General OUEI/Study InformationOUEI stands for "OpenEtG Unified Efficiency Index". It is an oetg-specific variation on vanilla's UEI. OUEI is reported in terms of gold per second, so "2 OUEI" means a deck should provide 2 gold per second, or 120 gold per minute, or 7200 gold per hour. "Per second" may seem like an odd choice, but it actually works pretty nicely once you get used to it, and if you really need to think of things per hour, it's easy to convert.
At the moment, OUEI is calculated by the
autostats spreadsheet using output from the autostats in-game function. It is calculated by taking the total amount of gold won plus the total wealth value of card drops obtained over a given time period, divided by the in-game time that occurred over that time period plus two seconds per game to account for the time between battles. This is all subject to change, and only provided for informational purposes, as all calculations are handled by the spreadsheet.
Because OUEI currently works based on actual results, any output will be
extremely personal. For the moment, no result here should be taken to mean anything besides "this is the results of this specific person at this specific time". One of the goals of these studies is to eventually gather enough data from diverse sources to be able to develop a more abstracted way of reporting results in order to make them more generally applicable and less personal. Because these tests will include raw autostats output, it will be possible to apply a more abstracted formula to previous tests whenever that more abstracted formula is developed.
Studies will run indefinitely. All test results will be tagged with their applicable balance patch, and tests from previous balance patches will be flagged as "old". A new set of studies will occur whenever a sufficiently major shift in the PvE environment occurs, but hopefully this will not be every balance patch.
How to ParticipateEssentially no effort is required to participate. The minimum that you need to do is play some games against the AI that meet the inclusion criteria, and then copy/paste the autostats output (Stats tab on top of chat) into one of the studies threads (ideally in a spoiler). If you are using the stats spreadsheet for your own information/records (which I strongly recommend), please include the summary information it provides (win rate, average plies/game, OUEI, etc), but it is not a requirement if you can't be bothered.
In summary,
raw autostats output must be included in any test results. No other information is
required to be reported, but it is appreciated.
To be considered a valid test, at least 50 games against the same target by the same person on the same computer in generally similar conditions must occur. For basic AI targets, these tests do not have to be completed in a single sitting, but all tests must be completed within the same balance patch. For arena, any given gameplay session must be at least 25 games long in a stretch, in order to ensure that the same arena meta is used. In other words, in order to report a test against arena, you must play two 25-game sessions (or one 50-game session) instead of five 10-game sessions or whatever. You may, of course, play longer than 25 games at a time - it is a minimum, not a maximum. If you are testing against arena, it would be appreciated if you screenshotted or copy/pasted that league's top 20 at the start/end of your session, but this is not a requirement.
In summary,
50 games minimum per test. For general AI, games must be in the same balance patch but otherwise do not need to be sequential. For arena, games must be in the same balance patch and must occur in blocks of at least 25 games.
Be sure to
include the deck code in every test result, even if the deck code already exists in the thread. This will ensure that we know what version of the deck is being used.
Some general notes on taking stats: Decide before starting a battle/session whether you will be keeping stats for it. Don't take a long break to chat in the middle of a match, since it will completely ruin your time - chat between duels. Try to end a test on a loss, so that you aren't ending in the middle of a win streak bonus. Do not "disqualify" tests because you misplayed or got terrible luck or whatever. Start with no win streak bonus.
A note on how the results will be reported, added 6 Mar 2015
Tests done by different people (or done by the same people in very different circumstances, e.g. one set done on a modern computer, another set done on an ancient netbook) will not be combined into one entry at this time. Results are currently so personal that there is no good way to do this without massively skewing things. Once more abstracted formulae are developed that let us use average spin drops values and figure out "typical" time/ply values, entries will be combined by deck. Multiple tests done by the same person in the same circumstances will continue to be combined into a single entry. If you need a notation added after your name (i.e. "cg (netbook)" or "cg (tablet)") just mention it when reporting your results.
Please help us with these studies in whatever way you can. Your effort investment is minimal if you're playing around already, and these will be of great help moving forward in figuring out what balance tweaks are necessary. Thanks!