Yesterday I pushed to rank 5 with my version of mid-range Shaman - Deck changed to this season from Tempo Storm with 2 cards changed acording to a comment. So it's this deck with these changes from fiat500 comment " My version of this deck doesn't have stormcrack or Argent Horserider and two faceless." But the decklist is also in my files

My winrate from last night hitting rank 5:


In the next few days i will perfect my files, look for misplays in the logs and post them in the silverfish tread. But wanted to share my files as they are now. since I hit rank 5 with it.


Here are the files

Yes sorry! it was Thunder Bluff - I changed it now

Al'Akir is a finisher in this deck esp combined with rockbiter, but if you don't have it i would considdering using doomhammer instead. But then u need to add this combo to your _combo.txt:

Don't feel too bad about it. I am at 48% WR at rank 14 with scottbreak midrange shaman. Meanwhile, someone already reached rank 5 this month with the same deck/settings. I think we are just unlucky, playing at the wrong times, or in a more difficult zone. I expect it will pick up in the next week.

I don't feel there should be any huge differences between the bots like that, assuming they're working properly. A bot that has no human emotion, can't tilt, plays the same constantly, should not be getting rank 5 with one person and rank 14 with another, all other things considered. Especially if the deck is tier 1. My bot is averaging a 57% win rate at rank 12, and I had to intervene in 2 games, because I'm seeing tons of misplays and I have no idea how to correct them or alternate it's behavior. But when people are hitting rank 5, and I'm stuck at rank 12 while my bot passes on 10 mana without playing Steward of Darkshire against an empty board with 1 card in the enemy player's hand, something is up. But to the above point, the last stats from Blizzard show that almost all classes are represented between 7-15%, so even if it feels like you're playing nothing but the same classes, statistically it evens out.

That's what I ended up doing :)

Thanks!

Right, you are correct, which is why I don't really know what to say. It seems odd that the bot would pass with that hand, but if it's doing the mathematical calculations it's possible I';m missing something., which makes it hard to determine if it's an error or I'm wrong. It also doesn't help that I'm not 100% sure it's set up properly, lol. I have the extern bot file path with the updated settings / combo / mulligan files, and also in the base folder as well, just because I was paranoid it wasn't working.

And you're right, with a 57% win rate it should be rising, and (theoretically) eventually capping at rank 5 as long as it stays over 50%, due to the win streaks. But after 32 games I had only gone up 2 stars, so wasn't sure what to think. I have no problem letting it run for 30-50 games a day if it's working properly. I'm not one to whine and complain about hitting Legend in 4 hours or anything like that, I was just trying to determine if it's working. I'm very thankful that it's rank 10, just right now, even if it never goes higher. I could push to 5 manually if necessary for the rewards. But if it gets to 5 on its own, I can try to push for Legend each month for the card back.

I suppose you're right about the trends. I remember last month I played rank 7-5 for an entire day and literally saw 100% warrior, warlock, rogue, shaman, despite the overall picture that they each make up roughly 1/9 of the meta on any given day. But I'm using Secret Paladin because it seems like most people in this thread agree that it's less computationally extreme than the other decks so there's less of a chance of roping, plus the games don't typically go 20 turns. It's just weird seeing things like playing Muster on an empty board, then passing without attacking face, even though Truesilver is in hand and the bot probably has a 95-100% chance to play it the following few turns so it's missing 1 damage for no reason, not playing Repentance on turn 6 with nothing else in hand, etc. But I'm going to just let it go a few days on its own, then I'll post the results, and hopefully all is well

Edit: After another 30 games, it's regressed down to 53% win rate, probably going to go under 50% soon. Doh

http://pastebin.com/raw/3ZmbUw0T Here's a game where the bot kept 5-6 secrets in hand, never playing them, ever

The AOE playaround code is simplistic and outdated for the current meta. I would not recommend using it unless you test and find it offers improvement. Brawl is not even considered in the AOE playaround code.

Enemy Class - Cards Considered:
Warrior - whirlwind
Mage - flamestrike, blizzard
Hunter - unleashthehounds
Priest - holynova
Shaman - lightningstorm
Paladin - consecration
Druid - swipe

Does anyone know how I can make it so it only plays reliq seeker only if it can activate its battlecry?

Also, how do I get it to play peddler more often? It seems to value tapping over peddler.

Put a large card value on it. You can modify the board in your test.txt file and see how much the boardvalue is increased by a 5/5 reliqseeker. That could serve as an upper bound on the card value. I expect you need >10.