Games Started All Stars

Earlier this week, I wrote about the importance of looking at a hitter’s performance on a per-game-started basis, particularly if you are in a daily lineups league. There are real risks to considering a hitter’s overall line without considering what they do when they start, and those risks can cut both ways. If you look at their rate stats and dream on the value of that .350 wOBA, you might miss that they lose plate appearances to pinch hitters, which might cost them 10% or more of their PA, even when they start.

If you look at their per-game stats, you might underrate a player’s power output or stolen base contribution because those numbers look bad overall, but they are actually on pace for 20/20 over 160 games. Today, we’ll account for those risks and look at a set of players whose overall and per-game performance might be hiding how good they have been on a per-game-started basis.

I am using Ottoneu FanGraph’s Points scoring for this exercise for two reasons:

1. Points leagues are arguably where this matters most. Whether it is Ottoneu or ESPN or Yahoo, when you look at a player in a points league, the first numbers you see are total points and average points per game, and those can be super misleading – if you have sorted by P/G on Ottoneu or Average on ESPN, any player with even a few appearances as a pinch hitter, pinch runner, or defensive replacement is going to be further down the list then they should be. On the other hand, a player who is being PH for often, like we saw with Michael Conforto and Kerry Carpenter in the last article, might lose 10% or more of their PA even when they start, which will, in turn, decrease their P/G by 10%, as well.
2. Points are a good shorthand way to find overall fantasy value. If you play in a 5×5 roto league, you’ll value steals more and OBP less than these scoring systems do, but there is enough overlap that if a player’s P/GS looks good, their roto value per game started probably will, too.

I pulled data from Baseball Reference’s Stathead, because they enable you to find “as starter” and “as substitute” stats. I started with a list of 366 players who have at least 100 PA as a starter, and then calculated their total points and points per game both as a starter and overall. I then compared their P/G overall to their P/GS, as well as their rank among hitters by P/G vs. their rank among hitters by P/GS. The table below shows the 17 hitters whose P/G is below 5.0, but whose P/GS is above.

I actually want to start at the bottom, because there are some names we can move through relatively quickly. Brandon Nimmo and Heliot Ramos are on this list almost more for optics than anything, as the gap between their P/G and P/GS is pretty small, but those P/GS numbers feel different because they start with a five. A.J. Ewing and Isaac Paredes have slightly bigger gaps, but they share one thing with Ramos and Nimmo: The P/GS number is bigger but is arguably worse.

The driver behind this is that P/GS is higher across the board. For these 366 bats, the average points per game is 4.11 the average points per game started is 4.53. When players come in as a sub, they get fewer chances to accrue points or other stats. It is possible to have a small number of really effective PA as a sub and run a higher P/G than P/GS, but it is rare. Only 25 of the 366 players in my sample are better on a per-game basis than per-game-started.

So Ewing, Paredes, Ramos, and Nimmo all look absolutely better but relatively worse when we consider them on a P/GS basis. The caveat to that is, the difference is small. Are you really valuing Ewing less because he is the 106th best bat instead of the 102nd? Even the 13-spot gap for Ramos or Nimmo feels small. But it’s worth noting – they appear on this list of potential Per Game Started All Stars and they really don’t belong.

I would make the same basic argument for Contreras, Clemens, Gonzales, and Gelof. Yes, they score more per game as a starter than you might notice just looking at a player page. Yes, they move up the ranks, as a result. But the shift in where they are ranked is pretty small.

And that leaves us an All Star team of nine players. It’s not an All Star team you would want to put on the field. We have two catchers, four first outfielders and three first basemen. Some of them can moonlight at other spots, but it is telling that there are no middle infielders on here. The main reason is that middle infielders are just less likely to be platooned and less likely to be used as a PH. They are more likely to be on the field for their defense, platoon splits or no, and they are less likely to be good enough bats to warrant use off the bench on a day off. The grass and the corners are just better places to find bats whose strong performance is hidden by their per-game numbers.

Despite being at the bottom of the group of nine on this table, Trevor Larnach is probably my favorite of the bunch. He has a few things going for him that not all of the rest do. While 2026 is on pace to be a career year for Larnach by wRC+, he has a track record of some success, including three other seasons with a better-than-average wRC+ and a career 108 wRC+. I think people tend to view him as a disappointment, but he hasn’t been bad at all. Partially as a result, his rest of season projections all have a wRC+ between 106 and 113 – yes, you should expect some regression, projections tell us, but that is still a useful OF.

He is something of a points-league or Ottoneu 4×4 specific player, though. He doesn’t run. He has 12 career steals (two this year) and has been caught eight times (twice this year). He looks unlikely to get to 15 HR, let alone 20+. His AVG and OBP are great, and his SLG isn’t bad, but he just isn’t doing a ton beyond that. But he is probably undervalued in points leagues and even that HR total is a little misleading. If you started Larnach or a Larnach clone in 162 games in the OF, you would be looking at 16 HR. That’s not world-beating, but it’s a lot better than the 12 he seems destined for.

