[1.32] False Gods Stats Thread

[the dictator]

I won't make it mandatory in any way, and the current formulae don't use it, but I will leave room for it, even if only so we can get a more accurate sense of the general turns/hour of the community.

'like turns where recorded when time was the main factor' is not completely true though, Statmasta doesn't have and never had any TTW (or TTL) entries. Pella's latest study kinda failed, in the sense that it tossed quite a bit of earlier improvements to the formula out of the window, however it did allow tracking TTW, thanks to which I'm able to use some of those numbers for my testing, if I can find a fair way to allow data with skips involved.

[rob77dp]

I guess my point didn't get across

I would be against a full elimination of time for a few reasons.

Just keep it a stat you still record just like turns were recorded when time was the main factor. I am honestly interested in seeing how this affects stats

I suppose we both misunderstood each other a bit then.  I am not suggesting NOT recording time, rather that the EI indices not use it as a base for the final efficiencies.  It must, somewhere, be included in calculations / considerations though, in my opinion.  Afterall, it is one three digit number to record that can have IMMENSE value later after a volume of tests are completed for comparison to the current idea of typical turns-per-hour settings.

[Leodip]

I've had 30 minutes of free time and did THIS.
It's the easiest and accurate way to calculate FGei I could come up with. Note that this isn't to keep stats, but purely to calculate the FGei of a deck.
The drop value isn't accurated (I used Ungaros' Drop Rate based on Sim, but used 1160 for card value) but for a more accurated one I would have had to spend lot of time on calculations, I don't even know if I could do that (even though it would be nice if someone who knows how to did that).

[the dictator]

I've had 30 minutes of free time and did THIS.
It's the easiest and accurate way to calculate FGei I could come up with. Note that this isn't to keep stats, but purely to calculate the FGei of a deck.
The drop value isn't accurated (I used Ungaros' Drop Rate based on Sim, but used 1160 for card value) but for a more accurated one I would have had to spend lot of time on calculations, I don't even know if I could do that (even though it would be nice if someone who knows how to did that).

I reran those simulations, to have accurate numbers until I find time to calculate the exact values, you can find them in the spoiler in the 4th post.

As far as you're spreadsheet goes: I don't see anything accounting for the 30 electrum cost for a lost game (unless you include it in the EG column), and nothing accounting for EM's, aside from that, the formula matches the formula used on our spreadsheet for directly calculating the UEI[FG], but this is not the complicated part. It gets much more interesting once you get to the point of trying to optimize skips, and even more complicated once you try building in some tolerance factors, you can find the sheet here.

[Leodip]

I've had 30 minutes of free time and did THIS.
It's the easiest and accurate way to calculate FGei I could come up with. Note that this isn't to keep stats, but purely to calculate the FGei of a deck.
The drop value isn't accurated (I used Ungaros' Drop Rate based on Sim, but used 1160 for card value) but for a more accurated one I would have had to spend lot of time on calculations, I don't even know if I could do that (even though it would be nice if someone who knows how to did that).

I reran those simulations, to have accurate numbers until I find time to calculate the exact values, you can find them in the spoiler in the 4th post.

As far as you're spreadsheet goes: I don't see anything accounting for the 30 electrum cost for a lost game (unless you include it in the EG column), and nothing accounting for EM's, aside from that, the formula matches the formula used on our spreadsheet for directly calculating the UEI[FG], but this is not the complicated part. It gets much more interesting once you get to the point of trying to optimize skips, and even more complicated once you try building in some tolerance factors, you can find the sheet here.

As far as the spreadsheet goes, you have to account for everything in the EG section (it really is a basic one).
I'll be optimizing it when I have other thirty minutes (and I'll be using your numbers for the simulations, thanks). I think I'll be adding a way to record for wins, losses and EMs only in order to keep stats, a basic skip utility and am unsure about a couple of things.

[Leodip]

'kay, modified it. Put a way to calculate a personal TPH (which shouldn't, as I said in the instructions, be posted on the forums) and a personal FGei with that. Added Lost and EM only in order to keep track of games. Time To Finish is also on for the TPH calculation. Edited the Card Value with The Dictator's calculations. Added a way to calculate skips (even though pretty rough).

Spoiler for Hidden:

Yeah, I know I used some silly colors, but oh well.

[the dictator]

Opening posts updated, more to come (see to-do list), but I decided to post this already, so everyone can have a look, and comment on it.

[Leodip]

*snip*
Total turns to Win, yes, that is TURNS to WIN, Not Losses, Not Finish, not time.[/li][/list]
*snip*

What do you mean by that?

Another thing I'm not okay with is the number of tests. I would add another rule: 100 games and AT LEAST one lost game against each FG. This is to prevent 100% WR due to few tests. If needed, the extra games can be tested in the Trainer to get that specific god, if you need it.

Aside that, I like it.

[the dictator]

I was merely stressing that I want Turns to Win. So not Turns to Lose, or Time to Finish, or Turns to Skip.

While at least 1 lost game (and at least 1 won game) per god would be idea, I won't require it as part of the initial data, because it can take quite a few games with the right deck against the right god. Regarding trainer, yes, it could be used, but I'm reluctant to advice it to people, zanz might decide to update the AI in the trainer, changing the difficulty of the opponents.
Even with VDB I have multiple gods at 80% or higher winrate, so that would mean an average of 5 games. With a more consistent deck like PDials the winrates can get even higher, which would mean a silly amount of games required against some gods (which would imbalance the data, because while it is normalized, it still isn't really fair to compare a god with 10 or 15 games played to a god with only 3, their winrates are at completely different levels of accuracy).

[Leodip]

Are you talking about a TTW which isn't the same as TTF? TotalTTF is, as a definition, TotalTTW+TotalTTL. However, if TTW is equal to TTL, TTF will also be equal to TTW.
What I was asking is, then, when you say Turns to Win you are talking about the turns it takes you to win a game or both winning and losing them (skips aside)?

Also, I'd either change it to TTF or edit what you said in the "Math" part, where you said you'd be using TTF.

[the dictator]

Testers are only required to log their turns to win, which will then, in the math part be used AS if it were turns to finish. Thus in the tester instructions I was stressing that it should include only the wins, that was the reason for the extra clarification in the first place.

Later, in the math post I do indeed switch from turns to win to turns to finish (same actually happens in the spreadsheet too): at the point where I 'correct' the TTF for the gods that don't have any won games, so no TTW of their own.

So no, I won't be changing either: Testers are required to track total turns to win. Which will then be used to calculate average turns to win. That is 'converted' into turns to finish. While (average) Turns to Win and (average) Turns to Finish, are the same as a number, they are distinctly different variables with different uses, and thus both used (in different parts).

[Leodip]

I look forward to see how this will go, then.