine that including playoff games might work against players on those teams, given that they have additional games against superior competition, but it’s not like we’re throwing off well-balanced regular-season schedules (Exhibit A: Austin), and anecdotally, even teams that lose in the playoffs typically play at least as well as they played, on average, over the regular season, whether as a result of more reps together as a team, playing up to the level of the competition, or other reasons. Of the eight teams competing in the Divisional Championship round, only Indy did not score more efficiently than its regular-season average. Overall, I think it’s unlikely that there’s a discernible bias either way.

The one other difference between this year’s methodology and last year’s is that in weighting the contributions of EDGE (production) and productivity (E100), I have returned to a 1:1 weighting instead of a 2:1. To quickly review, while EDGE-O and EDGE-B track season totals, E100 accounts for the fact that D-line players have less EDGE-earning opportunity than O-line players, and that O-line players on good teams with strong break-generating D-lines typically have far fewer possessions than those on weaker teams (and vice-versa). This year’s version of 2022’s Marx/Osgar Disequilibrium is the Smith/Martin Imbalance: Atlanta’s Matt Smith and Seattle’s Garret Martin both played 12 games: Smith had 228 possessions, Martin 361. Martin’s 133 extra possessions is seven games’ worth of possessions for Smith. E100 neutralizes the imbalance.

Pawel Janas has been among the top four in full-season EDGE in each of the three years we’ve been doing this, and maybe, just maybe, that kind of production has something to do with the combined 8-win swing in the fortunes of Chicago and LA this year. Pawel is hardly the only one who met his own lofty standards this year: Ryan Osgar, Jordan Kerr, and James Lewis were also among the top six in full-season EDGE last year. Two of those three were first-team All-AUDL, while the third did not get an Honorable Mention; acclaim comes hard if you miss the playoffs. Garret Martin more than tripled his per-game yardage this year after moving to Seattle, and is likely to get some consideration for BPOTY. Max Sheppard and Evan Swiatek also had huge seasons, with relatively little fanfare.

League experience is evidently not a key factor in getting blocks, as Ambrose and William Wettengel—the top two in EDGE-B this year—are both rookies, while Justin Burnett is in his second year. And for as lethal as New York’s offense has been, the Empire have twice as many leaders in EDGE-B—Jagt, Randolph, Babbitt, and Drost—than EDGE-O.

Productivity Rates

Table 2 shows the 21 top EO100, EB100, and E100 rates, and one prominent attribute of E100 is that it’s heavily populated with players on good teams. While only about half of the players in the EDGE categories were on playoff teams, 17 or 18 playoff-bound players fill the E100 rosters

This is where Ryan Osgar’s case for MVP comes into view. Osgar is so far ahead in EO100 (EDGE per 100 possessions) that Sean Connole in 2nd—yes, Connole, not Kerr—is closer to the 21st player on the list than the 1st. Some of those in his wake, including Kerr and Ben Jagt, have higher block rates, which help close the gap in overall E100, but Osgar still stands atop.

Similarly, one might start with Jeff Babbitt’s second-to-none EB100 block rate to make his MVP case, even though most would probably start with his goal scoring rate, not shown here but easily looked up on the AUDL site, where he is also first among players with more than 2 games played. For Wettengel, not only did he have a lot of blocks, he generated them at a high rate, whereas Ambrose’s comparable EDGE-B was built up over a league-high 420 Opponent Possessions, which is tiring just thinking about.

Ordering up the combo: EOz, EBz, and Ez

Table 3 shows the outcome of combining production and productivity in a 1:1 ratio. To combine values from different scales, we again use z-scores, a measure of how many standard deviations (SD) a given value is above or below the mean. To illustrate with one example, Ryan Osgar’s EDGE-O of 3.58 per team game (Table 1) was 2.75 SD above the mean, and he achieved that in relatively few possessions, so his EO100 of 17.2 (Table 2) is an incredible 3.41 SD above the mean. The average of the two (EOz) is 3.08, as shown below.

Before we further discuss this output in assembling our podiums, we have one more chart. Because the AUDL often highlights the offensive statistics of the players it puts on the DPOTY podium, BBSM prefers to break up DPOTY into two awards in order to clarify the basis on which we’re evaluating players: “Defenders,” evaluated on their ability to get blocks (and who could play predominantly on O-line), and “D-Line Players,” evaluated for their ability to generate breaks. For the former, we use the block-based EBz scores above, an insufficient measure of defensive prowess, but all we have. Break Generation is a combination of EDGE-B (block totals) and EO100 (offensive productivity) in equal proportion. I’ve set a qualification threshold of having played 60 percent D-line. Table 4 shows the full top 21 by this methodology.

The Podium Contenders

I now present what an end-of-season podium might look like if it were only based on EDGE methodologies and excluding Championship Weekend results. I’m not saying this is who should get the awards, but it’s a good starting point for integrating their statistical output. For players who get elevated far above what this baseline suggests, it will be intriguing to hear the case for doing so. Total seasons stats for every player is found in the table at the end.

MVP: Ryan Osgar

Contenders: Jordan Kerr, Pawel Janas

Osgar’s 2023 was not quite as amazing as his 2022 MVP year, but it was pretty damn close, and his Ez lead over Jordan Kerr is substantial. If the Empire get another title, it’s hard for me to see how Osgar’s production and precision is not seen as the most important component of that success. Now if Jordan Kerr has a couple of strong performances in leading the Shred to a championship, he’d also be worthy. Winning the championship is not a prerequisite, but given that our two main contenders are on a collision course, it’s reasonable to assume it will factor in. Pawel won’t win, but we’re highlighting here that his stats should keep him in the conversation. If you’re backing other horses, fine, but how does their statistical profile compare to these three?

