The first thing that comes to mind: "How do you skin a cat?"
The next thing that came to mind was ... "depends how you have your data represented". If you have all the games compiled in say a spread sheet. You could filter on various columns... ( Assuming you have game odds, and the scores).
Getting rid of (filtering) games where:
1) the home team was the underdog (homeMoney Line > awayMoneyLine)
2)the home team didn't win (homeScore < awayScore )
( the order wouldn't matter )
Take the number of resulting rows... and divide it by the total (unfiltered) number of games.
if its a database...the sql could look like this:
select count(*) as HomeGamesWonAsFavorite
from Game
where (HomeMoneyLine < AwayMoneyLine) and (HomeScore > AwayScore)
and (Game.Season="Regular Season" or Game.Season="PostSeason")
Does that help (and make sense)?