Computed Analytics Skills

Computed analytics skills run trusted numpy/pandas code over archive closes or agent-supplied inputs to produce derived metrics. They never execute agent-authored code. Every result includes the provenance envelope with source: "computed" or source: "archive-community" when the skill reads closes.

Skills that read closes pull them from the archive by symbol — you don’t pass a price series. Skills marked agent-supplied take inputs directly (e.g. cash flows for DCF).

Skill summary

Skill ID	Tool name	Source	Runtime	Description
`correlation.matrix`	`correlation_matrix`	archive-community	scan	Rolling correlation matrix for a ticker set
`factor.exposures`	`factor_exposures`	archive-community	interactive	Style factor loadings
`dcf.value`	`dcf_value`	computed	interactive	Discounted cash flow valuation
`portfolio.optimize`	`portfolio_optimize`	archive-community	optimization	Mean-variance / risk-parity optimization
`montecarlo.simulate`	`montecarlo_simulate`	archive-community	optimization	Path simulation for a portfolio or strategy
`scenario.project`	`scenario_project`	computed	interactive	Project returns under a named scenario
`risk.stress`	`risk_stress`	archive-community	backtest	Stress-test P&L under historical shocks
`tearsheet`	`tearsheet`	archive-community	backtest	Full performance tearsheet
`walkforward.run`	`walkforward_run`	archive-community	backtest	Walk-forward optimization over a strategy spec
`technical.indicators`	`technical_indicators`	archive-community	interactive	RSI, MACD, ATR, Bollinger, etc.
`seasonality.monthly`	`seasonality_monthly`	archive-community	interactive	Average monthly returns over N years
`seasonality.heatmap`	`seasonality_heatmap`	archive-community	backtest	Day/month return heatmap

correlation.matrix

Rolling correlation matrix for a set of tickers, computed from archive closes. Arguments

Name	Type	Required	Default	Description
`symbols`	string[]	yes	—	Tickers (2–50)
`window`	int	no	`60`	Lookback in trading days
`as_of`	ISO date	no	latest	End date
`method`	string	no	`pearson`	`pearson` or `spearman`

What it computes: pairwise correlation of log returns over the trailing window. Example

correlation_matrix({ "symbols": ["SPY","QQQ","IWM","TLT","GLD"], "window": 60 })

{
  "skill_id": "correlation.matrix",
  "data": {
    "as_of": "2025-06-27", "window": 60, "method": "pearson",
    "matrix": {
      "SPY": { "SPY": 1.00, "QQQ": 0.94, "IWM": 0.86, "TLT": -0.32, "GLD": 0.12 },
      "QQQ": { "SPY": 0.94, "QQQ": 1.00, "IWM": 0.78, "TLT": -0.28, "GLD": 0.10 }
    }
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 3, "tier": "basic" }, "warnings": [] }
}

factor.exposures

Style factor loadings — momentum, value, quality, size, low-vol, growth — for a ticker or portfolio. Arguments

Name	Type	Required	Default	Description
`symbol`	string	one of `symbol`/`weights`	—	Single ticker
`weights`	object	one of `symbol`/`weights`	—	Map of `{symbol: weight}` for a portfolio
`as_of`	ISO date	no	latest	Snapshot date
`lookback_days`	int	no	`252`	Factor model window
`factors`	string[]	no	core set	Factor names

What it computes: rolling regression of ticker returns against factor returns. Example

factor_exposures({ "symbol": "NVDA", "factors": ["momentum","value","quality","growth","size","low_vol"] })

{
  "skill_id": "factor.exposures",
  "data": {
    "symbol": "NVDA", "as_of": "2025-06-27",
    "exposures": { "momentum": 1.42, "growth": 1.18, "quality": 0.62, "value": -0.81, "size": 0.74, "low_vol": -0.92 },
    "r_squared": 0.68
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 3, "tier": "basic" }, "warnings": [] }
}

dcf.value

Discounted cash flow valuation. Agent-supplied cash flows. Arguments

Name	Type	Required	Default	Description
`cash_flows`	float[]	yes	—	Projected annual cash flows
`terminal_growth`	float	no	`0.02`	Perpetual growth rate
`discount_rate`	float	no	`0.10`	WACC
`shares_outstanding`	float	no	—	For per-share value
`net_debt`	float	no	`0`	Subtract from enterprise value

