And with a lot delay the wrapup of last season. My Bot Aicu actually performed horribly. It finished just in Champion I at 4753. It had pretty extreme Ups and Downs last season. Mainly because I decided to experiment heavily with monster distributions.
So, how did aicu-chan do?
Something like that, much better than last season. I haven't made big changes to aicu-chan that way it performed more like last season. It reached silver III with it's lvl 1 account.
Debugging
Like with many software development projects, debugging takes a large role in this one as well. I thought that by now, I had fixed most, if not all of the bugs in my Simulation. The simulation is the core part of my bot, and without a good simulation, the bot cant perform optimal. Thanks to @bubke I discovered today two more bugs in my simulation. And not even minor ones at that. I forgot to apply the damage buff from non summoner abilities like "Inspire". Without that, the damage boosting ability was completely ignored in my simulations. The next major bug was the incorrect calculation of the damage reduction from Void and Shield. I was looking for a quick way to calculate it and I noticed that the new damage is always the rounded down value of damage times two thirds. In a moment of derp I decided to apply that directly as the damage modifcation. So a melee damage of 3 was not reduced to 2, but 1. Because 3*(2/3) rounded down equals 2. 3 -2 -> 1. But It should have been 3-1 -> 2. Oh well. sometimes you can make major mistakes without even noticing them. I (hopefully) fixed those two bugs and I've pushed them to my github repository
I'm curious if those changes in the simulation reflect themselves in the performance of my two bots. If not, then it can't be changed. But it would be nice if it did.
Aside from that, I'll look a bit into deep reinforcement learning. It might actually perform pretty well with a game, where tracking the best teams already guarantees you a spot in the top 10 with a max level deck.
Giveaway
Like last time, this time I'm also doing a giveaway. The rules are as usual. If you comment below within the payout time I enter you in the random pool from which I draw the winners.
The cards this time are:
 |
 |
Oooo. A gold. That would be a nice win.
Your bots a most impressive! I wish I had the skill to make one. it looks like a fun puzzle to solve. 😁
Well , it might be yours if the RNG gods favor you ;)
And thanks! a puzzle describes the whole thing pretty well, a tricky but a fun one :)
Posted using Partiko Android
And the winners are:
@mano237 Grumpy Dwarf
@gillianpearce Gold Grumpy Dwarf
@handtalk5 Serpentine Soldier
Thank u 8)
Posted using Partiko Android
Thanks so much @bauloewe. A gold card's a great prize. I appreciate it. 😊
i am sure it was hard work to get so far so gz what you have achieved :)
Hey, thank you :) it's been three months since I started working on aicu, more time than I anticipated in the beginning :)
Posted using Partiko Android
Finishing in Champion 1 is far from horrible!
Posted using Partiko Android
You are right, champion one is usually top 100, so really far from horrible :)
Posted using Partiko Android
We should make it an official competition and have a leaderboard for botted starter accounts! The rules should be:
We could have a steem account that does weekly posts showing the leader board!
That sounds like a great idea :) I'd actually limit it to lvl 1 summoner cards. Then the level of the other cards doesn't matter and that way everyone has the same condition and results get comparable.
For other accounts a general bot ranking could be an option.
hi @bauolewe, please could you help me understand how the secret and team_hash fields are generated by sending the customJSON to the blockchain, my idea is to program a simple bot