360 Secure SkillHub
360 安全技能中心

Alphagbm Iv Rank

核心功能:计算个股隐含波动率(IV)排名、百分位、HV/IV比率等指标,基于IV区间给出买卖期权权利金的交易信号,提供252天IV历史数据。典型场景:判断IV高低、择时波动率交易、筛选IV极值、决策期权策略。输入为股票代码,可选回溯期;输出含指标、信号、历史数据的结构化JSON,支持AAPL等5个演示标的,无API KEY也可使用。

clementgu
clementgu
热度 13.76
最近更新 2026-07-23 14:14:00
分类:数据依赖:第三方 API / 读写文件 / 内部调用大模型应用场景:财务 / 数据处理 / 市场分析

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add alphagbm-iv-rank
Metadata
name
alphagbm-iv-rank
description
IV Rank and IV Percentile analysis showing where current implied volatility stands relative to its 252-day history. Returns IV rank (0-100), IV percentile (0-100), IV history data, and trading signals based on IV zone. Use when: deciding whether to buy or sell premium, checking if IV is high or low, timing volatility trades, screening for IV extremes. Triggers on: "IV rank AAPL", "is NVDA IV high", "IV percentile SPY", "historical IV TSLA", "is volatility cheap for META", "IV rank scan", "should I sell premium".
globs
["mock-data/*.json"]

AlphaGBM IV Rank

Prerequisites

  • API Key: Set env ALPHAGBM_API_KEY (format agbm_xxxx...).
  • Base URL: Default https://alphagbm.zeabur.app. Override with env ALPHAGBM_BASE_URL.

What This Skill Does

Calculates IV Rank and IV Percentile for any ticker, placing current implied volatility in historical context. Answers the key question: "Is IV high or low right now, and what should I do about it?"

Key Metrics

Metric Formula What It Means
IV Rank (Current IV - 52w Low) / (52w High - 52w Low) x 100 Where IV sits in its annual range. 0 = at the low, 100 = at the high
IV Percentile % of days in past year where IV was lower than today What % of the time IV was cheaper than now. 80 = IV was lower 80% of the time
Current IV 30-day ATM implied volatility The market's current expectation of annualized movement
IV 52w High Highest 30-day IV in past 252 trading days Peak IV -- usually during selloffs or events
IV 52w Low Lowest 30-day IV in past 252 trading days Trough IV -- usually during calm, grinding markets
HV/IV Ratio Historical Volatility / Implied Volatility >1 means realized vol exceeds implied (IV may be cheap)

IV Zones and Trading Signals

IV Rank Zone What It Means Suggested Action
80-100 Very High IV is near its annual peak -- options are expensive Sell premium: short strangles, iron condors, credit spreads
60-80 High IV is elevated -- above-average option prices Lean toward selling, but selective; good for covered calls
40-60 Moderate IV is in the middle -- neither cheap nor expensive Strategy-neutral; use directional view to decide
20-40 Low IV is depressed -- options are cheap Lean toward buying; good for debit spreads, long straddles
0-20 Very Low IV is near its annual trough -- options are very cheap Buy premium: long straddles, debit spreads, calendars (sell back month)

API Endpoint

IV Snapshot (instant, no quota cost)

GET /api/options/snapshot/<SYMBOL>

Returns: ATM IV, IV Rank, HV 30d, VRP, VRP level. This endpoint is free and does not count against your analysis quota.

Volatility Risk Premium (VRP)

VRP = Implied Vol - Historical Vol

VRP measures the gap between what the market expects (IV) and what actually happens (HV). It is a key signal for whether to sell or buy premium.

VRP Level Value Seller Buyer
very_high >=15% Very favorable Unfavorable
high 5-15% Favorable Slightly unfavorable
normal +/-5% Neutral Neutral
low -15% to -5% Unfavorable Favorable
very_low <-15% Very unfavorable Very favorable

How to Use

Input

  • Required: Ticker symbol
  • Optional: Lookback period (default 252 days), IV measure (30-day ATM, 60-day, or custom)

Output Structure

{
  "ticker": "AAPL",
  "price": 218.45,
  "iv_current": 28.5,
  "iv_rank": 42,
  "iv_percentile": 55,
  "iv_52w_high": 48.2,
  "iv_52w_low": 18.8,
  "iv_52w_mean": 30.1,
  "hv_30d": 25.2,
  "hv_iv_ratio": 0.88,
  "zone": "moderate",
  "signal": "No strong IV edge. Use directional conviction to choose strategy.",
  "iv_history": {
    "dates": ["2025-04-01", "2025-04-02", "..."],
    "iv_values": [32.1, 31.8, "..."],
    "hv_values": [28.5, 28.3, "..."]
  },
  "notable_events": [
    {"date": "2026-01-28", "iv": 48.2, "event": "Earnings spike"},
    {"date": "2025-08-05", "iv": 44.1, "event": "Market selloff"}
  ]
}

Example Queries

User Says What Happens
"IV rank AAPL" IV rank, percentile, zone, and trading signal
"Is NVDA IV high?" IV rank + zone classification + comparison to 52w range
"IV percentile SPY" Percentile with historical context
"Historical IV TSLA" Full 252-day IV history with HV overlay
"Is volatility cheap for META?" IV rank + HV/IV ratio + buy/sell recommendation
"Should I sell premium on QQQ?" IV rank-based answer with suggested strategies

Mock Data

Demo tickers available without API key: AAPL, NVDA, SPY, TSLA, META. IV history uses realistic 252-day data from mock-data/.

Related Skills

  • alphagbm-vol-surface -- Full 3D IV landscape across strikes and expirations
  • alphagbm-vol-smile -- IV skew for a specific expiration
  • alphagbm-options-strategy -- IV zone informs whether to buy or sell premium
  • alphagbm-options-score -- IV attractiveness is a key scoring factor

Powered by AlphaGBM -- Real-data options & research intelligence for traders and AI agents. 10K+ users.