Or let’s not be obtuse—what do we think of the fact that Babbitt still seems to be the presumptive favorite, given that he ends up 17th in our Ez rankings? Or put it the other way: what do we think of our rankings given that Babbitt is a popular favorite?

The first point is that EDGE values total production, and therefore is going to fade players like a Babbitt who have less of it. Babbitt has 2649 receiving yards, which is good, but it’s not in the top 30 for the league. Because he adds only a small amount of throwing yardage—228 net yards—his total yardage is outside the top 100. When Jagt won the MVP in 2021 (and likewise came out on top in this column), he led the league in receiving yardage—even on a per-game basis to account for the playoffs—such that he was among the league leaders in total yardage despite being outside the top 150 in per-game throwing yardage.

Similarly, leading the league in regular-season goals with 50 as Babbitt did is no better in EDGE’s eyes than having 25 goals and 25 assists, and so considering all the players who had more than Babbitt’s 52 total scores, he’s outside the top 25 on this account as well.

If you want a statistical case for Babbitt, here’s how you get there. You weight the rate of getting goals and blocks heavily and above all else. You also note that he has an excellent Completion Percentage Plus (CP+) of 96.2 (15th in the league), and conclude that 13 scores and four blocks for every turn is about as good as it gets (and overlook that Matt Smith has 52 scores for every turn, or that Lukas Ambrose has nine scores and eight blocks for every turn). The more straightforward approach is to simply declare him to be the most compelling player to watch in any game (no argument here), and award him the MVP on that basis.

Offensive Player of the Year: Jordan Kerr

Contenders: Pawel Janas, Sean Connole, Jacob Miller

If I said three guys from the same team would all be vying for OPOTY, you would of course assume that I was talking about the Shred, right? The point is, Kerr, Connole, and Miller have all had incredible offensive seasons. The West has been assumed to be an offense-happy division that gives a large advantage to offensive players such as the Shred Trio, but the West only averaged 87 more yards and 1.4 goals per game more than the East, so while there is some inflation, the appearance of the Shred players mainly has to do with them balling out (and being partly responsible for the inflation themselves). Connole has 3100 throwing yards, 6600 total yards, 57 scores, just 18 turns. Miller has even more throwing yards (3500) and fewer turns (9). Kerr essentially comes out on top because of his outrageous number of scores (92). However, this award is Osgar’s if he doesn’t not win MVP.

Defender of the Year: William Wettengel

Contenders: Justin Burnett, Jeff Babbitt, Brett Hulsmeyer

The rookie Wettengel recorded the second most blocks in the league at the third highest rate (per opponent possession), which is a winning recipe for Defender of the Year. Ambrose, the league leader in blocks, falls just out of contention here because of his league-leading number of opponent possessions, leading to a lower rate, but I’m sure others will not see that as so relevant.

D-Line Player of the Year: Ben Jagt

Contenders: John Randolph, Lukas Ambrose, Max Trifillis

Jagt is the kind of break engine this award is about: get blocks (5th in EDGE-B) and motor an offense (11th in EO100). That Jagt’s teammate Randolph is close behind says a lot about why the Empire are undefeated. Max Trifillis had the kind of block:turnover ratio one likes, the kind that doesn’t compute: 14 blocks, 0 turns.

All-AUDL

I’m going three teams deep, with each team including the top remaining Defender, Break Generator, and offensive player, and then filled in with players from the Ez list. The Third Team includes eight players because Khalif El-Salaam broke the system; I kept the game threshold low in order to include as many players as possible, and assumed that in using “EDGE per team game,” any player with fewer than 10 games or so would simply have season totals too low to viably contend for All-AUDL, which really should be limited to players who play at least ¾ of the games. But Khalif, despite only playing seven games, was so prolific that he ended up 12th in Ez. I can’t deny him the recognition he earned fairly, so I’ve just added an additional spot for Max Sheppard, another player who excelled on a weaker team.

First Team

• Brett Hulsmeyer
• Ben Jagt
• Pawel Janas
• Jordan Kerr
• James Lewis
• Ryan Osgar
• William Wettengel

Second Team

• Alex Atkins
• Justin Burnett
• Sean Connole
• Sean McDougall
• Jacob Miller
• John Randolph
• Brandon Van Deusen

Third Team

• Lukas Ambrose
• Jeff Babbitt
• Khalif El-Salaam
• Jonny Malks
• Zeppelin Raunig
• Max Sheppard
• Matt Smith
• Evan Swiatek

Rookie of the Year: Walker Frankenberg

Contenders: William Wettengel, Lukas Ambrose

This is a much more legitimate exercise this year, when there are not three new and good expansion teams to fill the list with a number of experienced players. This year, it really is among a set of young players, and it would seem logical to bestow the title on Defender of the Year and All-AUDL candidates Wettengel and Ambrose. But we’re going with the highest-ranking rookie on the Ez chart, and that’s Frankenberg, at a very impressive 24th. For the rest of the All-Rookie team, I’ve pulled the top rookies from among the Ez, Defenders, and Break Generators lists.

All-Rookie Team

• Lukas Ambrose
• Tanner Barcus
• Walker Frankenberg
• Max Gibson
• Liam Haberfield
• Saeed Semrin
• William Wettengel

As I said last year, the margins are thin, and moving players up and down based on alternative criteria or weighting is expected, including how much weight one puts on a team getting to the playoffs or Championship Weekend. But I’ll say this: if Championship Weekend plays out roughly the same way as the season up to this point, and Ryan Osgar is not MVP, I’m calling snub.

Full Division Championship Week Table