Note: The following remarks apply more to SPWW2 ( I forgot what forum I was on...
) but it also applies to MBT but not so specifcally for the Japanese or Germans in MBT. Still, the test I suggest can be applied to MBT but perhaps check the USA and Russia instead and double or triple the number of points
It's different for each nation that's why there's no "formula" stated in the GG. "Tank Heavy" for the Japanese isn't as "Tank Heavy" as it is for the Germans. The simple way to see what would be bought one way or the other would be to select a specific number of points then set whatever nation(s) you want to check to computer purchase and human deploy then try six purchases with Tank heavy on and six without and count the number of tanks bought each game. You should see a difference when you check the HQ screen but it will NOT always be the same purchases and sometimes, in the case of nations like Germany that have a more complicated picklist, you may even see cases where having tank heavy off may sometimes pick more tanks that when it is set to tank heavy on but on average there will be more but it still won't turn it into a tankfest. If we made it predictable we'd get more complaints
Also, really large point game may show up a bigger difference than more modest numbers of points but something like 2500 points would give you, on average in a meeting engagement the German player, maybe 16 tanks set to tank heavy off and 19 when tank heavy is set to on but that's the German OOB. Other tests with other nations may show a difference. Type of battle also affects that number.
Don