Looking at Fantasy Stats Per Games Started

Fantasy baseball is a stats game, both in terms of scoring (you are literally trying to maximize certain stats) and in terms of building a roster (you use stats to figure out who to roster). But if you play in daily lineups leagues, almost all of the stats we look at are presented in the wrong format for your purposes, particularly for hitters.

When you check out Michael Conforto’s player page, you will see he has 8 HR and a .243 average in 160 PA, which puts him on a 30-HR pace for a full season. You’ll also see him at 3.44 P/G in Ottoneu points leagues. But if you have been moving him in and out of your lineup based on when he does/doesn’t start, and you look at what he has given your team, you’ll see something more like 5 HR and a .223 average, with 4.06 P/G for Ottoneu points leagues. Why has Conforto been better for your team in points leagues (though, to be fair, still not good) and worse in roto?

His starter/substitute splits.

Conforto is an interesting example because he demonstrates multiple things that can be going on.

First, his overall numbers, on a rate basis, look good. Points per PA isn’t a stat we look at much, even in Ottoneu, but 1.33 is a strong number. A .243 average isn’t hurting you in 5×5, and prorated to 600 PA, he is on pace for 30 HR, 75 R, 82 RBI, and 4 SB, which is a very solid 5×5 line. No, he doesn’t play every day, but if you can plug him on when he starts and get even 50% of that, while playing someone else the other days, that’s pretty good.

Ah, but there is a catch here! All of those numbers assume that when he starts, you get a full starting OF workload from him. But you don’t! A hitter’s PA/G can vary by spot in the lineup, but among players with 100+ PA as a starter, the average PA per game started is 4.08. For the top 50 hitters in that group by OPS, it is 4.30. Conforto is only at 3.66 PA/G even when he starts, largely because he gets pulled for PH.

The impact of his limited PA/G is easy to see in the Ottoneu points scoring. It’s less clear in roto, but it is still there. I mentioned above that 1.33 P/PA is a good scoring rate, but that’s only true if you get enough PA. You can see that if you take Conforto’s P/PA and compare the P/GS for Conforto vs. the average player with 100+ PA as a starter or the average “locked on” starter (using the top 50 by OPS mentioned above as a proxy):

Conforto’s 1.33 P/PA overall, prorated to his PA/G as a starter, gives you a solid 4.87 P/G, which isn’t bad at all. Not a guy you have to have in your lineup, but a good option for sure. If instead he were getting an average number of PA for a starter, he would be a very solid regular starter (Michael Harris II, Cole Carrigg, Brandon Marsh, and Jake McCarthy are the four OF-eligible bats between 5.36 and 5.45). If he were getting that “locked on” starter playing time, he would be in a class with Ryan O’Hearn, Mickey Moniak, and Ronald Acuña Jr.

For roto, it’s harder to see, but getting about 10% fewer PA even as a starter means you have to discount those “per 600” numbers. At 4.08 PA/G, a starting hitter gets to 600 PA in about 147 games. Conforto over 147 games would get to 538 PA. If we prorate him to 538 instead of 600, he still has a .243 average, with 27 HR, 67 R, 74 RBI and 3 SB. That isn’t a ton worse than the per-600 numbers above, but it is worse.

The other thing that jumps out of that first table is that Conforto is just crushing when he comes off the bench. Given he has 4x as many PA as a starter (128 vs. 32) you would expect him to have 4x the HR, as well as 4x the R and RBI, and a similar average. But no. As a starter his average is nearly .100 points lower than as a sub. He has less than 2x the HR. He does have more than 4x the R, but not RBI.

There isn’t a ton of reason to assume this is predictive and that Conforto will continue to dominate off the bench. But he is being used as a PH strategically – against RHP, perhaps against RHP he matches up well with, etc. So we also can’t just assume that his overall numbers reflect what he will do as a starter.

And when you look at what he has done as a starter and prorate that to a full-season for him (538 PA in 147 games), you get that .223 average, 21 HR, 71 R, 71 RBI, 2 SB, and a .223 average. In Ottoneu points leagues, you get 4.06 P/G. That’s a far cry from the numbers above. In points leagues, that is an easy cut. In 5×5, there might be some value there, but not much. The player rater gives 2025 Adolis Garcia a $3.40 value, for 19 HR, 75 R, 58 RBI, 13 SB, and a .227 average. But those extra 11 SB are worth more than enough to suggest that Conforto isn’t adding value.

The truth probably lies somewhere in between prorating his as-starter splits and prorating his overall line. The overall line is a slightly larger sample and those are real PA, so we shouldn’t ignore them. But those sub PA are a skewed sample and less representative of the type of pitchers and situations he faces as a starter.

Conforto is just one example, and his overperformance off the bench makes him something of a unique one. Let’s quickly look at one other player – Kerry Carpenter.

Carpenter, like Conforto, suffers from a PA/G penalty, even as a starter. And – perhaps unsurprisingly given the propensity of Tiger’s manager AJ Hinch to manage aggressively – Carpenter suffers a worse penalty. His PA/G as a starter are even lower than Conforto’s, and he is less likely than Conforto to come in as a PH and get a second PA in that game. But his P/G as a starter is better than Conforto’s even with fewer PA/G.

Carpenter, however, isn’t impacted by his PH appearances the same way as Conforto. Conforto’s overall line makes him look better, especially in roto, because he has overperformed as a sub. Carpenter (more typically) has struggled off the bench, so his overall line – already deflated by limited PA/G – is even further deflated by poor numbers coming off the bench.

It’s tempting to look at 13 HR and 33 RBI in 231 PA and dream on Carpenter as a guy who pushes 40-homer, 100-RBI value prorated down to a platoon bat share. And, unlike Conforto, Carpenter’s HR do come mostly as a starter. But due to his limited PA/G (even as a starter), you can’t really prorate him to 600 PA.

If you want to prorate him to a full-timer’s share of games (147), you get to 501 PA. Prorating his starter performance to that 501 PA, you get 25 HR, 60 R, 80 RBI, and that .225 average. Or you get 4.14 P/G, in Ottoneu. That’s a better line than we get prorating the starter-version of Conforto, but it’s not 40/100.

The point of all of this is not that Carpenter is more valuable than Conforto or that neither of them are particularly valuable (especially in points formats where being pulled for a PH hurts a lot). The point is that when you look at a player’s season-long line, you need to consider how and when they are used. And how and when you will use them. If you are in a weekly league, the season-long line is a fine thing to look at, but you need to know you can’t replace the lost PA – these guys are either in your lineup for the week or they are not. If you are in a daily league, you can replace those lost PA, but the season-long line is going to be misleading, and it’s not as simple as “prorate this to 600 PA” or “what would they be doing in 4 PA/G as a starter.”

Later this week, I’ll look at some players whose performance as a starter means they deserve a second look.