I agree that BA RISP can be misleading. But this stat is even worse. The biggest problem is that it doesn’t distinguish between RISP and runners on first (RO-1). In most cases a single will drive in a runner from second or third. It will almost never drive one in from first. Thus “success or failure” here depends heavily on where the runners happen to be located when the batter comes up. It also depends on the speed of the runners, the number of outs, etc. These are all factors completely outside the batter’s control.

Walks are another problem. It’s hard to drive in runners when you get walked (or at least pitched around) whenever you come up with runners on. Just ask Barry Bonds (if you can ever find him).

Sacrifices create a similar problem. Batters often asked to bunt will do poorly with this stat.

How can the stat be improved? Using AB’s instead of PA’s would solve the BB and SH problems, although you’d have to add SF’s to the AB’s.

The only way to deal with the main problem is to compute two separate OBI’s: RISP OBI vs. RO-1 OBI. Of course you may want to go one step further, splitting RISP OBI between RO-2 and RO-3. And you could make a case for splitting RO-3 by number of outs (2 or less). But at that point the data gets a little overwhelming.

Let’s stick with WPA. Not only does it tell us how often a batter takes advantage of RBI Opps., but it also indicates how important those RBI Opps. were to the outcome of the game.