Unified Efficiency Index - How to compare apples to oranges

[ColorlessGreen]

I think there is an error at the pink-bold item in the quote above but I cannot quite place my finger on it.  Is SSV supposed to be a "per hour" value?  Or should unupped also be "* SSPH"?  Or should upped be GPS insetad of SSPH?  Am I way off here?!?

Thanks for catching that. The formula has it right, but my explanation kinda changed how I was explaining it halfway through. Clarified.

edit: Wow, actually I think I forgot to write down the odds of successfully spinning a gorram rare when I was transcribing the formula. I should probably fix that. And go through and redo all the formulae. And all the results. Sigh.

edit2: On the plus side, this gives me the opportunity to incorporate the work CuCN, the dictator, and pineapple have done on calculating the exact odds of SSPH. Also it gives me the excuse to correct a couple other tiny details I've been meaning to correct. There should be an update coming to the formula (and the stats threads) hopefully within the next week or so (i.e. as soon as I am able to).

[Leodip]

Well, if you do modify it, it'll be the excuse I needed since long to port my StatKeeper to GDocs.

[ColorlessGreen]

Major update. Anyone with a local copy of UEI should update their formulae.

-AI4 spin value updated per current stats in AI4 study.
-Corrected mistaken ideas about FG/AI4 non-EM electrum rewards (specifically, they don't go from half to whole; ai4 goes from 3/8 to whole, while FG goes from 7/12 to whole).
---Made AI4 a special case for electrum rewards (as compared to ai3/arena, which work with the standard half to whole).
---Changed all rewards to reflect a typical 50 hp at non-EM victory instead of 57.
------Changed EM factor to compensate
-Corrected my failure to include a special spin droprate in the actual formula despite knowing I needed to.
-Updated typical arena values to 15/60/115/240 to more accurately reflect where they've been for the last long time.
-Corrected all formulae.
-Clarified a bit of stuff.

edit: Don't feel like putting it in the OP, but here's a formula for AI2 UEI:

Code: [Select]

Unsimplified AI2:
(333 * WR * 10 * 0.75 * (1 + (EMR * 2.67)) * XHP / TTW) - (333 * (1-WR) * 5 / TTW) + (333 * 13.33 * WR / TTW)

Simplified AI2:
(2497.5 * WR * (1 + (EMR * 2.67)) * XHP / TTW) - (1665 * (1-WR) / TTW) + (4438.89 * WR / TTW)
edit2: Fixed incorrect LV for AI2 formula. Thanks TD

[Leodip]

Too fast. I've yet to finish writing part three of the tutorial and posting a couple of things D:
Well, GJ, it looks like it's ok (but I'm sleepy, may have gotten that wrong).

[ColorlessGreen]

Two things:

(1) Thanks to the dictator for pointing out my bad math in EM factor. EM factor was one higher than it should be because I am bad at math. All formulae have been corrected.

(2) Added the following:
In order to better compensate for differing play times across different targets/with different decks, all official studies will now report both UEI(333) and UEI(250). The official UEI formulas will still use 333 turns/hour as a basis, but the studies will report results for 333 turns/hour (listed as UEI(333)) and results for 250 turns/hour (listed as UEI(250)). 333 turns/hour is approximately 10.81 seconds/turn, while 250 turns/hour is approximately 14.40 seconds/turn. The formula for calculating UEI(250) is simply UEI/333*250. Note that it is very simple to determine personal exact time to compare decks by figuring out your average seconds/turn, then figuring out your average turns/hour (i.e. 3600/(seconds per turn)), then dividing UEI by 333 and multiplying by your personal turns/hour.

[serprex]

I've removed notes about including 250 TPH. Multiplying results by a constant (0.75) is the responsibility of the user

[SevenOfAKind]

It appears as though youve used 2 different values of TSP for AI4, initally you say it is 117.90 then later in a table state it as 104.77, is one of these a typo and if so, does that typo appear in the simplified formula towards the end of your post? When calculating UEI for an AI4 deck Ive been testing I get 3.2k UEI using 117.90 in a formula based on the very first formula you state and then I get 2.8k using the simplified version towards the end.

[serprex]

2038 cg @serp iirc there was an initial value used for ai4 prior to the ai4 value study being completed
2038 cg like iirc the ai4 stuff had results of like 300 tests used to compute some early numbers, then after 1000 tests i called it "close enough to accurate" and updated the formulae
2039 cg whatever he's talking about is probably me failing to correct some mention of the rate
2039 cg the ai4 study thread should explain which number is the correct number and which is the outdated number
2040 cg the number he's talking about iirc is the value per spin number
2040 cg and for ai4 that wasn't viable to simulate at the time
2040 cg so we had to base it off a large number of actuals

[SevenOfAKind]

Thanks, just checking cos I read over both, initially used the first formula and the AI4 thread to piece it together then plugged in the simplified version and got confused over where the large difference came from.

Out of curiosity, has anyone ever tried doing a study on the bonus electrum from matching the first 2 cards but not the last? I imagine it could add a just about barely noticeable amount, meaning faster win decks are being very marginally penalised here. Probably not worth the effort, I'm just curious if it's been done.