What it computes: NPV of cash flows + terminal value (Gordon growth), minus net debt, divided by shares. Example

dcf_value({ "cash_flows": [5000,6000,7200,8500,10000], "terminal_growth": 0.02, "discount_rate": 0.09, "shares_outstanding": 25000, "net_debt": 8000 })

{
  "skill_id": "dcf.value",
  "data": {
    "enterprise_value": 142000, "equity_value": 134000,
    "per_share_value": 5.36, "terminal_value_pct": 0.62
  },
  "provenance": { "source": "computed", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 1, "tier": "free" }, "warnings": ["cash_flows agent-supplied; not validated against filings"] }
}

portfolio.optimize

Mean-variance or risk-parity optimization for a set of tickers. Runs under the optimization profile (14,400 s). Arguments

Name	Type	Required	Default	Description
`symbols`	string[]	yes	—	Tickers (2–30)
`objective`	string	no	`max_sharpe`	`max_sharpe`, `min_variance`, `risk_parity`, `max_return`
`risk_free_rate`	float	no	`0.045`	For Sharpe
`lookback_days`	int	no	`252`	Covariance window
`constraints`	object	no	—	`{min_weight, max_weight, long_only, sector_caps}`
`as_of`	ISO date	no	latest	Snapshot date

What it computes: solves the selected objective over the covariance matrix from archive closes. Example

portfolio_optimize({ "symbols": ["SPY","QQQ","TLT","GLD","VNQ"], "objective": "max_sharpe", "lookback_days": 252 })

{
  "skill_id": "portfolio.optimize",
  "data": {
    "objective": "max_sharpe", "as_of": "2025-06-27",
    "weights": { "SPY": 0.32, "QQQ": 0.28, "TLT": 0.18, "GLD": 0.12, "VNQ": 0.10 },
    "expected_return": 0.112, "expected_vol": 0.121, "expected_sharpe": 0.55
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 8, "tier": "pro" }, "warnings": [] }
}

montecarlo.simulate

Path simulation for a portfolio or strategy spec. Runs under the optimization profile. Arguments

Name	Type	Required	Default	Description
`symbols`	string[]	yes	—	Tickers
`weights`	float[]	yes	—	Portfolio weights
`horizon_days`	int	no	`252`	Simulation horizon
`n_paths`	int	no	`10000`	Number of paths
`method`	string	no	`historical`	`historical` or `parametric`
`as_of`	ISO date	no	latest	Start date

What it computes: bootstraps or parametrically simulates portfolio return paths; returns percentile bands and terminal distribution. Example

montecarlo_simulate({ "symbols": ["SPY","TLT"], "weights": [0.6,0.4], "horizon_days": 252, "n_paths": 10000 })

{
  "skill_id": "montecarlo.simulate",
  "data": {
    "horizon_days": 252, "n_paths": 10000,
    "terminal_return": { "p5": -0.18, "p25": 0.02, "p50": 0.09, "p75": 0.17, "p95": 0.31 },
    "max_drawdown": { "p50": -0.08, "p95": -0.21 },
    "prob_loss": 0.22
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 10, "tier": "pro" }, "warnings": [] }
}

scenario.project

Project returns under a named scenario (e.g. rate_hike_2022, covid_crash_2020). Agent-supplied shocks applied to a portfolio. Arguments

Name	Type	Required	Default	Description
`scenario`	string	yes	—	Named scenario or `custom`
`shocks`	object	no	—	For `custom`: `{symbol: shock_pct}`
`portfolio`	object	yes	—	`{symbol: weight}`

What it computes: applies scenario shocks to the portfolio and returns projected P&L. Example

scenario_project({ "scenario": "rate_hike_2022", "portfolio": { "SPY": 0.6, "TLT": 0.4 } })

