AI3 EM True Efficiency Study ver. 2.0 (1.27)

[kirchj33]

While TTW data is a useful statistic, it only tells part of the story for score gain. Hence, data tracking should evolve towards efficiency.

If anyone has any suggestions as to how the entire community could get involved this time, please post below.

One idea I had was that I could list each deck (it will probably be around 10) that will be tested for efficiency and each person that wants to participate can do 50 games. If data is taken in equal proportions, time based data should be normalized. The only problem I forsee with this is how new v. 1.27 EM decks would be tested for efficiency when added.

Similar to the Fgei (http://elementscommunity.org/forum/index.php/topic,14626.0.html) used to calculate efficiency in false god farming, what needs to be calculated are the following:
A=Score Change
B=Total Time

The equations used would both be simple as a result:

Score/Time=A/B

*Note that time for losses should also be recorded for this to work.




As an example, I played USEM (http://elementscommunity.org/forum/index.php/topic,10563.0.html) for 50 games:
A=+1482
B=3972 secs/60/60=1.1033 hrs

Score/Time=1343.2430 score/hr       (1)

We will call this measure In-Game Score Efficiency (1)


You may say, however,  this is not the true efficiency!

You would be correct, as this number does not calculate time between games.  So let's take this one step further (all numbers from here are hypothetical, as I did not calculate actual time played when doing some statistical analysis of USEM way back when).  For this we need total time played.

To calculate this, you would write down the time you began playing, and the time you stopped to determine a total time played.  Let's say you were uber accurate and calculated this, down to the exact second.  (Taking the data I used above and adding a hypothetical 9 seconds between games(end game screen + spins + selecting level 3 at menu)):

*Important Note*- I believe this is better as an estimate, as when you are recording data opposed to farming, you will need time not included in the study to record data.  9 seconds between games should be used to standardize all data.

C=Total Time=4422 secs or 1.2283 hrs

We then have the following:
A/C=1206.5456 score/hr      (2)

The would represent our True Score Efficiency (2).  This latter "True" efficiency calculation helps normalize data for those decks which farm for speed and for those which are focused on speed + EM's.

In summary:

USEM (http://elementscommunity.org/forum/index.php/topic,10563.0.html) (50 games)

In-Game Score Efficiency	1343.243/hr
True Score Efficiency	1206.546/hr

I believe 200 games, similar to jmdt's (http://elementscommunity.org/forum/index.php?action=profile;u=3818) TTW Study 3.0 (http://elementscommunity.org/forum/index.php/topic,21865.msg299818#msg299818), should be the standard.  I will only be tracking upped decks as those are the only ones that work reliably for EMs.

I will continue to work on the data myself, as it was suggested that time-based efficiency is best performed by one person, as player patience and computer speed can vary quite a bit from person to person.

*I will post Efficiency data in score/hr below as that seems to be the easiest to conceptualize.
**Stats recorded on Fast AI with High Quality

[kirchj33]

Upgraded Deck Efficiency



-TADAbow (http://elementscommunity.org/forum/index.php/topic,22518.msg304907#msg304907): TSE: 1387.192/hr, Games: 200, Score Change: +6294, Electrum Change: +8089, Total Time:  242.2333 mins, Deck by - kirchj33
5 Quantum Tower, 6 Supernova, 3 Discord, 1 Flesh Recluse, 3 Black Hole, 2 Elite Graboid, 4 Improved Heal, 1 Lava Destroyer, 2 Abyss Crawler, 1 Elite Pegasus, 1 Gargoyle, 1 Phase Recluse, Entropy Mark.

Code: [Select]

6qq 6qq 6qq 6qq 6qq 6u3 6u3 6u3 6u3 6u3 6u3 6u5 6u5 6u5 713 74f 74f 74f 77g 77g 7ai 7ai 7ai 7ai 7dq 7gm 7gm 7jr 7tf 80g

	Raw	Rate
Games	200
Win	197	98.5%
EM	169	84.5%
Avg. TTW	7.6294
Avg. Time	72.67 secs
Total Time	4.0372 hrs
Total Score	6294
IG Score Eff	1558.993/hr
True Score Eff.	1387.192/hr

TTW
4	1
5	6
6	51
7	57
8	38
9	18
10	6
11	10
12	5
13	3
14	1
15	1
16	0
17	0
18	0
19	0
20	0

-USEM (http://elementscommunity.org/forum/index.php/topic,10563.0.html): TSE: 1386.2083/hr, Games: 200, Score Change: +5824, Electrum Change: +7559, Total Time:  222.083 mins, Deck by - jmdt
5 Emerald Tower, 5 Life Pendulum, 2 Long Sword, 6 Giant Frog, 6 Elite Cockatrice, 6 Improved Heal, Life Mark.

