At least as I read your posts, the differences between the non-proprietary models won't be an obstacle, so use BS if you have reason to prefer it. For proprietary models, you will have to have solid historical data, but again, I'm not convinced that spending time modeling such will be very revealing, unless your book is huge. I think with relatively little effort, you can come up with a scheme that gives you a satisfactory risk profile adding static long puts to a portfolio of equities. Now if you want to trade around that inventory, then that's where this whole business gets interesting.
Thanks for the insights, sonoma. Yeah, I was hoping that I could just use the BSM to generate the ATM put prices on the fly during backtesting, figuring that the differences between the actual and modeled prices would just average themselves out over time. I could also use a larger risk free rate and/or volatility to bias the premiums higher.