AI3 vs AI4, electrum grinding for new players

[kev]

A while back I ran a little experiment wherein I started a new account and tried to see how long it’d take to get a fully upped FG grinder.  I learned that getting the electrum to kickstart a deck is easy.  The bottleneck is the rares on which so many FG grinders rely.

During the experiment I used The Essential Halfblood Grinder against AI4, won upgraded cards quickly, and have been recommending AI4 over AI3 to new players since.  Recently I decided to test this thought using The Essential against AI4 versus MrBlonde’s Mono-Death against AI3.

Because there are 13 set AI3 decks, it’s fairly easy to calculate electrum earned per win.  But due to the huge number of AI4's possible decks, it’s messier.  Here I recorded the spins from 300 AI4 wins and found an upped spin rate of 7.7% per win and an unupped spin rate of 22.3% per win.  Now we can calculate and compare the average electrum earned.

Wins: 833
EMs: 119
Average TTW: 8.648259303721488

Electrum cost paid (remember cost is refunded if you win): -10*167=-1,670
Electrum won from EMs: 40*119=4,760
Electrum won from ordinary wins: We’ll call it half of 20 per win.  10*714=7,140
Sale value of earned cards (average of AI3 decks from this thread): 25.60*833=21,325
Five electrum for spinning XXY: I’m ignoring.
Total: -1,670+4,760+7,140+21,325=31,555

Wins: 765
EMs: 44
Average TTW: 13.423529411764706

Electrum cost paid (remember cost is refunded if you win): -20*235=-4,700
Electrum won from EMs: 80*44=3,520
Electrum won from ordinary wins: We’ll call it half of 40 per win.  20*723=14,460
Sale value of earned unupgraded cards (average of all cards, uses unupped spinrate above): 43*223=9,589
Sale value of earned upgraded cards (average of all cards, uses upped spinrate above): 1,170*77=90,090
Five electrum for spinning XXY: I’m ignoring.
Total: -4,700+3,520+14,460+9,589+90,090=112,959


Conclusion: The average game using The Essential Halfblood Farmer against AI4 earns you about 3.6 times as much electrum as the average game using Mono-Death against AI3.  Given that an AI4 game probably takes two to three times as long to play (the ttw disparity above is partially responsible), the real conclusion is that AI4 is probably only slightly ahead of AI3 in terms of electrum per hour.

[Marsu]

"only slightly ahead" is still a very interesting result, as almost everyone here advises new players to skip AI4, which I never liked. Thanks for proving that AI4 makes perfect sense. Besides, it's way more fun anyway.

[whatifidogetcaught?]

"only slightly ahead" is still a very interesting result, as almost everyone here advises new players to skip AI4, which I never liked. Thanks for proving that AI4 makes perfect sense. Besides, it's way more fun anyway.

If it wasn't for the 150HP (instead of 200), I would NEVER recommend AI4 to anybody. With that update awhile back, AI4 is certainly great to farm.

[Acsabi44]

What is hard to model is the rare card spins. My gut says that AI3 is better.... Who knows?

[Marvaddin]

Win rate and EM rate of your AI3 deck looks really bad, while, at other hand, your win rate vs AI4, and mainly upgraded card drop rate looks really high.

I dont know if these are the best unupgraded decks for each environment (of course I understand this is somethig to help newbies), but surely AI4 will never get close to AI3 grinding with a fast EM deck. Because of this, I recomend upgrading a good AI3 farmer before even to build any of those "almost unupgraded" FG decks.

I will do some AI3 games with upgraded deck and post the results later.

[Chapuz]

1) +rep for this intense study.
2) I still suggest making an unupped FG grinder as soon as you can (Firecell up to now) and possibly change to Limitless Speed or Rol-Hope after the first 9k. As they only give upped cards, electrum gain might be higher with unupped Firecell than AI4 with that mono 

[TribalTrouble]

To each his own.

Overall, I believe that pre-1.32 ai3 > ai4 farming.
post-1.32 ai4 is about = ai3 farming unupped. Upped, however, a good rush deck can go through ai4 so fast that it almost feels like ai3. Frankenshard was proof of this for me. However, ai4 spins are a lot more annoying >.>

[timetock]

AI4 never made sense anyways. A better EM rate with AI3 is a lot more consistent than the occasional EXTREMELY LOW CHANCE rare care from AI4

[Leodip]

Well, you forgot to calculate how long does it take for every game. From my (little) experience games are quite long, and could even be stressing against some bad match-ups or rough ones.
As I said, however, if you aren't against selling upped cards from the spins, AI4 could be better. However in all of my games I won just 12 cards from AI4 (I think they were something like 300, and of the 12 cards only 4 were upped, not that it's bad, though), but that's just me, probably.
As I said in the other topic, I think that the better route for a newbie would be: Start with darkness, build mono-dark, grind and, if you think it's right to sell upgraded cards from the spins, build the Essential Half-Blood Farmer, then get an Immorush for Bronze and so on.
However, recently, I've been into farming PvP1. If you're new to the game you can easily get tons of money in PvP1 with an Immorush that's easily buildable with the Darkness starter. Sure, this is even slower, but pays alot and Immorush works awesomely in there. You can even get shards and nymphs.

[kev]

What is hard to model is the rare card spins. My gut says that AI3 is better.... Who knows?

Using this and this you could calculate the difference.

I will do some AI3 games with upgraded deck and post the results later.

Sure.  I'd love to see a comparison of upped AI3 and upped AI4.  You can use the framework I've provided but I'd recommend you use Simulator for both rather than playing some games yourself.

2) I still suggest making an unupped FG grinder as soon as you can (Firecell up to now) and possibly change to Limitless Speed or Rol-Hope after the first 9k. As they only give upped cards, electrum gain might be higher with unupped Firecell than AI4 with that mono 

You may be right.  If you have a reliable winrate and EMrate for unupped Firecell, you could easily calculate electrum gain per game using the framework above and the drop rates DarkWeaver has provided.

Well, you forgot to calculate how long does it take for every game.

Actually I specifically commented on game length.  I'm aware your anecdotal evidence doesn't agree with my conclusion.  Also you can't win nymphs from PvP1.

[Leodip]

I said so, it's just me that can't spin well enough, I'm not against AI4 farming, instead, I'd suggest it to new players, too.

[Marvaddin]

Im doing the games myself already, Kev. Because I have to take note of all the progress, cards, double spins, etc, Im not considering real time (and Im not taking note of TTW, because this is probably already a known data for TADAbow). Now I have 44-1, 35 EM. But I will do at least 200 games.

I could do an AI4 test upgraded, too, but which deck to use is something that comes in mind. I have seen you have a small list here. Which one would you recommend (I care about high win rate)?

Looks like in your last simulation you got 23% less upgraded cards than in your previous sample. What a big difference. And even more considering that this was the main electrum source for AI4 games.


Edit: update of current AI3 upgraded test:
100 wins
82 EMs
3 losses (2 were very unprobable)
37 cards (later I will calculate money)
Double spins: 51 (+37 of the cards) = 88