Code: [Select]

6rk 6rk 7ac 7ac 7ac 7ac 7ac 7ae 7ae 7ae 7ae 7ae 7ae 7ag 7ag 7ag 7ag 7ag 7ag 7ai 7ai 7ai 7ai 7ai 7ai 7bu 7bu 7bu 7bu 7bu

	Raw	Rate
Games	200
Win	193	96.5%
EM	151	75.5%
Avg. TTW	7.0777
Avg. Time	66.625 secs
Total Time	3.7014 hrs
Total Score	5824
IG Score Eff	1573.4634/hr
True Score Eff.	1386.2083/hr

TTW
4	0
5	22
6	73
7	40
8	25
9	15
10	7
11	4
12	3
13	3
14	1
15	0
16	0
17	0
18	0
19	0
20	0

-Disc' EM (DEM) (http://elementscommunity.org/forum/index.php/topic,23307.msg316979#msg316979): TSE: 1386.09/hr, Games: 200, Score Change: +6001, Electrum Change: +7821, Total Time:  229.7667 mins, Deck by - kirchj33
5 Amethyst Tower, 6 Supernova, 1 Amethyst Dragon, 2 Werewolf, 3 Discord, 1 Flesh Recluse, 2 Black Hole, 1 Elite Graboid, 3 Improved Heal, 1 Minor Phoenix, 1 Abyss Crawler, 1 Elite Pegasus, 1 Phase Recluse, Entropy Mark.

Code: [Select]

4ve 4ve 6ts 6ts 6ts 6ts 6ts 6tv 6u1 6u1 6u3 6u3 6u3 6u3 6u3 6u3 6u5 6u5 6u5 713 74f 74f 77g 7ai 7ai 7ai 7ds 7gm 7jr 80g

	Raw	Rate
Games	200
Win	196	98%
EM	151	75.5%
Avg. TTW	7.62245
Avg. Time	68.93 secs
Total Time	3.8294 hrs
Total Score	6001
IG Score Eff	1567.068/hr
True Score Eff.	1386.09/hr

TTW
4	0
5	13
6	38
7	52
8	54
9	16
10	11
11	3
12	5
13	0
14	2
15	2
16	0
17	0
18	0
19	0
20	0

-TADA Blue (http://elementscommunity.org/forum/index.php/topic,22147.msg302391#msg302391): TSE: 1318.9958/hr, Games: 200, Score Change: +5823, Electrum Change: +7603, Total Time:  234.8833 mins, Deck by - kirchj33
6 Supernova, 3 Discord, 6 Entropy Pendulum, 1 Flesh Recluse, 2 Black Hole, 1 Graboid, 3 Improved Heal, 1 Minor Phoenix, 5 Abyss Crawler, 1 Elite Pegasus, 1 Phase Recluse, Water Mark.

Code: [Select]

6u3 6u3 6u3 6u3 6u3 6u3 6u5 6u5 6u5 6ve 6ve 6ve 6ve 6ve 6ve 713 74f 74f 77g 7ai 7ai 7ai 7ds 7gm 7gm 7gm 7gm 7gm 7jr 80g

	Raw	Rate
Games	200
Win	194	97%
EM	144	72%
Avg. TTW	7.5619
Avg. Time	70.465 secs
Total Time	234.8833 hrs
Total Score	5823
IG Score Eff	1487.4619/hr
True Score Eff.	1318.9958/hr

TTW
4	0
5	13
6	42
7	43
8	55
9	21
10	9
11	8
12	1
13	0
14	1
15	0
16	1
17	0
18	0
19	0
20	0

-Aerial Smooshing (http://elementscommunity.org/forum/index.php/topic,21643.msg295915#msg295915): TSE: 1234.7089/hr, Games: 200, Score Change: +5502, Electrum Change: +7232, Total Time: 237.3667 mins, Deck by - kirchj33
10 Wind Tower, 6 Damselfly, 6 Azure Dragon, 2 Long Sword, 6 Improved Heal, Life Mark.

Code: [Select]

6rk 6rk 7ai 7ai 7ai 7ai 7ai 7ai 7ms 7ms 7ms 7ms 7ms 7ms 7ms 7ms 7ms 7ms 7mt 7mt 7mt 7mt 7mt 7mt 7mv 7mv 7mv 7mv 7mv 7mv

	Raw	Rate
Games	200
Win	189	94.5%
EM	134	67%
Avg. TTW	7.6455
Avg. Time	71.21 secs
Total Time	3.9561 hrs
Total Score	5502
IG Score Eff	1390.7597/hr
True Score Eff.	1234.7089/hr

TTW
4	0
5	6
6	44
7	60
8	36
9	14
10	16
11	7
12	2
13	2
14	1
15	0
16	1
17	0
18	0
19	0
20	0

-GAEM: Giant Amphibi'EM (http://elementscommunity.org/forum/index.php/topic,21620.msg295673#msg295673): TSE: 1168.041/hr, Games: 200, Score Change: +5470, Electrum Change: +6985, Total Time: 250.9833 mins, Deck by - jmdt
6 Supernova, 6 Entropy Pendulum, 3 Chaos Power, 1 Flesh Recluse, 1 Elite Graboid, 4 Giant Frogs, 4 Improved Heal, 1 Minor Phoenix, 1 Abyss Crawler, 1 Elite Pegasus, 2 Vampire Dagger, 1 Phase Recluse, Life Mark.

