Historically I've played by building units at the start, then playing those subunits for missions after.
For example I may randomly roll up a regular heavy mech regiment. From there I determine all the company weights, and use faction tables for the appropriate timeframe to determine the makeup of each company.
Then games over the next few months are simply one or more elements, chosen randomly, from that initial force.
Sizes are also random, for example each player could roll 1d3 and get that many lances of their force. We rarely made it through a whole regiment, but if we did we'd just roll up a new one and continue.
No BV was ever used until victory conditions determined. If you had 3.56x the BV of your opponent, each kill you got was worth 3.56x the BV of the unit, and each kill the enemy got was 1/3.56, or 0.28x the BV of the unit they killed. This meant even if you lost all 4 of your mechs when facing 12, if you killed 4 in return you probably won the game. It was very fulfilling to have a light lance take out an assault mech, even if you lost everything. The cheers and excitement far outweighed "loosing", and could be far more humiliating for the "winner" :)
At the end of the year VP was calculated, looser bought the winner a new miniature lance of his choosing.
This was a LONG time ago though, but that is how we played it, and it was really fun.
The only other way I've ever played and really enjoyed was a full campaign between a clan galaxy and an AFFS RCT. The "pick X BV and X number of units" types of battles are just "slugfests" or "die rolling contests", as with completely equal forces the winner is nearly always determined by luck, and not skill, or perhaps skill in picking forces, but not so much in tactics which is why I want to play games.