What the binary does not settle on its own. The whole cartridge is explained byte by byte; this is what remains to be measured or decided.
Between the platforms and the fruit there are eight lists of X positions ending in 0xFF, one per stage, and no instruction in the trace points at them. By their place and shape they look like an earlier layout of flowers or fruit that went unused. No instruction that reads them has been found in the trace.
GENERA_ECUACION computes a random offset from 0 to 7 to choose among several scripts per level and discards it with a xor a. The offset tables that remain are one byte long, so today there is one script per level. Whether there were ever more, or whether the xor a is what left level 5 with its single five-symbol script, the binary does not say.
The stage (E051) is binary; PINTA_FASE (0x4A24) converts it to decimal for display and resets it to 0 on reaching 100. The tables of platforms, cards and fruit are indexed by (stage − 1) & 7, so the eight stages repeat. What grows is how many crabs are in play (one up to stage 7, two from stage 8 to 15, all three from stage 16; 0x4850) and how little they wait hidden (from stage 32, one frame). It remains to actually play that far to see whether the game becomes unplayable before that or whether the wrap to stage 0 has any effect the code does not show. One measurement so far: a novice player's game, at levels 4 and 5, lasted 5 min 37 s from LEVEL SELECT to GAME OVER, with one stage cleared and 8,360 points (replay in work/omsx/partida).
ESTADO_6_DEMO_MONTA (0x42FF) puts digits 2 to 7 in cards 4 to 9, leaves 0 to 3 with digit 0, and the hidden digit is 2. The input script at 0x4B78 takes the demo to the right card, but what happens if the fruit or the crabs —which are random— push it off course has not been checked: the script is fixed and the crabs are not.