Volatility Skills

Volatility skills are computed analytics — they run trusted numpy code over archive closes to produce derived volatility metrics. They never execute agent-authored code, and every result includes the provenance envelope with source: "computed".

These skills read real archive closes — no agent-supplied price series. Pass a symbol and a window; the skill pulls the closes from the archive and computes the metric.

Skill summary

Skill ID	Tool name	Source	Runtime	Description
`vol.realized`	`vol_realized`	computed	interactive	Annualized realized volatility from archive closes
`vol.term_structure`	`vol_term_structure`	computed	interactive	IV term structure by expiry
`vol.iv_rank`	`vol_iv_rank`	computed	interactive	IV rank and percentile
`vol.vrp`	`vol_vrp`	computed	interactive	Variance risk premium (IV − RV)
`vol.anomaly_score`	`vol_anomaly_score`	computed	interactive	Z-score of current vol vs. history
`vol.character`	`vol_character`	computed	interactive	Vol regime classification

vol.realized

Annualized realized volatility from archive closes. Computed as the annualized standard deviation of log returns over the window. Arguments

Name	Type	Required	Default	Description
`symbol`	string	yes	—	Ticker symbol
`window`	int	no	`20`	Lookback in trading days
`as_of`	ISO date	no	latest	End date for the window
`annualization`	int	no	`252`	Trading days per year
`method`	string	no	`close_to_close`	`close_to_close`, `parkinson`, `garman_klass`, `yang_zhang`

What it computes: RV = std(log_returns[-window:]) * sqrt(annualization) Example

vol_realized({ "symbol": "SPY", "window": 20, "method": "close_to_close" })

{
  "skill_id": "vol.realized",
  "data": {
    "symbol": "SPY", "window": 20, "as_of": "2025-06-27",
    "realized_vol": 0.112, "realized_vol_pct": 11.2,
    "method": "close_to_close", "annualization": 252
  },
  "provenance": { "source": "computed", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 1, "tier": "free" }, "warnings": [] }
}

vol.term_structure

IV term structure — implied vol by expiry, derived from the archive options chain. Arguments

Name	Type	Required	Default	Description
`symbol`	string	yes	—	Ticker symbol
`as_of`	ISO date	no	latest	Snapshot date
`strike`	float	no	ATM	Strike (defaults to spot)
`moneyness`	float	no	`1.0`	K/S ratio if `strike` omitted
`max_expiries`	int	no	`12`	Max expiries to return

What it computes: pulls the archive options chain, fits IV per expiry (Black-Scholes), returns the curve. Example

vol_term_structure({ "symbol": "NVDA", "moneyness": 1.0 })

{
  "skill_id": "vol.term_structure",
  "data": {
    "symbol": "NVDA", "as_of": "2025-06-27", "spot": 128.42,
    "curve": [
      { "expiry": "2025-07-18", "dte": 21, "iv": 0.42 },
      { "expiry": "2025-08-15", "dte": 49, "iv": 0.39 },
      { "expiry": "2025-09-19", "dte": 84, "iv": 0.37 },
      { "expiry": "2025-12-19", "dte": 175, "iv": 0.35 }
    ],
    "slope": -0.0007
  },
  "provenance": { "source": "computed", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 2, "tier": "basic" }, "warnings": [] }
}

vol.iv_rank

IV rank and percentile — where current IV sits relative to its own trailing history. Arguments

Name	Type	Required	Default	Description
`symbol`	string	yes	—	Ticker symbol
`lookback_days`	int	no	`252`	History window
`as_of`	ISO date	no	latest	Snapshot date
`strike`	float	no	ATM	Strike

What it computes:

iv_rank = (current_iv - min_iv) / (max_iv - min_iv) * 100
iv_percentile = rank of current IV in the trailing distribution

Example

vol_iv_rank({ "symbol": "TSLA", "lookback_days": 252 })

{
  "skill_id": "vol.iv_rank",
  "data": {
    "symbol": "TSLA", "as_of": "2025-06-27",
    "current_iv": 0.58, "iv_rank": 72, "iv_percentile": 0.78,
    "lookback_min_iv": 0.31, "lookback_max_iv": 0.92, "lookback_days": 252
  },
  "provenance": { "source": "computed", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 2, "tier": "basic" }, "warnings": [] }
}

vol.vrp

Variance risk premium — the gap between implied and realized vol. Positive VRP means options are priced richer than realized. Arguments

Name	Type	Required	Default	Description
`symbol`	string	yes	—	Ticker symbol
`rv_window`	int	no	`21`	Realized vol lookback (days)
`iv_window`	int	no	`30`	Implied vol horizon (days)
`as_of`	ISO date	no	latest	Snapshot date

What it computes: VRP = IV^2 − RV^2 (in variance space), plus the ratio IV / RV. Example

vol_vrp({ "symbol": "SPY", "rv_window": 21, "iv_window": 30 })

{
  "skill_id": "vol.vrp",
  "data": {
    "symbol": "SPY", "as_of": "2025-06-27",
    "iv": 0.142, "rv": 0.112,
    "vrp_variance": 0.0078, "iv_rv_ratio": 1.27
  },
  "provenance": { "source": "computed", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 2, "tier": "basic" }, "warnings": [] }
}

vol.anomaly_score

Z-score of current realized vol against its trailing distribution — flags vol spikes. Arguments

Name	Type	Required	Default	Description
`symbol`	string	yes	—	Ticker symbol
`window`	int	no	`20`	Realized vol window
`lookback_days`	int	no	`252`	Baseline history
`as_of`	ISO date	no	latest	Snapshot date

What it computes: z = (current_rv - mean_rv) / std_rv over the trailing lookback_days. Example

vol_anomaly_score({ "symbol": "NVDA", "window": 20 })

{
  "skill_id": "vol.anomaly_score",
  "data": {
    "symbol": "NVDA", "as_of": "2025-06-27",
    "current_rv": 0.48, "baseline_mean_rv": 0.35, "baseline_std_rv": 0.08,
    "z_score": 1.625, "is_anomaly": false, "threshold": 2.0
  },
  "provenance": { "source": "computed", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 2, "tier": "basic" }, "warnings": [] }
}

vol.character

Vol regime classification — labels the current vol environment. Arguments

Name	Type	Required	Default	Description
`symbol`	string	yes	—	Ticker symbol
`lookback_days`	int	no	`252`	History window
`as_of`	ISO date	no	latest	Snapshot date

What it computes: classifies the regime as low, normal, elevated, high, or crisis based on realized vol percentile and trend; also returns term_structure (contango / backwardation) and trend (rising / falling / flat). Example

vol_character({ "symbol": "SPY" })

{
  "skill_id": "vol.character",
  "data": {
    "symbol": "SPY", "as_of": "2025-06-27",
    "regime": "low", "rv_percentile": 0.18,
    "term_structure": "contango", "trend": "falling",
    "iv_rank": 22
  },
  "provenance": { "source": "computed", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 2, "tier": "basic" }, "warnings": [] }
}

Next steps

Options Skills

Black-Scholes greeks and implied vol — pure computed from agent-supplied inputs.

Computed Analytics Skills

Correlations, factor exposures, DCF, portfolio optimization, Monte Carlo, and tearsheets.

​Skill summary

​vol.realized

​vol.term_structure

​vol.iv_rank

​vol.vrp

​vol.anomaly_score

​vol.character

​Next steps

Options Skills

Computed Analytics Skills

Skill summary

vol.realized

vol.term_structure

vol.iv_rank

vol.vrp

vol.anomaly_score

vol.character

Next steps