"OAT has a higher delay, which somewhat offsets it's lower rounds per weaponskill. Overall weaponskill cycle time is 1095 delay for OAT, compared to 1400 for str polearm. So OAT weaponskills 28% more often than Str polearm."
This is where your problem is. There is no way, not a chance, that OAT is only WSing 28% more often. Yes, the longer delay offset the hit build slightly, but couple the lower hit build with the 40% OAT it should at least be 35% more often if we account for TP overflow. Not to mention you get 40% more TP from jumps as well.
Also, if conserve tp is such an important factor and that 5 hit is seemingly not as big of an advantage, you should try putting this in your spread sheet:
Using sword strap for OAT polearm and do a 6 hit. That way, you can better use your conserve tp and should allow for a better comparison.
Like I said, there is literally no way OAT only WS 28% more.
PS: You also need to account for extra TP making OAT polearm WS harder (even if just slightly). All of those conserve tp proc that didn't help OAT cut a hit from the hit build should result in 15+ extra TP on its WS as well as any TP overflow from OAT proc with 1 more hit to WS. I am not doubting your spread sheet since I know you from BG, but there has to be something that was not set correctly cuz I have parsed against STR polearm before and the result wasn't close.