Replace P/G with P per Start

Now that we have the lineup eject button (which has been working great for me so far! Thanks for that Niv!) I feel like P/G is a particularly irrelevant stat. Can we instead get P/S (Points/Start)? P/G unfairly penalizes pinch hitters. Between P/S and Total Points, a reasonably adept user should be able to see the whole picture.

Open to feedback on this. I will say I am not 100% confident we have historical GS data for position players.

My two cents is that there are really only a handful of Ottoneu-relevant players for whom this is an issue (i.e., they make appearances as pinch hitters with any regularity) and those players tend to be platoon players who can easily be identified.

For example, Joc Pederson often sits against LHP starts but will later appear as a pinch hitter, although he might stay in the game and bat a second time. So P/G would tend to underestimate the value that he brings in those starts.

So it seems to me that if you were to switch from P/G to P/GS for position players, the measure would go from underestimating his contributions to overestimating his contributions. For most players, it’s not going to matter. But for the handful of Ottoneu-relevant players where it could matter–like Pederson–you’re not really improving the situation; rather, just changing the direction of the imprecision.

For example, with Pederson (5.13 P/G in 2019), if you simply divided up his total points (764.2) by games started (109), then you would get 7.01. But this is an overestimate, as I’ll show with the following set of calculation.

What I do to analyze a player like Pederson is first figure out PA/GS and then estimate P/G using the following formula (this was posted on Slack a while back, not my own creation):

P/G=(7.25 x wOBA-1.16) x PA/GS

Note that this formula does not include points from SB and CS, but for most players that’s negligible.

In 2019, Pederson had a wOBA of .362 and started 109 games with 456 PA (note: you need to get this from BBRef’s splits, not FG’s). So his PA/GS is 456/109=4.18.

Plug that into the formula to find his P/G as a starter:


You could also simply input the actual stats and calculate the exact points, but the formula based on wOBA and PA/G is close enough, IMHO.

So we have 5.13 (total points divided by total games played, i.e., what Ottoneu presently provides), 7.01 (total points divided by games started), and 6.12 (estimated points scored as a starter divided by games started). Essentially, what Ottoneu provides is a lower bound on P/G, but what the OP is suggesting would represent an upper-bound.

As I mentioned, there are very few Ottoneu-relevant players where this is a substantive issue (i.e., Pederson is exceptional for nearly a quarter of his games played being as a PH, among Ottoneu-relevant players). And if you’re going to use platoon bats, then you should put in a little work to figure out if/when they make sense to be used. That is, I certainly wouldn’t complain if Ottoneu presented additional data (i.e., the points accumulated as a starter divided by games started), but I’m not sure it’s absolutely necessary and I certainly wouldn’t want it to replace the standard P/G.