Win Time Study - Pro Stats for Pro Grinders.

[hrmmm]

As an academic person I have to try to find flaws in this study to make it better.

i respect this. i (and i think jmdt and kuross aswell) am interessted in a "flawless" study.

If you look at the results, you can see that decks tested by jmdt have the most top spots, while decks tested by hrmmm and kuros have the bottom spots. 5 decks in top-6 were tested by jmdt, while only 1 deck in bottom-7 was tested by jmdt. There looks to be clear correlation between time to win and whether or not it was tested by jmdt. So my question is, do decks tested by jmdt get better score because of computer performance or style of playing?

please note that we are still need more testing (more decks and maybe more games played per deck).

please also note that i tested the shrieker rush. actually on rank 3 by sec/game.
jmdt had Sec/game: 57.36 and Avg ttw: 5.98
i had Sec/Game: 57.68 and Avg ttw: 5.780
even with a higher ttw rate, jmdt was slightly faster. you can clearly see the computer performance difference here.
both are shrieker decks, both decks are in the same ttw range as in the ttw study. so we can eliminate a playstile-factor here.

with such information we can determine a performance factor. im sure jmdt will imply such difference in his analysis (still in work of corse).

Or did jmdt just happen to pick all the fast decks by chance, while others happened to pick slow decks?

we decide in the chat about who is testing what deck.
as you see jmdt tested the most decks now:
jmdt 8
kuross 5
me 4
we started with the fastest decks first. i think they are the most interesting decks for the community. so its natural that he took more of the actually high-ranked decks.

Wouldn't this study be more "scientific" if only one person tested everything? (I'm not saying a one person should do it because it's a huge task).

this, or we all test the same decks to negate the performance difference.
... or we can calulate a performance factor to get better rounded results.

...
i was thinking about to mix up some testings results:
means calculate the average results from different testers
->hrmm is testing 50 games->jmdt is testing 50 games... = average datas from different testing conditions = a better determination of a decks speed?
what do you think about this method?

but since non of us reached "jimmizle-status", im not sure if all of us have the upgraded cards to test every deck.
and this would mean, all of us need to do this "huge task".
we have to care about the "work" this needs.
i needed a test free time. i was actually working on/enjoy a animation film festival in my town last week. and now i need a test free week to get the fun for testing back .

[Scaredgirl]

FACT 1. Decks in this study are listed as "Sec/game", meaning that the "fastest" deck is the one that used the least amount of time to win.

FACT 2. Computer performance is directly related to how long your games last.

FACT 3. The computer performance of all 3 testers is different.

FACT 4. jmdt has the highest performance computer of all 3 testers. You (hrmmm) said it yourself, and the Graboid deck thing pretty much proves it.



CONCLUSION 1.
Because jmdt has the highest performance computer, and computer speed is directly related to how fast you win, decks tested by jmdt will get better results than decks tested by the other two testers.

CONCLUSION 2.
This study does not give accurate results when comparing results from 2 or more different testers.

CONCLUSION 3.
Deck list on the first post does not give an accurate picture on the true speed of all the decks



HOW TO FIX THIS?
Option 1. One person tests all the decks with the same computer
Option 2. All three testers test all the decks and count the average (better option)

[jmdt]

While it is true I have the faster computer of the three of us, I did also look at several of the decks with faster ttw.  All three of us play on low quality with the sound off so differences shouldn't be that great as if we were playing a graphic intensive game.  We haven't tested the same deck yet, but all 3 of up did test the time for the spins and independantly we all came up with 8 seconds over a number of runs each.  This tells me permorfance is relatively standard between our computers.  I do agree however that more data is definately needed here as with the ttw study, we did 50 tests first to find the contenders and then went back for 200 games later.  The same thing will likely happen here eventually.  Also all 3 of us will independantly test the same deck and see what differences arise.  The 3 of us essentially took a break from testing, but will get back to it here soon when we have more time.  Much more work to be done here for a thorough study.  I have an analysis that is in the works that I have yet to post.

As to the the 2 shrieker rushes in question, I copied the stats for further analysis:

Antlion/Shrieker Rush (http://elementscommunity.org/forum/index.php/topic,664.0.html) - jmdt
Sec/game: 57.36, Sec/Turn: 9.592, Clicks/Game: 12.82, Clicks/Turn: 2.144, Score/Min: 18.577, Elec/Min: 18.766
Time overall (min.): 47.800, Score: 888, Electrum: 891, Wins: 50, Ems: 1, Losses: 0, Avg ttw: 5.98Mono-Earth Shrieker Rush (http://elementscommunity.org/forum/index.php/topic,482.0.html) - hrmmm
Sec/Game: 57.68, Sec/Turn: 9.979, Clicks/Game:13.14, Clicks/Turn: 2.273, Score/Min: 19.244, Electrum/Min: 19.723
Time overall (min.): 48.067, Score: 925, Electrum: 948, Wins: 50, Ems: 3, Losses: 0, Avg ttw: 5.780
While I see why you have a problem with the stats for these two decks since they came from 2 testers, I may have an explaination within these stats for why these numbers are actually pretty close.  As we see, the antlion version has a ttw of 5.98, versus a lower 5.78 for the nonantion version, while the aqnltion version has a faster 57.36 sec/turn versus 57.68 for the non antlion version.  When I first looked at these, I thought it was also a case of different cpu's.  Further analysis of the data shows where the difference comes in.  The antlion version has has a lower number of clicks per game, which leads to a lower clicks per turn, which leads to a lower time/turn, hence making the deck faster with a lower ttw.  The antlion version has fewer clicks since it plays a net fewer amount of graboids overall, and graboid requre 2 clicks to play versus 1.  Compounding the clicks thing, there is an animation time and delay when evolving a graboid which slows the game slightly.  I think the time difference comes from evolving more/less graboids, but as always more data will tell for sure.

