Using ATC’s Inter and Intra Measures to Evaluate Hitters

2023’s PitcherList PitchCon was a great time. If you are unfamiliar with the event I highly recommend going back and watching any of the videos from any one of the four day’s recordings. FanGraphs’ own Ariel Cohen presented a few aspects of his ATC projections that I had never been aware of and think could be very useful in draft preparation. I encourage you to go back and watch Cohen’s talk on The Value of ATC Volatility Charts. In this post, I’ll take a look ATC’s InterSD, InterSK, and IntraSD metrics. Join me by opening a new tab, clicking on over to the ATC projections page, shuffling to the “Fantasy” tab and noticing three columns all the way to the right of the spreadsheet.

–

InterSD

ATC projections are known for a “Wisdom of the Crowd” and Many Model Thinking approach. This means that ATC utilizes a wide range of projection systems to project player performance. Some of these projection systems, we don’t have posted here on FanGraphs simply because there are, nowadays, so many unique systems. If you are going to source the wisdom of the crowd, you need to have a big enough crowd. Because of this, you won’t be able to replicate ATC’s “InterSD”. Trust me, I’ve tried. We aren’t made aware of exactly what systems ATC uses and how much each one is weighted. But, what we can do is use the projections we have on FanGraphs to exemplify how the measurement works. Joey Meneses is a good example. Go to his player page and you’ll notice big discrepancies in his projections, specifically looking at THE BATX, Steamer, and Depth Charts:

Using those same projection systems to determine an auction price can further prove there are some big differences in expectations:

THE BATX: -$1.48
FGDC: $6.36
Steamer: $15.40

Steamer and FGDC are higher on offensive output and Steamer is higher on playing time. Joey Meneses has a high InterSD using only these three projection systems and while this is an oversimplified version of what ATC’s InterSD accomplishes, the ATC version picks up on it as well. This tells us that Meneses is a difficult player to project and that various systems differ in his value. While it would be best to draft players on whom the crowd agrees, you may be able to use InterSD to identify sleepers with who projection systems just can’t agree. If anything, sorting on InterSD can really make you think. Below, I’ve hand-picked a few players who I found interesting when doing some sorting of my own. You’ll notice that younger, unproven players with low PA projections tend to have high InterSD, and rightfully so. For this exercise, I focused on players I thought were more fantasy relevant with higher PA projections. Here are three players with high InterSD, high variability in projected value, and three players with low InterSD, low variability in projected value:

High InterSD

Tommy Pham – 8.1
AJ Pollock – 6.4
Yuli Gurriel – 5.70

What do you notice about these three players? Ding, ding, ding! They’re getting old. Pham, age 34, Pollock, age 35, and Gurriel, age 38, are all experienced players who father time is reaching for. On one hand, they have showcased skills to justify rostering on fantasy teams in the past. On the other hand, their skills could be declining, they are each joining new teams (though Yuli Gurriel remains a free agent as of this writing), and their playing time is even harder to pin down than usual. It’s not surprising that projection systems disagree. Not only are age and role in question for these players, but they each have struggled with injury and haven’t necessarily been consistent in recent years. Take a look at Yuli Gurriel’s stats in the past few seasons:

After a down year in 2020, Yuli rebounded in 2021 (though his BABIP was .336) and showcased why he’s a legendary hitter. Yet, he did not repeat in 2022. He did steal 8 bases. Seriously, he did. Yet, consider his rising K% matched with a career-low SLG (.360) in 2022 and the fact that he is still unsigned and you can see why projections differ.

How about Tommy “Smack You During Batting Practice” Pham? You won’t be able to identify the discrepancies in Pham’s projections by looking at our site projections. They are all pretty much in alignment. But, the whole story isn’t being told on his player page. Pham stole 25 bases in 2017 and again in 2019, but slowed to only 8 in 2022. His O-Swing% is up, his O-Contact% is down, and his SwStr% and CStr% are both up from 2021 to 2022. But, here’s the kicker; his HardHit% increased, his Pull% increased and he’s maintained a 2022 MaxEV (112.6) in line with his career average (113.3). But how often will he play in New York? If you can answer that accurately there are a lot of projection systems that could use your help.

No love for A.J. Pollack? He accumulated 527 plate appearances with Chicago in 2022, his highest since 2015 with Arizona. But his ISO was down, his SLG was down, his wOBA was down, and he finished the season with a 92 wRC+. Right now, Roster Resource has him lined up in the DH spot, but his PA projections differ quite a bit.

