SuperAPI AI Directory

zettaranc

Writing & research

View repository

Updated 10-16 00:02

About this skill

Zettaranc Trading System Skill

This skill provides the complete knowledge of the Zettaranc (Zhixing Xiaocainiao) A-share trading system, and can be used to query trading rules, analyze buy/sell signals, and ingest new material into the knowledge base.

System Overview

The Zettaranc trading system was created by Brother Z (Zhixing Xiaocainiao), and is arule-driven, discipline-firstA-share short-term/swing trading system. Core ideas:

  • Wait by the stump for a rabbit: only enter at fixed buy points (B1/B2/B3), do not chase highs
  • Iron rule of stop loss: 3-5 price levels below the low of the buy K-line, range ≤3%
  • Position discipline: supermarket mode diversified holdings, a single stock not exceeding 20% of total position

Three-module architecture

┌─────────────────────────────────────────────┐
│           07-心法哲学(道)                   │
│   交易心理 · 纪律执行 · 资金管理哲学          │
├─────────────────────────────────────────────┤
│        04-战法策略(法)  05-资金仓位(法)     │
│   少妇战法 · 坑口战法 · 双枪战法 · 量比战法    │
├─────────────────────────────────────────────┤
│  03-买卖信号(术)  02-底层工具(器)  06-市场筹码│
│  B1/B2/B3/S1/DSZ  KDJ/筹码/砖形图  主线/题材  │
└─────────────────────────────────────────────┘

Data source hierarchy

This Skill adopts a three-tier data access strategy, upgraded on demand:

Tier Data source Trigger condition Access method
L1 quick query references/ directory Routine questions (rules, parameters, processes) Direct read
L2 deep query Complete wiki pages Need source tracing, knowledge conflicts, complete context Local path or GitHub
L3 raw query raw/ raw material Need to see Brother Z's original words, verify citations Local path or GitHub

Data source detection (by priority)

  1. Environment variable:ZETTARANC_KB_PATH exists → points to the local knowledge base root directory
  2. Local sibling directory: the plugin sibling directory contains wiki/ → local knowledge base mode
  3. Remote access: if neither exists → access the repository via GitHub API chenleiitaz/zettaranc-knowledge

Detection method:

# 1. 检查环境变量
echo $ZETTARANC_KB_PATH

# 2. 检查本地同级目录(相对 plugin 安装位置)
ls wiki/zettaranc/index.md 2>/dev/null && echo "本地知识库可用"

# 3. 检查 GitHub 远程仓库
gh repo view chenleiitaz/zettaranc-knowledge --json name 2>/dev/null && echo "远程仓库可用"

GitHub remote access method

When the local knowledge base is unavailable, read remote files via gh api :

# 读取 wiki 页面(完整概念页)
gh api "repos/chenleiitaz/zettaranc-knowledge/contents/wiki/zettaranc/concepts/{分类}/{文件名}" --jq '.content' | base64 -d

# 读取 wiki 索引(总目录)
gh api "repos/chenleiitaz/zettaranc-knowledge/contents/wiki/index.md" --jq '.content' | base64 -d

# 读取 raw 原始素材
gh api "repos/chenleiitaz/zettaranc-knowledge/contents/raw/01-zettaranc/{子目录}/{文件名}" --jq '.content' | base64 -d

Note: Chinese characters in file names need URL encoding (e.g. B1建仓波.md → B1%E5%BB%BA%E4%BB%93%E6%B3%A2.md). If the file name is uncertain, first read index.md to get the complete list.

Query workflow

When the user asks a trading-related question, answer according to the following process:

  1. Determine the question type: which category it belongs to (signals/battle strategies/tools/mindset/chips)
  2. Load the corresponding Reference: read as needed references/ the corresponding file under
  3. Comprehensive answer: give a precise answer based on the reference content, annotate the source of information
  4. Cross-reference: proactively associate related concepts (e.g. when asked about B1, associate B2, stop loss, position)

Answer specifications

  • UseSimplified Chinese
  • Use for key rules bold annotation
  • When numerical thresholds are involved, give exact numbers (e.g. J<13, increase 30%, stop loss 3%)
  • If the knowledge base has no related content, explicitly state "The knowledge base does not contain this content," then provide general advice
  • When the answer exceeds 2 paragraphs and has analytical value, proactively ask whether to save it as a knowledge page

Deep Query Workflow

When L1 quick query (references) cannot meet the need, automatically escalate to L2/L3.

