After a whole day of testing 2 pirate 2 mech and 5 oil decks, writing and customizing mulligan and testing settings over and over again. i finaly found something that has a 66% win in 10 game (on primetime) at rank 15 though.

Big testrun will be tonight, so hopefully good results tomorrow, what kind of deck it is, is still a secret ;)

Damn Rogue is rly a hard one.

Deck, specific: secredin

External AI used y/n: y

Settings and datasets used (settings.txt, _combo.txt and _mulligan.txt):

Silverfish a generic set of rules (_mulligan.txt & _combo.txt)

Number of games played: 472

What levels was being botted at : r10 -> r4 -> r8 ->r7

Time of day played: 2.12 days = 50.77 hours = 3046 minutes

Timezone, +/- CET: Asia
Measured winrate, percentage: 56.8%

Potential screenshots (put in spoilers, cover anything identifiable):
Hide Spoiler

Deck, Task Statistics

---

enfacehp = 20; // hp of enemy when your hero is allowed to attack the enemy face with his weapon
maxwide = 5000; // numer of boards which are taken to the next deep-lvl
twotsamount = 1500; // number of boards where the next turn is simulated
simEnemySecondTurn = true; // if he simulates the next players-turn, he also simulates the enemys respons
playarround = true; //play around some enemys aoe-spells?
playaroundprob = 50; //probability where the enemy plays the aoe-spell, but your minions will not die through it
playaroundprob2 = 80; // probability where the enemy plays the aoe-spell, and your minions can die!
enemyTurnMaxWide = 24; // bords calculated in enemys-first-turn in first AI step (lower than enemySecondTurnMaxWide)
enemyTurnMaxWideSecondTime = 160; // bords calculated in enemys-first-turn BUT in the second AI step (higher than enemyTurnMaxWide)
enemySecondTurnMaxWide = 16; // number of enemy-board calculated in enemys second TURN
nextTurnDeep = 6; //maximum combo-deep in your second turn (dont change this!)
nextTurnMaxWide = 16; //maximum boards calculated in one second-turn-"combo-step"
nextTurnTotalBoards = 160;//maximum boards calculated in second turn simulation
useSecretsPlayArround = true; // playing arround enemys secrets
alpha = 50; // weight of the second turn in calculation (0<= alpha <= 100)
simulatePlacement = false; // set this true, and ai will simulate all placements, whether you have a alpha/flametongue/argus
useExternalProcess = true; // use silver.exe for calculations a lot faster than turning it off (true = recomended)
passiveWaiting = false; // process will wait passive for silver.exe to finish
behave = control; // select the behavoir of the ai: control, rush, face (new) or mana (very experimental, dont use that :D)
concedeOnBadBoard = false; // set this to true and ai will concede if enemy is to good
speed = true; //set this to true, to support sending multiple actions to bot

---

Other input/comments:

End of the Season, got the [Sea Giant's Treasure] .

So, I've gone spally->cw->a little spally again with that successful deck posted->yesterday I picked up from reddit someone predicting dragon priest winning in the current meta, so, I made that dragon priest deck. Meticuously updating to the very latest datasets.

It's all the same to me, every single morning after a full night of botting I'm somewhere between rank 11-10- no more than 9. Always the same.

I'm beginning to think blizz has boxed me into some bot hell, my HR efforts only serve as guinea pigs in their ai labs. lol.


Edit: tried several iterations per deck archetype too

Thats allot better then my deck did! iam curious :)

Thank-you for your data it's a shame only 40% win-rate =/

a deck should pretty much be the exact opposite huh

Hey guys, I installed the auto class switch tool and configured the directories, and installed all the proper data sets. But my win loss ratio has taken a dive. I think I might have something set up incorrectly. Was going for the dragon priest ruleset. But I don't think the logic is working as intended.

Here is a copy of a log.

What are common stumbling blocks when things don't seem to be working right?
I'd assume you'd have to create a deck that is card for card the exact same as the logic you'd be using in the data set.
But It's a little difficult to tell if silverfish is actually using that data set.

Thanks for testing the Pirate Rogue deck with me, I'm happy with a 50%+ win-rate on ladder and have added to the upload.

@everyone With the new silverfish update botmaker has added the support for hero power in combo.txt so I have put those back in for testing your feedback would be great :) also I have added a FaceHunter by luffyDmonkey 50% win-rate on ladder nice work :)

it says combo reader cant read combo txt (I already unblocked it )

Secret pally pretty much