Concerning skipping - FYIHere are the thoughts we had on the skipping issue.
What kind of study is the FG-efficiency study?- A high performance study geared towards finding the very best deck under optimal conditions?
As we know by now, the FGei value can be tipped significantly by seemingly minor circumstances:
- the performance of your PC
- whether you skip at the right time or not
- if you are a superior or average player
- plain old luck with the RNG (Rainbow does drag the FGei down ... oooh yes)
In order to find the best performing deck, testers would have to be selected and balanced
while taking all of these aspects into the strictest possible consideration.
In fact, you wouldn't even need a bunch of testers or a large sample size because in
peak performance-studies all you need is ONE top-guy who proves how far the gear can go.
Using a large array of testers, quickly brings up questions about how certain samples
are being "polluted" by mediocre conditions.
Having, say, Essence the deck-creator prove that the Essential-newbie-FG-farmer
can get an FGei of 4000 is as interesting for the average user as marveling at the
skills of a professional Ralley-pilot driving a 1960 pick-up truck.
Interesting for sure, but can he ever relate?
- An average performance study geared towards finding the best decks under regular farming-conditions?
Who farms FGs? -> More or less everybody or at least everybody who wants to get electrum.
Does everybody have a top-PC? Is everybody a seasoned vet? How many people know
every trick in the book when playing a certain deck and would also quit at the perfect moment?
How many people pay 100% attention when farming to get the best out of their deck?
I guess you see where we are going here:
We decided on designing the study as a kind of broadband-study that will give
anybody who wants to farm FGs a realistic impression what he can expect from
certain decks. Of course not every greenhorn is testing the decks but the above-
mentioned pollutions are so to speak intended since they do happen in everyday
farming all the time.
-> A FG-decks purpose is not to win a championship, it is to get many people rich.
-> A FG-deck is just as good as the performance of its entire user-crowd.
What can/can't the skipping policy do?- Purging god-based data / the "skip-god" segment
Yes, playing Hermes, EternalPh etc. with mono-aether most likely isn't worth the time.
When a certain sample-size (at least 500, hopefully 1000 games) is reached, for each
deck ALL games against gods where a deck couldn't breach a certain FGei-threshold
will be purged from the data: ALL games will be set to "skip" and the game-time to
e.g. 5secs for quitting and loading up the next god.
This is already a giant leap towards a decks suggested optimal performance,
since with a large sample size the deck-build and it's performance against
certain gods deck-build will really show. Also, insta-quits are much more efficient
than those quits you do in turn 10 after realizing you will
probably lose.
This study will yield an accurate farming guide for each deck which is based on
data that has been gathered by at least trying to make the best of every god.
- Purging single bad games from the "doable god" data-segment
This is the tough spot and we really can't do it.
The occasional "should have gone well but then I got RoFed at the wrong time"-game
will unfortunately remain in the data as long as the particular gods data hasn't been purged.
Fortunately, the chances are high that many of those games actually are in the above
"skip-god" segment already and have indeed been purged already.
The way, to approach the games against "doable" gods is the "95% sure you can't
win"-policy which gives some freedom of choice to the more experienced testers.
Once again, especially if you are playing against a normally doable god (which won't
get purged), chances are that the time you have spent so far was well worth trying
and that those extra 30secs you spend actually losing/reaching the 95%-state
won't have a very large impact on the FGei.
The only alternative to such a procedure would be to overly careful select the testers
to have a proven "elite" test the decks, which immediately results in a study-design
of the "best-optimal-performance"-type.
Otherwise, you can never be sure if a player is actually good enough to know his skips.
He may be just a bit too impatient or not favour skips at all ... a lot of variables for
an aspect that counts large, because:
-> Actually winning a game is worth a lot of FGei, no matter if it was tough.
-> Saving those 30secs four times by skipping but therefore missing out on a
single lucky win in those four games is actually counterproductive.
We are positive that with
1. The high chance that semi-skip matches are in the "skip-god" segment anyways
2. The chance that lost time in the "doable-god" segment is somewhat balanced by lucky wins
a realistic and yet optimized performance can be shown for each and every deck.
----------
Please comment and suggest for any future improvements.
