The Z Files: Normalizing the ATC Pitching Projections

When we last were together, in The Wisdom of Crowds and ATC Projections, the ATC version of the Steamer 600 was presented. For those unfamiliar, the Steamer 600 assigns 600 plate appearances to each hitter, then prorates their Steamer projection accordingly.

Now it's time to do something similar to Ariel Cohen's ATC projections for pitchers. Doing so is tricky, since assigning the same number of innings to starters and relievers skews the counting stats. As such, all starters will be scaled to 200 innings with relievers adjusted to 65 frames each.

STARTING PITCHERS

Even with this adjustment, there is still an issue using conventional valuation, since the replacement numbers are skewed after the normalization. Don't worry, the projected earnings aren't integral to the analysis. The primary evaluation looks at how much each pitcher moves relative to the others after the adjustment. Starters will be compared to starters and relievers will be compared to relievers.

As was the case with the hitters, the elegance of this approach is everyone can look at the results differently to glean the information they feel is most relevant. The tables are sortable to aid in this endeavor.

As someone who does projections for (part of) my living, it's always a struggle deciding how many starts to assign to rotation members. Do I make it 30? 31? Over the last three seasons, this guy's started 30, 32 and 30 games. Hmm...

A difference of one or two starts between closely skilled pitchers can mean multiple spots in the rankings. A guy can be on a playoff team and have his last start skipped or truncated. Maybe someone pitches three frames and the game is suspended. Even worse, he warms up and the game is called off due to weather, and he ends up losing a start. There are many ways for 32 starts to end up as 30 or 31, with no fault to the pitcher.

Is it smart to strictly adhere to rankings based off projections, when the difference between 31 and 32 is a coin flip? Using ATC 200 identifies the pitchers offering the most fantasy potential on a per-inning (or per-game) basis. On my cheat sheet generated from projections, I note the ATC 200 ranking for the top 50 starters. Picks are still made in context with my roster, as it's a needs and risk profile, but having both styles of rankings helps guide the choice.

Zack Wheeler is the perfect example. He's No. 3 via standard ATC, and his March ADP in the NFBC RotoWire Online Championship is essentially a tie for third with Garrett Crochet. However, if Wheeler were to lose one or two starts, he falls to the ninth-best fantasy hurler according to the ATC 200. There is a reason Wheeler is No. 3 in the unadjusted ATC; he's started 32 games in three of the last four seasons, including the last pair. However, in 2022, he logged only 26 outings. He's going to be 35 years old this season with a lot of regular season and playoff mileage on his right arm. It's risky to pay for another 32-start campaign. Maybe take another pitcher instead? If there is not another pitcher worthy of the pick, grab a hitter and wait a round or two for an arm. On the other hand, the veteran has made 32 starts for two straight seasons. Fading him could result in losing 15 strikeouts and one or two wins, which combined could be a few roto points.

Some other pitchers relying on volume include Logan Gilbert, George Kirby (currently injured) and Framber Valdez, then we get to Corbin Burnes and Logan Webb. Burnes and Webb significantly depend on innings.

Burnes standard ATC rank and ADP are both 11, but ATC 200 plummets him 26 spots. Not only is his falling strikeout rate a concern, but he needs all of the 32.3 games started and 196.2 innings he's averaged over the last three seasons to just break even at his draft spot. That's a big risk.

Webb's 45-spot ATC 200 drop is even more dramatic, but his cost of acquisition changes the algebra. Via the standard ATC projections, Webb is the 10th-most useful starting pitcher, but the market recognizes the risk and ranks him 42nd. Is that enough of a discount to make him your SP4? Your team, your call.

Narrator: Todd thinks it is plenty, and is all over Webb as his SP4

Again, the beauty of this is you can evaluate each pitcher within the confines of your league, personal philosophy and risk aversion, using some of the approaches just presented (and your own).

Once you get past the "pick and stick" starters (those you're leaving in your active lineup all the time) and into the streamers, the ATC 200 is less useful. That may seem counterintuitive, since the ATC 200 essentially compares on a per-start basis, and that's what we care about when streaming. Having a feel for straight-up skills is important, but I want to embrace variance with my streamer.

There is an organic home field advantage for pitching.

This is park neutral. Using just ERA, a league-average Mariners pitcher should post an ERA in the 3.40 neighborhood, compared to a 4.70 mark for a Rockie. Of course, these are the extremes, but it points to streaming (hence drafting) starting pitchers with favorable home parks. On a week-to-week basis, matchups are paramount, but my primary filters for SP4-SP7 is quality of pitcher and favorable home park. The ATC 200 adjustment is further down the list. Your mileage may vary.

RELIEVERS

Confession time: I don't find ATC 65 for relievers especially useful. There is too much context driving the value, and draft slot, of closers. That said, I love doing the ATC 65 for relievers after removing saves and wins from the process. This isn't for drafting saves but rather for pinpointing relievers most adept at protecting ratios while helping strikeouts.

Setup men like Griffin Jax, Jason Adam and Bryan Abreu are routinely drafted as insurance policies. You don't need this version of the ATC 65 to help identify players of that ilk. Cade Smith, Matt Strahm, Jeremiah Estrada and Dylan Lee (and others) merit more attention when drafting relievers solely to support ratios while adding some punchouts.

Thus concludes the two-part series on leveraging the ATC projections

Questions? Comments? Fire away!