DOCSEARCHbyAlgolia
Submit Docs

AlphaGBM Skills

repository·main·Indexed 23 days ago

https://github.com/alphagbm/skills

A real-data options and research intelligence layer providing 29 specialized skills for traders and AI agents. It includes the alphagbm-cli (v0.1.0) for stock and options analysis, offering verifiable market data such as IV, Greeks, and VRP to prevent LLM hallucinations in workflows like Claude Code and Cursor. Key features include stock analysis (G=B+M model), options scoring, volatility surface analysis, and P&L simulation.

Tokens
34.5K
Snippets
66
Records
169
Agent score
80%

What's inside AlphaGBM Skills

  1. Overview of AlphaGBM Skills

    main

    AlphaGBM provides 29 AI skills categorized into several functional areas to bring real-time options and research intelligence into AI workflows (Claude Code, Cursor, Windsurf, etc.).

    Core Analysis

    • Stock Analysis: Fundamentals, momentum, and risk scoring (G=B+M model).
    • Options Score: 0-100 scoring across 4 strategies.
    • Options Strategy: Strategy builder and scanner.
    • Vol Surface / Vol Smile: 3D IV and skew analysis.
    • Greeks: Greeks calculator and IV solver.
    • P&L Simulator: Position what-if analysis.

    Data Intelligence

    • IV Rank: IV percentile vs 252-day history.
    • Earnings IV Panel: Implied move and crush history.
    • Unusual Activity: Large block and smart money detection.
    • Market Sentiment: VIX, Put/Call, and Fear & Greed.
    • VIX Status: 5-tier fear thermometer.
    • FearScore: Per-ticker panic composite.

    Workflow & Risk

    • Compare: Side-by-side analysis.
    • Watchlist / Alert: Monitoring and notifications.
    • Hedge Advisor: Scenario-driven hedging specs.
    • BPS Backtest: Bull Put Spread backtesting.
    • Take-Profit Lab: Exit strategy backtesting.

    Investor Masters

    • Duan-Yongping / Buffett / Marks / Tepper: Mechanical translations of specific investment philosophies into actionable tools.

    Knowledge Base

    • Company Profile / Investment Thesis / Macro View / Theme Research / Health Check: Tools for building and auditing a personal research workspace.

  2. Backtest Bull Put Spread (BPS) strategy

    main

    The alphagbm-bps-backtest skill performs a walk-forward backtest of a Bull Put Spread strategy (short put + long put at a lower strike) from 2018 to the present.

    It executes two simultaneous passes to quantify the value of the FearScore signal:

    1. With Signal: Only enters trades when the per-ticker FearScore is $\ge$ your specified fear_threshold.
    2. No Signal (Control): Enters trades unconditionally every Monday.

    This allows you to compare whether the signal actually delivers alpha or if you are simply paying for noise. The skill returns equity curves, key performance indicators (KPIs), a full trade ledger, and a plain-language summary in English and Chinese comparing the two passes.

  3. Understand Volatility Risk Premium (VRP) levels

    main

    The Volatility Risk Premium (VRP) is calculated as VRP = Implied Vol - Historical Vol. The skill categorizes the VRP level to help determine if volatility is a favorable environment for sellers or buyers:

    VRP LevelThresholdSeller SentimentBuyer Sentiment
    very_high>= 15%Very favorableUnfavorable
    high5-15%FavorableSlightly unfavorable
    normal+/- 5%NeutralNeutral
    low-15% to -5%UnfavorableFavorable
    very_low< -15%Very unfavorableVery favorable

  4. Analyze Volatility Risk Premium (VRP)

    main

    The Volatility Risk Premium (VRP) measures the gap between Implied Volatility (IV) and Historical Volatility (HV).

    Formula: VRP = Implied Vol - Historical Vol

    Use the VRP level to determine if there is an edge for selling or buying premium:

    VRP LevelValueSellerBuyer
    very_high>=15%Very favorableUnfavorable
    high5-15%FavorableSlightly unfavorable
    normal+/-5%NeutralNeutral
    low-15% to -5%UnfavorableFavorable
    very_low<-15%Very unfavorableVery favorable

  5. Understand AlphaGBM metrics and computation

    main

    AlphaGBM provides verifiable financial metrics computed directly from market data rather than LLM estimations. Key metrics include:

    • IV (Implied Volatility): Computed via Black-Scholes on actual bid/ask prices.
    • IV Rank: Current IV relative to 252 trading days of history.
    • VRP (Volatility Risk Premium): Implied Vol - Historical Vol, measuring option overpricing.
    • Option Score: A weighted score (0-100) based on premium yield, support/resistance, safety margin, trend, PoP, liquidity, and time decay.
    • Stock Score: A 0-10 score calculated as G = B + M, where B (Basics) includes PE, PEG, growth, and margins, and M (Momentum) includes VIX, technicals, and flow.
    • Risk: An additive score (0-10) based on valuation, growth, liquidity, market, and technical factors.
    • EV (Expected Value): Calculated as 50% × 1w + 30% × 1m + 20% × 3m expected value.

  6. The Signal Logic and Quality Filter Criteria

    main

    For the signal to return signal: true and level armed, three specific gates must all pass:

    1. VIX ≥ 35: Represents extreme market fear.
    2. FearScore ≥ 80: A multi-indicator panic index (reusing the alphagbm-fear-score module).
    3. Quality filter: The target ticker must have a market cap > $50B AND profit margin > 0. This prevents the signal from triggering on 'meme stocks', concept stocks, or pre-revenue small caps.

