Since some said this matchmaking was tested in Beta: it wasn’t. There are not enough players in Beta for properly testing matchmaking anyway.
In theory, this doesn’t look bad:
But apparently 27 vs. 30 wars happen, and sometimes between rather unequal alliances. So I’m not sure if the algorithm is working well for all matches. We the players have little to no way of testing it.
A small improvement might be to weight more the top 5 heroes of each player (to account for defense team strenth). This was suggested somewhere above in this thread.