Here's another idea: punish losses.

I.e. consider if rewards were only given at the end of the season. Stack increases with each win, and decreases with each loss. The more wins >50%, the higher the rewards. So it rewards a combination of win rate, as well as quantity of wins. However a player who's <50% with a massive number of battles wouldn't necessarily reap the highest rewards.

It discourages fleeing and discourages de-ranking for liquidity and exploiting weaker opponents.