Code: [Select]

6u2 6u2 6u2 6u3 6u3 6u3 6u3 6u3 6u3 6ve 6ve 6ve 6ve 6ve 713 77g 7ae 7ae 7ae 7ae 7ai 7ai 7ai 7ai 7ds 7gm 7jr 7tb 7tb 80g

	Raw	Rate
Games	200
Win	192	96%
EM	129	64.5%
Avg. TTW	7.6094
Avg. Time	75.295 secs
Total Time	4.1831 hrs
Total Score	5470
IG Score Eff	1307.657/hr
True Score Eff.	1168.041/hr

TTW
4	8
5	24
6	50
7	28
8	23
9	26
10	10
11	10
12	2
13	3
14	3
15	2
16	1
17	2
18	0
19	0
20	0

-Smokey Bear- "Only You Can Prevent Forest Fires" (http://elementscommunity.org/forum/index.php/topic,21375.msg291949#msg291949): TSE: 1132.644/hr, Games: 200, Score Change: +5510, Electrum Change: +7163, Total Time: 261.8833 mins, Deck by - kirchj33
2 Crimson Dragon, 2 Phoenix, 1 Long Sword, 2 Giant Frog, 6 Improved Heal, 6 Cremation, 4 Minor Phoenix, 6 Damselfly, Earth Mark.

Code: [Select]

5f2 5f2 5fc 5fc 6rk 7ae 7ae 7ai 7ai 7ai 7ai 7ai 7ai 7dp 7dp 7dp 7dp 7dp 7dp 7dq 7ds 7ds 7ds 7ds 7mt 7mt 7mt 7mt 7mt 7mt

	Raw	Rate
Games	200
Win	189	94.5%
EM	136	68%
Avg. TTW	7.3757
Avg. Time	78.565 secs
Total Time	4.3647 hrs
Total Score	5510
IG Score Eff	1262.394/hr
True Score Eff.	1132.644/hr

TTW
4	9
5	28
6	52
7	30
8	27
9	13
10	6
11	5
12	10
13	4
14	3
15	0
16	1
17	0
18	1
19	0
20	0

-SCEM (http://elementscommunity.org/forum/index.php/topic,18665.0.html): TSE: 1124.916/hr, Games: 200, Score Change: +5014, Electrum Change: +6509, Total Time: 237.4333 mins, Deck by - jmdt
6 Entropy Pendulum, 6 Supernova, 3 Chaos Power, 4 Elite Charger, 1 Flesh Recluse, 1 Elite Graboid, 3 Improved Heal, 1 Minor Phoenix, 1 Abyss Crawler, 1 Elite Pegasus, 2 Vampire Dagger, 1 Phase Recluse, Gravity Mark.

Code: [Select]

6u2 6u2 6u2 6u3 6u3 6u3 6u3 6u3 6u3 6ve 6ve 6ve 6ve 6ve 6ve 713 745 745 745 745 77g 7ai 7ai 7ai 7ds 7gm 7jr 7tb 7tb 80g

	Raw	Rate
Games	200
Win	189	94.5%
EM	109	54.5%
Avg. TTW	7.1376
Avg. Time	71.23 secs
Total Time	3.9572 hrs
Total Score	5014
IG Score Eff	1267.05/hr
True Score Eff.	1124.916/hr

TTW
4	5
5	29
6	43
7	46
8	29
9	19
10	5
11	8
12	3
13	1
14	1
15	0
16	0
17	0
18	0
19	0
20	0

*All time statistics include those from losses (except TTW of course)
** Decks posted in order of efficiency, top to bottom.
***Some decks that were less efficient from AI3 True Efficiency Study ver. 1.0 (1.26) (http://elementscommunity.org/forum/index.php/topic,21530.msg294197#msg294197) were removed with the interest of only including the top AI3 score farmers.

[kirchj33]

Some people had expressed a desire to help out, so I will link the spreadsheet I use to googledocs.

TSE Document (https://spreadsheets.google.com/ccc?key=0Am2NpVDyToyBdDlCRnFjcFZKTXpIUnJZajNNM19xMHc&hl=en&authkey=CIPAiLsO#)

It calculates TSE automatically for you.

