来自量化系列视频1的技能：如何利用对冲基金方法赢得每一笔交易。

框架由Roan（@RohOnChain）制作——我是安装摄像头的人。

---

安装（标题路径——两个命令）

在Claude代码中：

就这些。技能现已安装。随时在任何资产上调用：

…或者直接用通俗的英语问：“detection the regime on BTC-USD”， “在我的SPY动量策略中添加政权确认过滤器”， “AAPL的长期政权组合如何——是否尾部过重，难以交易？” 克劳德自动释放了regime技能。

没有API密钥。没有账户。没有sudo。依赖关系在首次运行时被解析 uv（PEP 723 内联元数据）——无需自行安装 PIP。

---

技能的作用

它回答了任何资产的一个问题：我们处于什么体制，粘性有多强 这对风险和方向意味着什么？

• 每天通过滚动回报规则标记多市/熊市/横盘（默认20天，±5%）
• 从资产历史（最大似然）构建3×3转换矩阵
• 通过将矩阵提升为幂来预测n步（Chapman-Kolmogorov）

Markov Hedge Fund Method

Skill from video 1 of the Quant Series: How To Use The Hedge Fund Method To Win Every Single Trade.

Framework by Roan (@RohOnChain) — I’m the guy installing it on camera.

---

Install (the headline path — two commands)

In Claude Code:

/plugin marketplace add jackson-video-resources/markov-hedge-fund-method
/plugin install markov-hedge-fund-method@markov-hedge-fund-method

That’s it. The skill is now installed. Invoke it any time, on any asset:

/markov-hedge-fund-method:regime

…or just ask in plain English: “detect the regime on BTC-USD”, “add a regime confirmation filter to my SPY momentum strategy”, “what’s the long-run regime mix of AAPL — is it too tail-heavy to trade?” Claude fires the regime skill automatically.

No API keys. No accounts. No sudo. Dependencies are resolved on first run by uv (PEP 723 inline metadata) — nothing to pip-install yourself.

---

What the skill does

It answers one question for any asset: what regime are we in, how sticky is it, and what does that imply for risk and direction?

• Labels every day Bull / Bear / Sideways via a rolling-return rule (default 20-day, ±5%)
• Builds a 3×3 transition matrix from the asset’s history (maximum-likelihood)
• Forecasts n-steps ahead by raising the matrix to powers (Chapman-Kolmogorov)
• Computes the long-run stationary distribution (baseline regime mix)
• Emits a signed signal: bull_prob − bear_prob → direction + conviction
• Runs a walk-forward backtest (no lookahead) → reports Sharpe + max drawdown
• Optionally fits a Hidden Markov Model via hmmlearn (graceful degrade if it can’t compile)

It takes either a ticker (--ticker BTC-USD, fetched via yfinance) or your own CSV (--csv my_prices.csv, just a date + close column) — so it drops into whatever data pipeline you already run, on whatever asset you trade.

It’s built to compose: slot it into a trading agent you already have as a confirmation layer, a standalone signal, or a tail-risk filter — without rewriting your strategy. See skills/regime/SKILL.md for the JSON contract and three worked composition patterns.

---

The on-camera build / zero-trust manual path

markov-hedge-fund-method.md is the original one-shot onboarding prompt — the version built live on camera. Paste it into Claude Code (agent mode) and it builds the whole skill from scratch in front of you: detects your OS, installs uv, writes every file, runs the sanity check.

It’s kept here as the zero-trust path: if you don’t want to install a plugin from a marketplace, this builds the identical logic locally so you can read every line as it’s written. Most people should use the two-command plugin install above — this is the transparent fallback and the on-camera artifact.

---

Pine Script bonus