<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Balatro on Wang Huijiu's Site</title><link>https://styleofwong.cn/en/tags/balatro/</link><description>Recent content in Balatro on Wang Huijiu's Site</description><image><title>Wang Huijiu's Site</title><url>https://styleofwong.cn/apple-touch-icon.png</url><link>https://styleofwong.cn/apple-touch-icon.png</link></image><generator>Hugo</generator><language>en-US</language><lastBuildDate>Sun, 28 Jun 2026 19:30:00 +0800</lastBuildDate><atom:link href="https://styleofwong.cn/en/tags/balatro/index.xml" rel="self" type="application/rss+xml"/><item><title>Taming Balatro: Pushing White Stake Clear Rate to 99% with Reinforcement Learning</title><link>https://styleofwong.cn/en/posts/balatro-rl-design/</link><pubDate>Sun, 28 Jun 2026 19:30:00 +0800</pubDate><guid>https://styleofwong.cn/en/posts/balatro-rl-design/</guid><description>If we want to achieve a 99% Ante 8 clear rate on Balatro White Stake, how should the RL algorithm be designed? This article presents a concrete technical blueprint from game-mechanics modeling, state/action spaces, reward shaping to algorithm selection, and explains why pure PPO won&amp;rsquo;t work and why a hybrid MCTS+RL architecture is necessary.</description></item></channel></rss>