You will be wanting to read "Numbers, prediction, and war: Using history to evaluate combat factors and predict the outcome of battles" by Dupuy. You'll have to find a copy at a used-book dealer because it was published in 1979.

Awesome book. Controversial then and now, but still an awesomely comprehensive approach. He adopts the approach of working out what the "+X for a defensive position" type numbers ought to be, extracting data from history and then using the outputs to retroactively predict historical results and get the right answers.