Games across the generations are programmed differently, have different RNG implementations, and different patterns.
Gen3 uses a single RNG; a 32-bit Linear Congruential RNG. To simplify things, the 32-bit RNG produces four 16-bit random values to generate the PID (32-bits) and IVs (30 bits). The RNG itself has 2^32 different states, so the PID (2^32) and IVs (2^30) are restricted in the possible results they can be found with. A 32-bit LCRNG cannot output more than 2^32 different combinations; the PID/IV has 2^62 different states possible. Hence, we can check if an RNG seed can yield the current PID/IV through backwards calculation.
Gen5 uses two RNG instances to generate a Pokémon. A 32-bit seeded Mersenne Twister generates IVs, and a 64-bit main RNG to produce everything else. The mersenne twister can produce all possible IV combinations, so there is no correlation/restriction to detect for those. The remainder of the "random" properties such as PID-Gender-Ability (32-bits), Nature (0-24) are possible to obtain in any permutation, as the 64-bit RNG is orders of magnitude more random. It is important to note that there are still restrictions; if the game generates things in a specific pattern (like truncating certain bits of the PID) then the Pokémon must still follow the patterns of how the game generates a Pokémon to still be legal. The Entree Forest Pokémon require the ability bitflag of the PID (bit 16) to be 0 so that the game properly remembers it has that ability slot rather than a random ability.
If all pkm-property-combinations are randomly possible from the game's RNG, then there is no correlation to detect. It's not worth the time/computation to guess & indicate an RNG seed, when there are billions+ of possible seeds that can originate it.