Low InterSD

Amed Rosario – 1.2
Elias Díaz – 1.3
Bo Bichette – 1.3

Rosario and Bichette are projected by most to have stolen base totals in the mid-teens, accumulate over 600 plate appearances (well over in Bichette’s case), and hit above .270. These are two very projectable players. They’ve proven their skills and we know what to expect. Keep in mind that having a low InterSD doesn’t mean you should draft that player. Elias Díaz is a 32 year-old catcher projected to hit somewhere around .250 with an OBP just over .300. I think he’s great as a second catcher in two-catcher leagues and InterSD says what you see is what you get.

–

IntraSD

Think of a player who is a one-category player. Who do you come up with? Sort by IntraSD and you’ll find a perfect example, Adalberto Mondesi. With a projected 30 stolen bases by ATC, Mondesi is appealing. Yet IntraSD tells us that his value is coming from only those stolen bases. His ATC .221 average, 10 home runs, 38 RBI and 46 runs explain why. IntraSD looks at a player’s ATC projection and tells us how balanced or unbalanced a player is. The flip side of Mondesi? Austin Meadows. Now, you may think that seems odd, but remember that IntraSD doesn’t tell us about value, it tells us how equally distributed a player’s category values are.

–

InterSK

InterSK is probably the most useful of these three measures. InterSK tells us the skewness of projected player value. A player with a negative skew has their value being brought down by one outlier projection. In other words, most everyone agrees that the player will be good, yet there’s one system that’s low on his value and that outlier can explain why his ATC value is lower than others. This is advantageous because it can help us identify players who the crowd generally agrees on yet could have a doubter that points out things others may have missed. Here are three players with a negative skew in their projected value:

Negative skew
Take a look at Adley Rutschman’s player value based on the default settings of our auction calculator:

ATC: $11.8, Steamer: $14.4, FGDC: $10.3, THEBATX: $12.3

While these four projections don’t fully explain ATC’s InterSK for reasons explained earlier, it does help. If we considered ATC’s $11.8 valuation as the average, then FGDC’s valuation is pulling that average down. No one is questioning Rutschman’s skills, but perhaps systems are low on how many runs he can score in Baltimore’s lineup. Yes, the O’s are on the up and up, but some are probably higher on the team’s offensive abilities than others.

It’s hard to tell why Jeremy Peña’s value distribution is negatively skewed. Most systems on his player page are closely projecting over 20 home runs and over 10 steals. I love it. This is the type of player the InterSK really works for in my mind. Perhaps there’s a projection system out there that questions where the power comes from. Peña’s 2022 HardHit% was below average, he put the ball on the ground more often than average and his lower-than-league-average launch angle helped explain why. Yet, he’s a young player who may just be a tweak or two away from taking a step forward.

Someone out there doesn’t believe in Rowdy Tellez repeating a 30+ home run season? I’m ok with that. Pulling down his value will only help me add him to my roster in more leagues.

–

If you’ve made it this far, you can probably surmise what a player with a positive InterSK might have in their projected value distribution. These three players again have a common theme. A bullish projection is bringing the ATC value up. Beware of these players as almost everyone agrees on the lower value.

Positive skew
Maybe someone out there thinks a full season in Toronto will help Merrifield return to being a top fantasy player in 2023 while others see his wOBA trends and think otherwise:

Perhaps this high outlier projection saw something from Merrifield at the end of the season that justifies a return?

Sleep on Paul Goldschmidt. Sleep on Paul Goldschmidt. Sleep on Paul Goldschmidt. While many projection systems are calling for regression, I’m taking the upside of that one outlier believer. He’s never not been a good pick.

With two offensive catchers in Alejandro Kirk and Danny Jansen in Toronto, it’s hard to estimate how much Brandon Belt will DH. His injury history adds to that difficulty. Yet he’s another player with established skill and baseline. What if he’s healthy all year? What if Toronto’s park really benefits his bat? What if being off his feet in a DH role more often keeps him in the lineup? All these what if’s have led to an upside outlier. What a gift a $1 Belt could be!

Whew! This took longer than expected and if you’re still here, I thank you. Let’s sum this up:

– Players with low InterSD are projectable and projection systems are in agreement on their 2023 expectations
– Players with medium to high value and low InterSD should be targeted.
– Beware of high InterSD as these players are harder to pin down. Can you find a way to use that to your advantage? I’m sure you will try.
– Players with medium to high value and low IntraSD are balanced 5-category players who should be drafted.
– It’s probably best to avoid the rabbits with high IntraSD unless you really missed out on a category and need to make up for it late in drafts.
– InterSK allows you to bet with the crowd, or find players with upside. Someone out there believes in high/positively skewed players, do you?

Thanks to Ariel Cohen for allowing me to pick these measures apart and for generating such useful tools for fantasy baseball. Now go use ATC to build your roster targets!