so i ran some simulations like i said i would.
using a normalized distribution curve:
0: 5
1: 10
2: 20
3: 35
4: 20
5: 10
assigning the round winner berries randomly in an instance of a tie
assigning all the bonus berries randomly
the results are in;
(might have to enlarge the images to see the numbers lol)
yeeeaaahh even the best scorers are barely scraping 40 here, so that seems a bit high. 30 is fine anyways since i don't necessarily want to have all 10 rounds if i don't have to, that's just the upper bound. i may adjust the EP round bonuses accordingly to account for this? i don't see the roulette going the full 10 rounds if i use 30 berries just based on the sims but 35 is an ugly number and 40 is too high so 30 is fine. then 2 bonus berries for every round ahead of the final round that early scorers make it in.
with 30 berries it's technically possible to qualify for the finals in 4 rounds but that's, again, exceedingly unlikely and if you score 4-5 points every round, win 3 or 4 rounds in a row, and get 6-8 bonus points in 4 rounds (which is to say, going with risky themes), you probably deserve to win anyways
edit: updated. I also added some further details in the rules & a tiebreaker criteria for the EP round