  7. Dimensions compared by AlphaGBM Compare

    main

    The comparison skill evaluates tickers across the following four dimensions:

    • GBM Five Pillars: Momentum, Value, Quality, Volatility, and Sentiment scores.
    • Options Metrics: IV rank, IV percentile, VRP, skew, and term structure.
    • Technicals: RSI, MACD, moving averages, and support/resistance levels.
    • Valuations: P/E, P/S, EV/EBITDA, and PEG ratio.

  8. Understand Earnings IV concepts

    main

    The alphagbm-earnings-crush skill uses several core financial concepts to drive its analysis:

    • IV Crush: The sharp drop in implied volatility that typically occurs immediately after an earnings announcement.
    • Average Crush %: The mean IV decline from the pre-earnings peak to the post-earnings trough, calculated over the last 8 quarters.
    • Implied Move ±X%: The price range the options market is pricing for the earnings move, derived from ATM IV × √(DTE/365).
    • IV Rank: The current ATM IV percentile relative to the 20-day Historical Volatility (HV) over the last 2 years. This drives the strategy recommendation:
      • IV Rank > 70: Short-IV plays (e.g., Iron Condor).
      • IV Rank < 30: Directional plays (e.g., Long Call/Put).
      • IV Rank 30-70: Wait/Neutral.
    • Iron Condor Quote: A ready-to-trade 4-leg spread with short strikes positioned at ±1× the implied move.

  9. How the Buffett Analysis skill works

    main

    The alphagbm-buffett-analysis skill applies Warren Buffett's specific investment framework to US stock tickers. Unlike generic momentum-based analysis, this skill focuses on four weighted dimensions to determine if a stock is suitable for long-term holding:

    1. Business (20% weight): Evaluates the 'circle of competence' based on sector simplicity (e.g., consumer staples score higher than complex tech/healthcare).
    2. Moat (30% weight): Measures durable advantage using thresholds like Gross Margin > 40%, ROE > 20%, Profit Margin > 15%, and Market Cap > $100B.
    3. Management (15% weight): Uses dividend continuity, payout ratios (ideally 15-60%), and 5-year average dividend yield as proxies for capital allocation quality.
    4. Valuation (35% weight): Checks for fair pricing by comparing PE, PEG, and FCF yield against the 10Y Treasury yield.

    Based on the weighted score, the skill returns an overall verdict:

    • ≥ 75: HOLDABLE (Green)
    • 55-74: WATCHABLE (Amber)
    • < 55: AVOID (Red)

  10. How the Chokepoint Analysis methodology works

    main

    The alphagbm-chokepoint skill implements a "Chokepoint Theory" framework to identify physically irreplaceable bottleneck suppliers in the AI supply chain. A true chokepoint is a supply-chain node that satisfies five specific criteria simultaneously. The methodology focuses on finding small-cap, near-monopoly suppliers buried 4–7 layers deep whose capacity constraints force violent repricing when demand outgrows supply.

    The 5-Factor Chokepoint Test

    Each factor is scored 0–100, and the final Chokepoint Score is a weighted composite:

    1. Concentration (25%): Measures market share (e.g., Top 1–3 suppliers hold $\ge$ 70% market share; HHI > 2500).
    2. Irreplaceability (25%): Measures the material-science or physics moat (no viable second source or drop-in substitute).
    3. Qualification Gate (20%): Measures the design-in/qualification cycle (e.g., $\ge$ 12 months).
    4. Discovery Gap (15%): Measures institutional ownership and analyst coverage (e.g., Institutional ownership < 40%, analyst coverage $\le$ 3).
    5. Demand Tension (15%): Measures demand growth vs. supply constraints (e.g., Demand CAGR $\ge$ 50%, capacity utilization > 85%).

    Scoring Thresholds

    • $\ge$ 80 $\rightarrow$ CORE: Highest-conviction chokepoint, full position.
    • 60–79 $ ightarrow$ BUILD: Strong candidate, scale in on confirmation.
    • 40–59 $\rightarrow$ STARTER: Early signal, small position, monitor closely.
    • < 40 $\rightarrow$ PASS: Does not meet chokepoint criteria.

  11. Understand the Rollercoaster Rate metric

    main

    The rollercoaster_rate is the core metric used by the alphagbm-take-profit skill. It quantifies the probability that an entry's paper profit reaches +50% and then falls more than 50% from that peak before the exit occurs.

    A high rollercoaster rate indicates a volatile journey where significant paper gains evaporate, suggesting that a tiered exit strategy is necessary rather than a simple long-term hold.

    Typical Rollercoaster Rates by Asset Class:

    • Broad-index ETFs (SPY, VTI), Blue chips (AAPL, MSFT), Sector ETFs (SOXX, XLK): ~0% (Hold forever)
    • Large-cap mega-caps (META, AMZN): ~47% (Tiered exit preferred)
    • High growth (NVDA, TSLA, AMD): ~85% (Tiered exit mandatory)
    • Crypto-related (COIN, MSTR): ~90% (Tiered exit mandatory)
    • Leveraged ETFs (TQQQ, SOXL): ~97% (Structurally un-holdable; tiered exit mandatory)

  12. Interpret IV Zones and Trading Signals

    main

    The skill classifies IV into zones to suggest trading actions based on whether options are expensive or cheap:

    IV RankZoneMeaningSuggested Action
    80-100Very HighNear annual peak; options expensiveSell premium (short strangles, iron condors, credit spreads)
    60-80HighElevated; above-average pricesLean toward selling; good for covered calls
    40-60ModerateMiddle rangeStrategy-neutral; use directional view
    20-40LowDepressed; options cheapLean toward buying (debit spreads, long straddles)
    0-20Very LowNear annual trough; options very cheapBuy premium (long straddles, debit spreads, calendars)