Low luck was perfectly described. The idea is that in normal dice games, many times a huge fight (around 70-100 IPC on each side, sometimes more) happens and the luck potential of those fights is huge in comparison to the normal smaller fights. For exemple if both sides average hit number would be 10 hits, variance makes it quite likely to make it different then that, but even a small difference on turn 1 of attack (like 12 hits against 6 lets say) keeps affecting the following turn of attacks (more units left = more probable hits), therefore increasing the final effect (for exemple a 6 units variance early could end as a 12 units difference in the end, making for around 30-50 IPC depending of the planes in the fight (and if battleships are involved it can get ugly !). In a smaller fight, a lucky roll is unlikely to win you more then 10 direct IPC (usually quite less) and maybe a couple more of circonstancial IPC due to the fact that a territory is not traded.
Oddly enough, the odds work in exactly the opposite way you are describing. The larger the battle is, the more closely it will follow low luck. Variations in hit results will follow the central limit theorem (an interesting google if you have the time). It’s small battles that will be most effected, and actually somewhat break the system.
Think about 2 tanks vs. 1 tank
2 Tank hits:
Low luck 1 hit 100% of the time
With Luck 0 hits 25%, 1 hits 50%, 2 hits, 25%, Thus the expected value is 0*.25+1*.5+2*.25= 1 hit. BUT, you can’t really hit twice against 1 tank, so the actual expected value is .75. So low luck gives an unintended boost in hits.