Anyone want to chime in on the mathematical soundness of my NBA simulating model

My NBA simulator model has been doing well, enough to turn me a profit using it to bet, but I was still unnerved by the fact that at least on totals, it was skewing too heavily toward the under, by an aggregate of about 1.25 points.

So what I did is that in my stat calculations, I switched over to the basketball-reference.com style of calculating possessions(have been using the stats.nba.com/ESPN's style of calculating possessions, which is FGA + (FTA * .44) - ORB + TO). Basketball Reference is a bit different, 0.5 * ((Tm FGA + 0.4 * Tm FTA - 1.07 * (Tm ORB / (Tm ORB + Opp DRB)) * (Tm FGA - Tm FG) + Tm TOV) + (Opp FGA + 0.4 * Opp FTA - 1.07 * (Opp ORB / (Opp ORB + Tm DRB)) * (Opp FGA - Opp FG) + Opp TOV)).

So while this made game possession calculations more accurate, it also means my simulator needs to take into consideration a couple of things. One, this style of calculating possessions tries to factor in the team offensive rebound(something that doesn't show up on a box score). It uses 1.07 * ORB to calculate the team rebound. Second, it changes FTA to 0.4 for a possession usage, down from 0.44 on the previous method.

So I changed it up and was still getting too low totals, and I realized two things, my simulator was still using FTA as a full possession usage, and was not factoring in the team rebound. So in my simulator I decreased the running possession count(the simulated "game" ends when the running possession count is greater than the estimated possession count for the game) for free throws from 1 to 0.8(0.4 per free throw times 2). This helped some but was still short.

Then I realized that FGA attempts were the culprit of the low totals, as they were too low based on expectancy. So for every offensive rebound that occured in the simulation, I reduced the running possession counter by .07(to factor in the team offensive rebound). This also helped, but the total was still just a bit too low for my liking.

What I finally did was realize that box score turnovers and season team stat turnovers are not equal because of the rare "team turnover". For example if you add up the Hawks total turnovers individually from the box scores for all their games this year, you get 628, but their season turnover total is 647, meaning they have 19 team turnovers that didn't show up in box scores. So what I did was multiply their game turnovers by the percentage increase of team turnovers (3.025%). This was necessary because my possession and tempo calculations use turnovers in the equation, so naturally being short on turnover totals was causing tempo numbers to also be a bit too low. This seemed to do the trick and now the simulated totals in a macro sense are more in line with the posted totals.

Is there anything else I need to be factoring in here? I factor in back-to-backs, use both dynamic home advantages and static home advantages, but nothing with injuries.