Hello! First time posting after much lurking and searching.
Leveling roulette has been a big part of my leveling towards omni 100 DoW/M over the past few years, but I often had the feeling its results weren't very consistent. So recently, I decided to start keeping track properly. Here are some preliminary results! (see images).
I say preliminary, because it will take a lot more data to be able to draw meaningful conclusions. But there are some fun results to point out already:
Roulette bonus outcomes really are inconsistent. Even at levels where the instance EXP is negligible and it's entirely up to the roulette bonus to give you relevant EXP, results vary greatly. Players around the 60-80 range appear to get especially stingy roulette bonuses. As just by chance these players are over- or underrepresented in some instances in my data so far, the horizontal red lines indicating mean bonuses in the % of next level graphs should be taken with a generous lick of the Ishgardian salt rock until I have broader coverage of different player levels!
The roulette bonus is fixed... probably? It's common knowledge that EXP buffs only apply to such sources as direct mob/boss kills and FATEs, which acknowledge the buffs explicitly in the chat log by mentioning the buff amount "(+xxx%)". Conversely, the roulette bonus does not mention any such buffing. If it really isn't affected by buffs, then it should never vary between players of the same level doing the same instance (assuming that in itself is deterministic!). I have captured two such cases so far. Two 66 players in Ifrit NM got the same bonus -- actually 476000 and 476400, but I'm choosing to believe the former player truncated the number because an 0.084% buff makes no sense to me -- while two 62 players in Gubal Library got noticeably different bonuses, 269800 and 353050 (~ +30.86% vs the lower). That doesn't match any buff amount I can think of, and the "buffed" number is still suspiciously nice and round.
Related to the above, when player level <= 80, the roulette EXP bonus is always a nice clean multiple of 50. No counterexample to that in n=39 bonuses awarded to level <= 80 players. To me, that suggests these are probably constants. Or they're being rounded, but that seems both unnecessary and unlikely given that past 80, not-so-round numbers appear regularly. I suspect whoever was in charge of setting those constants just stopped bothering to round the result of some guideline calculation, and took the actual result (still rounded to integers at least).
So, some fun things going on, but it's not nearly enough data. Being a daily, I can only gather a single datapoint a day myself without getting into alts. So as hinted above, I've started asking others to share their numbers at the end of a roulette. It helps, but it's still slow going.
That's why I'd like to ask for your help! I've set up an easy form here: https://forms.gle/ne8Kks3pEABpJnys9 . All it needs is your instance, level at the time you enter, and your bonus at the end. Accepts up to a whole party's results at a time, if you can get others to report their numbers!
As a bored data scientist, every little bit is much appreciated! If this ends up helping to gather meaningfully more data, I will look to make a followup post and include the raw data. Feedback and ideas are welcome too!