Mitchell is probably the favorite of some of you reading this list, and that is totally fair – my preference for Larnach is, in part, that I think he is flying further under the radar and therefore easier to acquire. Mitchell strikes out a ton (32.9% this year, 33.5% for his career) but hits the ball very hard (51.2% hard-hit rate this year, 43.8% for his career). He keeps the ball on the ground at is something of a BABIP merchant, but as he nears 800 career PA with a .390 career BABIP, maybe we shouldn’t expect too much regression. He offers similar power to Larnach (one more HR in four more GS) but adds a little bit of speed — though with 6 SB and 5 CS, maybe he shouldn’t keep trying to add speed? And, like Larnach, his 12 games as a substitute are skewing the surface level view of his points league scoringa and other counting stats. He looks like he’s barely on pace for 10 steals, but on a per-game-started basis, he is closer to a 15-steal pace.

Canzone came up in the comments of the precursor to this piece, as he has gotten some more looks against LHSP recently. That could get him to break out of this group and would cause his P/G to slowly, over time, creep up towards hi P/GS. That will reduce any discount you can get on Canzone, as it would make his surface numbers much more appealing. As you can see in that comment section, I am unsure that Canzone is, in fact, breaking out of his platoon role, but it almost doesn’t matter. He’s an interesting buy even if his role never changes. If you can pay for a 4.43 P/G bat or a guy on pace for 23-25 HR per projections; you would be getting a guy who has been putting up 5.67 P/G when he is in the lineup and who is hitting homers at more like a 35 HR pace.

France and Lowe I am lumping together because, to be honest, I am just not that into them. Lowe had a great start to the year, but even his 5.12 P/GS isn’t all that exciting for a 1B, and he has had a below average wRC+ in both June and July, so far. I’m not inclined to believe April and May are coming back, and even if he settles in somewhere between the torrid spring and the tepid summer, you aren’t getting much to be excited about. France’s per-game-started numbers look better, but he has also cooled off and his xwOBA suggests he’s more the guy he has always been than the guy he has looked like so far this year.

Sánchez looks quite good here, but his usefulness is really league-dependent and, to be honest, might not extend outside Ottoneu. He doesn’t play that often, having only 12 starts in the last month. In Ottoneu season-long leagues, where you have two catcher spots but only need to fill 162 games, you could plug him in as your C2, collect that per-start performance, and be fine. In most formats, however, if you are counting on him in your lineup, you are getting zeroes far too often. And if you keep him as a bench option and start him when he plays, you are using up a precious bench spot on a part-time catcher who might not even be in the lineup the day your primary catcher sits and you need him.

If you do want a catcher from this group, Rushing is the obvious choice, given better performance and better role (20 starts over the same stretch that Sanchez had 12). The thing to note with him is that projections expect heavy second-half regression. Not enough to make him a cut or a problem for your roster, but enough to make continuation of his current performance look a little optimistic.

That leaves the two most extreme examples on this list. Andrew Vaughn goes from an unusable 1B by P/G to a star by P/GS. The risk with him is that the sample size is so small. We are still at just half a season, so all samples are small, but due to injury and role, Vaughn has just 55 games played and just 38 as a starter. If you want to bet on him maintaining this small sample performance, you can use his 2025 numbers to support your cause. Ever since he moved to Milwaukee, he has been this guy. But you also need him to be in the lineup to help. Do you want to roster a 1B — even a star-level 1B — who has only five starts thus far in July and only 15 in the last month? I am not sure how useful that is. In Ottoneu or other formats with deep benches, I could see holding him if you don’t have a better option at 1B. Using him half the time could be a big boost. But if you don’t have a ton of bench room, that’s a hefty price to pay for a part-time 1B with both injuries and poor performance tempering his track record.

Lastly, we have Mangum, whose performance goes from “unrosterable even if he got catcher eligibility” to “hey, that’ll play!” when you only look at his starts. Mangum has a number of things working in his favor. First, he is playing daily these days. He has started all but three games since June 7. Second, he has seemingly taken over the leadoff spot in Pittsburgh. He’s hit first and only first since July 2. Yes, that is only because Konnor Griffin hit the IL, but he won’t be back to reclaim that spot anytime soon. Third, he puts the ball in play, puts it on the ground, and lets his legs do the work, allowing him to post high BABIPs and making him a solid play for both average and OBP. Toss in a ton of speed (18 steals despite not being a full-time player all year) and you have a recipe for a solid 5×5 player who is probably undervalued right now.

With 65 games left in the Pirates season, a fantasy manager looking at his season to-date (76 games played) and assuming Mangum will play 60 the rest of the way might expect this line (prorating his season-to-date to 60 more games) the rest of the way:

• .310, 28 runs, 1 HR, 15 RBI and 14 SB

But if we assume 60 more starts at his per-game-started pace, we get this:

• .315 (his average as a starter), 34 runs, 2 HR, 17 RBI and 19 SB

That’s quite a bit better. Now, the big caveat to this is that those numbers all assume he just keeps doing what he has been doing. Projections aren’t buying that, but the drop in performance they foresee is heavily tied up in a big drop in BABIP, from his current .371 to something between the .320s and .340s. Mangum, however, has .354 career BABIP in 681 career PA and a profile that fits the bill for high BABIPs. Projections are always going to be conservative on BABIP, but this is a moment where I might bet on the career number rather than the projected number. So, yes, some regression, but not as much as you might think.

Oh, and we haven’t even accounted for the fact that he could easily get an extra 10% more PA per game the rest of the way by leading off instead of hitting 7th, 8th, or 9th. Given that, I think he is a pretty fun pickup in 5×5 formats, though limited power and the lack of value in his speed tempers my excitement in Ottoneu (outside 5×5).