Very good thread!
I will use this formula later for my tests.
A thing i really feel to be considered in calculating efficiency is standard deviation from the average.
A deck with lower standard deviation has a more reliable behaviour and can be confronted with later tests.
Variance also points out if the data of the test are stable or it require further analisis.
I can provide data for the deck about TTW, win rate, EM rate and gain before spins.
Anyway all these data can be mined from a tester if he keeps the data comma separated like:
turns, gain
10,16
9,40
7,17
Then it's easy to import the data in excel and do the math.
I'd consider also the turns to loss, most when dealing against FG; some decks can drag out for 30 turns before losing, or others just see the odds aren't right and surrender the first turns.
As for the standard deviation, I'm working on a little update, I could as well put it, but, IMHO, it's not that needed and could, as well, confuse someone.
As for the TTL, I felt the need for it for a little time, it could be implemented but:
1.I'm following CG's formulas, I can't deviate from it too much, even though CG itself said it wouldn't be too hard to insert TTL;
2.TTL isn't really relevant, mainly because we're not for FG, the biggest differences would be for Plat, whom is filled with Skips for certain decks.
However, calculating the TTW even when you lose basically fixs it up.
What happens when we enter TTL in the TTW slot when we lose?
Let's say we have a 75% WR, a 10 TTW and a 2 TTL. If we simulated 100 games, we'll be having, ideally, 75 games with 10 TTW, e 25 with a 2 TTL. If we average those we obtain a 8 TTF (Turn To Finish) on average.
Now let's try using 8 TTW as an average and then try calculating it with 10 TTW e 2 TTL.
The gain from win is ALWAYS 30, and the cost is 15. There are no spins, no special spins and no variation in the gain, neither from the HPs nor the opponent, not even EM.
Let's say that we're playing awesomely fast, 4,5 seconds per turn, 800 TPH. This way, we won't have to calculate TPH in the formula, because 100 games are exactly 1 hour.
Now, take the 75 games, multply for 30, we get 2250. Those were the electrums gained.
Now take the 25 games lost, multiply for -15, we get -375, the amout of electrums gained (and, because it is negative, the amount of electrums lost).
2250-375=1875 Electrums.
Given that the 75 games have a 10 TTW average, it is a total of 750 turns, while the losses are 50 turns. 800 turns at a 800 TPH are exactly 1 hour of gaming.
Said this, the UEI is 1875 Electrum/Hour, using TTW and TTL.
What happens if we use 8 as an average TTF?
100*8=800 turns at an 800 TPH is exactly 1 hour of gaming.
1875 Electrum/Hour, isn't that exactly the same? This works because we're not barely averaging TTW with TTL, but we took the amount of games won and lost.