{
  "skill_id": "scenario.project",
  "data": {
    "scenario": "rate_hike_2022",
    "portfolio_pnl_pct": -0.082,
    "by_symbol": { "SPY": -0.064, "TLT": -0.109 }
  },
  "provenance": { "source": "computed", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 2, "tier": "basic" }, "warnings": [] }
}

risk.stress

Stress-test P&L under historical shocks (2008 GFC, 2020 COVID, 2022 rate hike, etc.). Reads archive closes. Arguments

Name	Type	Required	Default	Description
`portfolio`	object	yes	—	`{symbol: weight}`
`scenarios`	string[]	no	all	Named stress scenarios
`as_of`	ISO date	no	latest	Snapshot date

What it computes: replays each historical shock window over the portfolio’s current holdings. Example

risk_stress({ "portfolio": { "SPY": 0.7, "QQQ": 0.3 }, "scenarios": ["gfc_2008","covid_2020"] })

{
  "skill_id": "risk.stress",
  "data": {
    "results": [
      { "scenario": "gfc_2008", "window": "2008-09-15 to 2009-03-09", "pnl_pct": -0.42, "max_drawdown": -0.48 },
      { "scenario": "covid_2020", "window": "2020-02-19 to 2020-03-23", "pnl_pct": -0.31, "max_drawdown": -0.34 }
    ]
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 5, "tier": "basic" }, "warnings": [] }
}

tearsheet

Full performance tearsheet — returns, drawdowns, risk ratios — for a ticker or portfolio. Runs under the backtest profile. Arguments

Name	Type	Required	Default	Description
`symbol`	string	one of `symbol`/`weights`	—	Single ticker
`weights`	object	one of `symbol`/`weights`	—	Portfolio weights
`benchmark`	string	no	`SPY`	Benchmark symbol
`from`	ISO date	no	`now - 5y`	Start date
`to`	ISO date	no	latest	End date
`risk_free_rate`	float	no	`0.045`	For Sharpe/Sortino

What it computes: CAGR, vol, Sharpe, Sortino, max drawdown, Calmar, win rate, monthly returns table, drawdown series. Example

tearsheet({ "symbol": "QQQ", "benchmark": "SPY", "from": "2020-01-01" })

{
  "skill_id": "tearsheet",
  "data": {
    "symbol": "QQQ", "benchmark": "SPY", "from": "2020-01-01", "to": "2025-06-27",
    "cagr": 0.182, "vol": 0.221, "sharpe": 0.62, "sortino": 0.85,
    "max_drawdown": -0.32, "calmar": 0.57, "win_rate": 0.55,
    "alpha_vs_benchmark": 0.041, "beta_vs_benchmark": 1.12
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 5, "tier": "basic" }, "warnings": [] }
}

walkforward.run

Walk-forward optimization over a strategy spec. Runs under the backtest profile. Arguments

Name	Type	Required	Default	Description
`strategy_id`	string	yes	—	Strategy spec ID (from `propose_strategy`)
`train_window`	int	no	`252`	Training window (days)
`test_window`	int	no	`63`	Out-of-sample test window
`step`	int	no	`63`	Rebalance step
`param_grid`	object	no	—	Parameter grid to search

What it computes: rolls train/test windows, optimizes on train, evaluates on test, aggregates OOS performance. Example

walkforward_run({ "strategy_id": "strat_abc123", "train_window": 252, "test_window": 63 })

{
  "skill_id": "walkforward.run",
  "data": {
    "strategy_id": "strat_abc123",
    "n_folds": 12,
    "oos_cagr": 0.094, "oos_sharpe": 0.51, "oos_max_drawdown": -0.18,
    "is_oos_decay": 0.32
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 12, "tier": "pro" }, "warnings": [] }
}

technical.indicators

Standard technical indicators from archive closes — RSI, MACD, ATR, Bollinger, SMA/EMA, ADX. Arguments

