GEN King's Bounty any% speedruns (ft. RNG seed tracking)

Played on console - Sega Genesis, Model 1 (TMSS). Mission - Palmer: Impossible.
This video is a demonstration of speedrunning King's Bounty on the Genesis using the RNG seed tracking method. It highlights a few successful runs, the fastest of which is at 11:30 in the video. I've also uploaded this fastest run separately with timer on display, here: • GEN King's Bounty spee...
Since RNG manipulation is not viable on actual console, I propose the most efficient method to speedrun this game is seed tracking. There are a total 16384 possible RNG seeds the game uses to generate much of its object data. Once the player character & difficulty have been selected, the list of possible seeds is automatically narrowed down to 1024. From these, a seed is randomly selected and used to generate the sceptre coordinates, as well as many other object locations.
Though now obsolete, my original RNG manipulation video can still be found here: • GEN King's Bounty spee...
The selection of Palmer/Impossible in these runs is slightly arbitrary and was mainly chosen for testing purposes. Keep in mind, not all combinations have potential seeds that put the sceptre at the player's starting location. For purposes of the any% speedrun, it may be more advantageous to run Crimsaun/Easy which has 2 such seeds possible. However, note that seed tracking can be used for any speedrun category, in which case sceptre location may not be the most important criterion.
ROUTE NOTES
All seed data was collected ahead of time and organized in spreadsheets. This data was extracted by myself using custom Lua scripts in BizHawk. All of this is currently available on Google Docs:
docs.google.co...
From there, you can download a copy of the documents and modify the data to suit your needs. All the data should be accurate, but please let me know if you come across anything that looks wrong or confusing.
Once you've selected your character & difficulty, you'll have an initial list of 1024 seeds. This list can be further narrowed down by omitting seeds based on certain criteria, such as sceptre location. For instance, some seeds may have not have the sceptre placed on the 1st continent (map 0) or else it may be in a difficult to reach location.
The most important aspect of routing using this method is being able to quickly locate coordinates on the map. This is also the most tedious part of routing. It involves going through all potential locations and writing short descriptions of them. I did this by assigning names to geographic features on the map and describing locations relative to fixed objects such as towns or signs.
For example, "Passage - 1st span - right-most tile" is my description for (19,34) on map 0. It probably doesn't mean a whole lot to anyone else, but you can likely come up with a similar description of your own by learning the map and naming things accordingly. Some locations are much harder to describe -- some of my descriptions are extremely verbose in some cases quite confusing. Refinement of these descriptions is all part of the routing process.
During a run, 3 points of information are used to narrow down the list of possible seeds, namely:
1. The object in slot 1 (due southwest of the starting location)
2. The monster/character type that appears in Hunterville
3. The villain that rules Castle Vutar (which is learned in Hunterville)
Additional info can be gathered during the course of the run -- specifically, objects that appear in other slots. In order to make this information useful, you would need to scout seeds -- look up the desired slot data for a particular seed and make notes in the sheet.
There's also one frame-saving maneuver that I employ: Since I decide to dig as my first action (just in case), you are able to hold down the A button beforehand then release it when player control is gained in order to bring up the menu early.

[@RedArremer]

Your spreadsheet has 16385 total lies, which means there's a mistake in there somewhere. Also my father found 68 specific lies for a certain set of paramaters for one of the characters on impossible, and your sheet only listed 67, we triple-checked it, and that's another error in the database.

[@TascoDLX]

It definitely has 16384 on all tables; I'm not sure how you're counting that. As for the 68, you'll have to be a little more specific about the data point you're looking at. All my data was extracted via emulation using lua script, therefore I cannot guarantee it's free of error. But if you let me know which data is incorrect, I may be able to re-extract and verify.