The result would still not be comparable, because if your players are tired in overtime you will get a lower mean in OT games than in regular length ones, only the difference will be even slightly harder to understand.
I agree with what you're saying, I think. The difference in our stances:
- Me: The OT ratings should try to present what people think they measure (a rate), even if they're a little bit inaccurate
- You: The OT ratings do measure something accurately (an integral), so better to leave it as it is, rather than present an inaccurate derived measurement.
I think I'm more sympathetic to your position now.
I still think when I calculate so-called bbstat ratings in an automated way, I'll just naively reduce the ratings by a factor of 48/(48 + OT minutes).