50 games of each deck and posting would be a great place to start.

[kirchj33]

/quadruple post

All the posted decks are now updated for 100 games.  The men (USEM) are starting to separate from the boys.

The gameplan that follows is this:

1. Do some testing, tweaking, and post 100 games on some BH/Discord decks.  There are some really efficient decks that lack data out there in this category.  My estimate is that one or two of them can compete with USEM for efficiency and are alot more fun (personal opinion) to play.

2. Follow up with 200 games of each deck, add USEMosis.

[kirchj33]

200 games posted for all decks included.  I will start from zero with USEMosis as I did not find results competitive with those posted by Willng3 through 100 games.  I will post 200 games of USEMosis within the next couple of days and update.

[frisbee1989]

Could some1 test Essence's Mastery Despoiler? http://elementscommunity.org/forum/index.php/topic,11412.0.html

I'm working on getting all the upgrades for it, but with only the Vamps and all the non creature cards upped I got a true score efficiency of approximately 1108. I averaged about 10.5 turns to win and in an earlier study the ttw of the fully upped version was less than 8.5. I EM'ed about 80% of my wins. Will be much faster upgraded, b/c then you can play devourers on your first turn and start the quanta denial ASAP. So less damage from your opponent and more for you eventually.

Some variations that might be better are replacing 2 gargs with steals or siphon lifes. I only had a couple of losses, but all them were b/c of annoying shields that i couldn't do anything about. Steals would probably be better than siphon, because i was generally short on quanta as it was. Taking the AI's shield or weapon would kill him faster so it will nullify the siphon life healing anyways.

[kirchj33]

Could some1 test Essence's Mastery Despoiler? http://elementscommunity.org/forum/index.php/topic,11412.0.html

I had included this deck in the first Efficiency Study (http://elementscommunity.org/forum/index.php/topic,21530.msg296140#msg296140).  While this study was less complete (<200 games), it featured more decks, and I chose the cutoff to be placed at 1000 TSE for the update.  I will probably go back and fill in these gaps in the future, but with war coming up, it will probably be a few weeks.

Note that all of these TSEs were much lower due to the update of AI speed.

[Wolfunit]

Just wondering but isn't this the excate same study which jmdt started with decks of 3 versions of the game already been studied ,He started a version 1.27 already.

[kirchj33]

Just wondering but isn't this the excate same study which jmdt started with decks of 3 versions of the game already been studied ,He started a version 1.27 already.

No, you're missing the point entirely.  Jmdt's study focuses on TTW, which is only one component of score farming.  This study is a review of which decks produce the best score/hour in the form of a measured index called TSE.  TTW translates well to PvP rushing, even though it is researched vs. AI3 decks, whereas the decks here focus entirely on gaining EM's for score.


On a side note, I added an additional deck with stats today: Disc' EM (DEM) (http://elementscommunity.org/forum/index.php/topic,23307.msg316979#msg316979)

[Amilir]

I think you may have missed the whole 'TIME to win' stuff that's been going on.  Really, this just factors in spinning and time to start another game.

[kirchj33]

I think you may have missed the whole 'TIME to win' stuff that's been going on.  Really, this just factors in spinning and time to start another game.

You may want to go up above and re-read what this study is all about then.  I would not have spent hundreds of hours testing if there were not a need for such data within the community.

1.  The first "T" in TTW has always represented, and will continue to represent, "Turns" as opposed to "Time".  Time can be a result of turns and the two are correlated, however, a different approach is taken here to focus on score generation.

2.  If you look at the TTW studies, score and electrum are represented in several posts, and theoretically, the score should be the same thing as represented by the in-game efficiency here.

3.  The TTW study focuses primarily on speed represented in terms of turns.  Which is only a small component of what is represented here.  This study focuses in on decks which generate high score/hour as opposed to speed in terms of turns.  The decks here are "EM decks" meant for farming AI3.

4.  The score per hour is a result of not only "turns to win", but ending HP, EM bonuses, and time per game overall.  The more "efficient" a deck is at performing well in each of these categories, the better the deck will be for farming AI3.  The raw score per hour that is measurable is represented by the in-game efficiency, but to give players a good idea of what they will actually be producing per hour if they choose to spin and restart games (required) a true score efficiency is posted as well and thus, is the focus of this data.

5.  While it is true that playing equal amounts of games should normalize the data between each deck, using the TSE allows individuals to compare decks that have been played in varying amounts of games.  I believe data represented by TSE is accurate within plus or minus 50 score/hour when 200 games are played from my experience.  This is my best guess, but the idea is to give players a good representation of how "good" of a farming deck they are actually using in terms of score when they set off to do so.

I hope this helps to clear some things up and thanks for helping to highlight what is a common misconception made by the community.

[FuryGhost]

Wow great post, even though a bit dated.