Name	Type	Required	Default	Description
`symbol`	string	yes	—	Ticker symbol
`indicators`	string[]	no	core set	`rsi`, `macd`, `atr`, `bollinger`, `sma`, `ema`, `adx`
`periods`	object	no	defaults	Map of indicator to period(s)
`from`	ISO date	no	`now - 1y`	Start date
`to`	ISO date	no	latest	End date

What it computes: each indicator over the archive close series. Example

technical_indicators({ "symbol": "AAPL", "indicators": ["rsi","macd","bollinger"], "periods": { "rsi": 14 } })

{
  "skill_id": "technical.indicators",
  "data": {
    "symbol": "AAPL", "as_of": "2025-06-27",
    "indicators": {
      "rsi_14": 62.4,
      "macd": { "macd": 1.82, "signal": 1.41, "histogram": 0.41 },
      "bollinger": { "upper": 218.20, "middle": 212.55, "lower": 206.90, "width": 0.053 }
    }
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 2, "tier": "free" }, "warnings": [] }
}

seasonality.monthly

Average monthly returns over N years — detects recurring seasonal patterns. Arguments

Name	Type	Required	Default	Description
`symbol`	string	yes	—	Ticker symbol
`years`	int	no	`10`	Lookback years
`as_of`	ISO date	no	latest	End date
`statistic`	string	no	`mean`	`mean`, `median`, or `both`

What it computes: groups archive returns by month, averages per month over the lookback. Example

seasonality_monthly({ "symbol": "SPY", "years": 20 })

{
  "skill_id": "seasonality.monthly",
  "data": {
    "symbol": "SPY", "years": 20,
    "monthly": {
      "Jan": 0.012, "Feb": 0.004, "Mar": 0.008, "Apr": 0.015, "May": 0.001,
      "Jun": -0.004, "Jul": 0.011, "Aug": -0.006, "Sep": -0.009,
      "Oct": 0.008, "Nov": 0.018, "Dec": 0.012
    },
    "best_month": "Nov", "worst_month": "Sep"
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 2, "tier": "free" }, "warnings": [] }
}

seasonality.heatmap

Day-of-month × month return heatmap — finer-grained seasonality. Runs under the backtest profile. Arguments

Name	Type	Required	Default	Description
`symbol`	string	yes	—	Ticker symbol
`years`	int	no	`10`	Lookback years
`as_of`	ISO date	no	latest	End date
`resolution`	string	no	`daily`	`daily` or `weekly`

What it computes: average return for each (month, day) bucket over the lookback. Example

seasonality_heatmap({ "symbol": "SPY", "years": 10 })

{
  "skill_id": "seasonality.heatmap",
  "data": {
    "symbol": "SPY", "years": 10, "resolution": "daily",
    "buckets": [
      { "month": "Jan", "day": 2, "mean_return": 0.0042, "n_obs": 10, "hit_rate": 0.80 }
    ],
    "best_bucket": { "month": "Dec", "day": 31, "mean_return": 0.0061 },
    "worst_bucket": { "month": "Sep", "day": 15, "mean_return": -0.0048 }
  },
  "provenance": { "source": "archive-community", "freshness": { "as_of": "2025-06-27T20:00:00Z", "lag_seconds": 0 }, "cost": { "credits": 4, "tier": "basic" }, "warnings": [] }
}

Next steps

Financials & Earnings Skills

Income statements, balance sheets, cash flow, transcripts, earnings calendars, dividends, splits.

Strategy & Backtest

Turn your research into a declarative StrategySpec and run a deterministic backtest.

​Skill summary

​correlation.matrix

​factor.exposures

​dcf.value

​portfolio.optimize

​montecarlo.simulate

​scenario.project

​risk.stress

​tearsheet

​walkforward.run

​technical.indicators

​seasonality.monthly

​seasonality.heatmap

​Next steps

Financials & Earnings Skills

Strategy & Backtest

Skill summary

correlation.matrix

factor.exposures

dcf.value

portfolio.optimize

montecarlo.simulate

scenario.project

risk.stress

tearsheet

walkforward.run

technical.indicators

seasonality.monthly

seasonality.heatmap

Next steps