The first three lines are from Starsector's core files Commodities.CSV. The last three lines are from the mod's Commodities.CSV:
This taken from the mod's configurations settings at data/config
"MF_OreConversionRate": 10,
"MF_OreCost": 185,
"MF_RARE_OREConversionRate": 10,
"MF_RARE_ORECost": 62,
"MF_ORGANICSConversionRate": 10,
"MF_ORGANICSCost": 95,
"MF_HeavyMachineryCost":1,
Now referring to the charts found on the OP, you can see the breakdowns of each commodity that matches the above conversion numbers.
So that being said, let's take the Palladium conversion as an example: Rare_Ore is set by starsector at 75. I have Palladium set at a base price of 75. Now multiply that by the Utility of 64 which gives it a conversion price of 4800. Now this does effect market demand because the game sees this 1 product as 64 rare ore. I originally had the custom commodities set as their own demand class non-econ commodity but as was pointed out by slowpersun it did not convert properly so I changed the conversion to use the demand class of the originating ore. This replicates the bulk value (which is always lower) of the rare ore sold at a station versus a 1 v 1.
[key HM=Heavy Machinery]
Now back to the OP charts vs the mod's configurations settings I have the RARE_ORECOST set to use 62 versus the the above UTILITY of 64 from the Commodities.CSV. The 2 difference accounts for the HM cost (HM cost 150 vs Rare ore cost 75 x2 =150). The downside of this is that HM has it's own demand and market fluctuations. I can not factor that in because I cannot do split demand classes so that fraction of the overall that represents HM does account for some loss in overall value of the Palladium.
Now the only factor I can see maybe effecting the price is that I lowered the price variability by 1 in the mod's CSV vs the game's CSV. The variability affects how price fluctuates with demand. I don't remember the exact formula, but it factors into an exponent somewhere. Variability 0 means no price variation at all. Variability 10 is something like 16x the price when no demand is met, and a comparably sharper dropoff after all demand is met. So by lowering the variability I lowered the variance fluctuation but that my not be working as intended because math is hard and I suffer from dumb brain.