Escalation Trigger Conditions

  • The user asks "What exactly did Brother Z say," "What is the source," "Which article is it from"
  • The user asks "Is this concept controversial," "What do different sources say"
  • The user wants to see the complete wiki page for a concept (including related links, Mermaid diagrams)
  • The specific content the user is asking about cannot be found in references
  • The user requests "in-depth analysis" or "detailed expansion"

L2 Deep Query (complete wiki page)

1. 先读 wiki/index.md 或 wiki/zettaranc/index.md 获取页面列表
2. 定位目标页面:wiki/zettaranc/concepts/{分类}/{概念名}.md
3. 读取完整页面内容
4. 提取:完整定义、知识冲突区块、关联连接、来源列表
5. 回答时标注来源页面路径

wiki directory structure:

wiki/zettaranc/
├── concepts/
│   ├── 01-体系总览/      ← 知行交易模块、交易闭环、框架式交易...
│   ├── 02-底层工具/      ← 白线黄线系统、砖形图、MACD...
│   ├── 03-买卖信号/      ← B1建仓波、B2突破、S1信号...
│   ├── 04-战法策略/      ← 少妇战法、坑口战法、双枪战法...
│   ├── 05-资金仓位/      ← 底仓与动态仓、去弱留强...
│   ├── 06-市场筹码/      ← 筹码战争、顺周期轮动...
│   ├── 07-心法哲学/      ← 防守哲学、交易心理...
│   └── 08-操作手册/      ← 短线/长线操作手册
├── entities/              ← Zettaranc、国家队、麒麟会...
├── sources/               ← 7 个来源摘要
└── syntheses/             ← 综合分析报告

L3 Raw Query (raw source material)

1. 从 L2 页面的 frontmatter `sources` 字段获取 raw/ 文件路径
2. 读取原始文件
3. 提取 Z 哥原话
4. 用引用格式回答:> "原文内容" — [[来源文件名]]

raw directory structure:

raw/01-zettaranc/
├── 知行小菜鸟/            ← Z 哥本人(高可信度)
├── 知行课代表/            ← 课代表笔记(中可信度)
├── 复盘专用z/             ← 直播复盘(中可信度)
├── TANGOO/                ← 学生笔记(补充参考)
└── 大富翁小菜鸟/          ← 舰长班整理(高可信度)

Source Credibility Ranking

When answering, prioritize citing high-credibility sources:

Priority Source Description
1 Zhixing Xiaocainiao/ Brother Z himself, most authoritative
2 Dafuweng Xiaocainiao/ Systematically organized by the Captain Class
3 Zhixing Kedaiibiao/ Class representative notes
4 Fupan Zhuanyong z/ Livestream review
5 TANGOO/ Student notes, supplementary reference

Key Rules Quick Reference

Buy Signal Hierarchy

Signal Core Conditions Timeframe Priority
B1 Position-Building Wave KDJ J value < 13 Daily/Weekly Highest (wait for the rabbit)
B2 Breakout Confirmation bullish candle, trend established Daily High
B3 Buy Point Continuation confirmation, position-adding signal Daily Medium

B1 Core Iron Rules

  • J value < 13 (the lower the value, the better; -10 and -14 both count)
  • Two 30% Principles: Gain of about 30%, cumulative turnover rate ≤30%
  • Turnover rate >40% = invalid B1
  • 3-Day Discipline: Must resume rising within 3 days after buying, otherwise reduce position
  • Stop Loss: 3-5 price levels below the low of the buy candle, range ≤3%

Sell Signals

  • S1 Signal: High-level volume expansion with stalled gains, signs of main force distribution
  • DSZ Strategy: Combination of top divergence + shrinking volume + high-level doji

Position Management

  • Supermarket Mode: Diversified holdings, single stock ≤20%
  • Golden Pit Formula: (Neckline - Pit Bottom) + Neckline = Target Price
  • Do Not Chase Highs: Only enter at B1/B2/B3; other positions have no basis

Ingestion Workflow

When the user provides new material that needs to be incorporated into the knowledge base, load 05-ingest-workflow.md to execute the standard process.

Reference File Index

Load the following reference files as needed:

File Content When to load
01-Signal System.md Detailed rules for buy and sell signals such as B1/B2/B3/S1/DSZ When the user asks about buy/sell signals
02-Trading Strategies.md Details of strategies such as Shaofu/Kengkou/Shuangqiang/Liangbi When the user asks about specific strategies
03-Tool System.md Tools such as KDJ/chips/white line yellow line/brick chart When the user asks about technical indicators/tools
04-Mindset Philosophy.md Trading psychology/discipline/capital management philosophy When the user asks about mindset/discipline/philosophy
05-ingest-workflow.md Detailed workflow for ingesting new material When the user requests to incorporate new material

When references cannot satisfy the query needs, automatically escalate todeep query workflowto access the complete knowledge base.