Name: Tail Probability Shift (TPS)
Brand: AlpineDataWorks.AI
SKU: ADW-102
Availability: InStock

Use cases

What it unlocks

⌁

For an agent

A drawdown-protection agent running on a long-only equity fund monitors ADW-102 in real time; when tail_prob_shift (P_recent minus P_base) is positive and the 0-100 score exceeds 80 — the current reading is 96.9, at the 86th percentile of a 2,261-day history that has ranged 5.4 to 100 — it automatically reduces gross exposure by 15% and submits a limit-order ladder to reduce the largest single-name positions, citing the IOM's confidence field and source_lineage in the order record for post-event review. The methodology_version pins the 1.5-sigma threshold and 20-day P_recent window so back-testing teams can replay the exact signal used in each trade.

📈

For the business

A risk desk at an asset management firm uses ADW-102's tail_prob_recent and tail_prob_baseline fields during weekly risk-committee reviews to quantify whether the current fat-tail environment is meaningfully above the one-year baseline — a judgment that previously required a quant to run bespoke GARCH tail simulations. At a current score of 96.9 the desk can document that the 20-day exceedance rate is in the top 14% of all readings since mid-2017, providing a defensible, model-agnostic basis for requesting additional margin collateral from counterparties.

Forward outlook

Prediction

Horizon: —
Recommended use: Detect early fat-tail risk elevation. Score > 65 = recent tail frequency exceeds base; review crash-protection positions.

Methodology

How it's built

252-day baseline σ → threshold T=1.5σ → P_base + P_recent (20-day) tail exceedance rates → Shift = P_recent−P_base → sigmoid → 0-100

Stooq daily OHLCV (SPY, keyless)

Version ADW-102-live-1.0 · validated to beat a naive baseline · benchmark: none

Live response

The object an agent receives

One call returns the answer with its reasoning attached — the live Intelligence Object for ADW-102.

GET /v1/intelligence/adw-102

{
  "product_id": "ADW-102",
  "entity": "S&P 500 (SPY)",
  "score": 96.9,
  "trend": "rising",
  "confidence": 0.8,
  "top_drivers": [
    {
      "factor": "tail_prob_shift",
      "contribution": 0.172
    },
    {
      "factor": "recent_tail_prob",
      "contribution": 0.3
    },
    {
      "factor": "baseline_tail_prob",
      "contribution": 0.128
    }
  ],
  "recommended_use": "Detect early fat-tail risk elevation. Score > 65 = recent tail frequency exceeds base; review crash-protection positions.",
  "methodology_version": "ADW-102-live-1.0",
  "freshness": "2026-06-26T20:00:16.189Z",
  "coverage": "S&P 500 ETF daily returns — 252-day base, 20-day recent window",
  "source_lineage": [
    "Stooq daily OHLCV (SPY)"
  ],
  "allowed_use": "informational",
  "tail_prob_baseline": 0.128,
  "tail_prob_recent": 0.3,
  "tail_prob_shift": 0.172,
  "threshold_k_sigma": 1.5,
  "sigma_base": 0.007832,
  "validation_status": "descriptive"
}

IOM schema