[wavedash]

That said, it would be ideal to have one person (or multiple people) test ALL decks.
Of course, that's not necessarily possible.

[jmdt]

That said, it would be ideal to have one person (or multiple people) test ALL decks.
Of course, that's not necessarily possible.

I agree with the chart

It would be nice if I had time/cards to test everything but unfortunately I do not yet.

[Scaredgirl]

It would be nice if I had time/cards to test everything but unfortunately I do not yet.

One way to "fix" the situation would be to list the results by tester, not to mix them all up. For example all your results would be in one list, followed by a separate list of results by the next tester, etc.

Also it might be good to include one paragraph explaining the possibility of computer specs and style of playing affecting the results so that we cannot really compare results from two different testers.

While this might be easy to understand to many, there are also a lot of people who just look at the list and assume that the decks are listed from the fastest to the slowest, which is most likely not 100% correct. Like I said, I find it to be too big of a coincidence that you just happened to pick most of the fastest decks. More likely explanation is that either your computer performance or style of playing is skewing the results.

[hrmmm]

(...)
... there are also a lot of people who just look at the list and assume that the decks are listed from the fastest to the slowest, which is most likely not 100% correct. Like I said, I find it to be too big of a coincidence that you just happened to pick most of the fastest decks. More likely explanation is that either your computer performance or style of playing is skewing the results.

the "Pro Stats for Pro Grinders." in the title should be a hint that this data we gathered here needs a closer look to understand the "decks speed".

as you see jmdt picked also the fastest (in turns) decks. no surprise for me he get also the fastest games.

i was wrong in my post before. i dont think that computer performance can be that mutch different. we are not talking about a game you need to buy a new system for playing.
and we both are testing decks in low quality.
if this would still slowing a deck for avg. 0.3 sec/game, its still the same speed.
there are other inaccuracys in the time gathering method that may have a bigger impact and can also vary in a test by the same person.
is someone really caring for +-1 minute per hour?

playstile: the most decks we tested here cant be played in that different ways to get that big impact aswell.
jmdt, kuross and i had about the same ttw in the ttw study, if we tested the same deck.

its still just 50 games we tested for the most decks. you can see in the ttw study that results can vary alot between 50 games. of corse this list isnt 100 % correct .

if still some player think, this study is now the ultimate answer for the universe and everything, i just can say: 42
... and maybe jmdt should mark some passages in the introduction in red or yellow.

[hrmmm]

... something (almost) completly different:

more stats for:
quadruple machinegun -- little things get dangerous (http://elementscommunity.org/forum/index.php/topic,11894.0.html)

Games:   200
   
avg. Sec/Game:   82.17
Sec/Turn:             12.65
Clicks/Game:   18.43
Clicks/Turn:   2.84
Score/Min:   11.93
Electrum/Min:   11.99

Time overall (min.):    273.88   
Score:   3268
Electrum:   3283
Wins:   192
Ems:    3
Losses:   8

Avg ttw:   6.495
ttw Details:   
ttw 3:   0
ttw 4:   14
ttw 5:   43
ttw 6:   69
ttw 7:   32
ttw 8:   18
ttw 9:   8
ttw 10:   6
ttw 11:   6
ttw 12:   4

[willng3]

Angry Santa: Christmas for your Score!!! (http://elementscommunity.org/forum/index.php/topic,12659.0.html) (50 games)
Sec/Game: 75.84, Sec/Turn: 9.65, Clicks/Game: 15.84, Clicks/Turn: 2.02, Score/Min: 23.04, Electrum/Min: 25.73
Time overall (min.):  63.200, Score: 1456, Electrum: 1626, Wins: 50, Ems: 34, Losses: 0, Avg ttw: 7.86 

[Sir Valimont]

Although we've already had this discussion jmdt, I'd like to point out for the record that my slight variation for the Ultimate Speed Rush deck (Life EM) actually works better for me. In any event I think the statistical margin is minuscule. The only difference is: I switch out your two Long Swords for two Jade Staves. They hit for 2 less damage but the 5 healing per turn has given me EM in places where Long Swords would not have. I've also experimented switching an Improved Heal for a Green Nymph with good success ... but while I have yet to find a statistically significant difference in efficiency I actually suspect that latter implementation to be a bit slower.

[tinkady]

Angry Santa: Christmas for your Score!!! (http://elementscommunity.org/forum/index.php/topic,12659.0.html)
Sec/Game: 75.84, Sec/Turn: 9.65, Clicks/Game: 15.84, Clicks/Turn: 2.02, Score/Min: 23.04, Electrum/Min: 25.73

Code: [Select]

5f2 5f2 5f2 6rk 7ai 7ai 7ai 7ai 7ai 7dg 7dg 7dg 7dg 7dg 7dg 7dg 7dg 7dp 7dp 7dp 7dp 7dp 7dq 7dq 7ds 7ds 7ds 7ds 7ds 7ds

that is pretty darn good. not as good as SEM but very good. what was the EM rate? or did you not record that

[Toge111]

playstile: the most decks we tested here cant be played in that different ways to get that big impact aswell.

- I've been playing with Speed EM life deck and I can say that there are several player choices that affect the speed. Mostly it's about reducing redundance.

* Playing pillars when you don't need quanta
* Estimating / Not estimating how many actual turns it takes to win and saving quanta for heals instead of playing creature cards.
* Choice between playing cockatrice and frog.

Speed EM life deck is as straight-forward as they come, but even then there's room for improvement. Training, as well as general gameplay experience affects results. Sample size should be bigger and testers should be given time to familiarize themselves with deck before taking measurements.