In general, wouldn't it make sense to have different formulae for different lifts?
For example, the OH Press vs Bench. I figure given an accurate 1 RM, I can kick out more reps at 80% of that on Bench than I can for OH Press.
The formula in Wendler's book (which is not original to him) can be generalized to:
1RM = Weight * (1 + Reps/k)
Where the value of 33 for k is pretty common, and is meant to hit the middle of the bell curve for general fast/slow twitch mix. For those with generally more slow twitch, who are better at endurance, you want to predict a lower number, so k goes up, making each rep worth less in predicting the 1RM.
This formula works pretty well for me on bench and squat, suggesting I'm average. It falls off for dead and press, in opposite directions. Not surprisingly, it overpredicts for me for Press, which involves smaller muscles than bench and squat. For dead it undepredicts, which also makes sense because there are more fast twitch muscles working there.
That's all I know.