Skip to content

causal-tree.md

Latest commit

 

History

History
82 lines (57 loc) · 3.26 KB

causal-tree.md

File metadata and controls

82 lines (57 loc) · 3.26 KB
Raw

Causal Tree

A causal tree is a structured probabilistic model that breaks down a complex event into a hierarchy of independent, verifiable sub-conditions. Each node has a probability, and the tree computes the overall probability of the root event.

Explanation

Causal Trees: SimpleFunctions' Core Analytical Framework

A causal tree is how SimpleFunctions thinks about prediction markets. Instead of asking "what's the probability of X?" directly, we decompose the question into smaller, more answerable components.

Structure

A causal tree has:

  • Root node: The ultimate question (e.g., "Will Iran-Israel conflict escalate to direct confrontation?")
  • Branch nodes: Conditions that contribute to the root outcome
  • Leaf nodes: Observable, verifiable facts that can be tracked

Example: Iran Conflict Thesis

Root: Iran-Israel direct confrontation (p = 35%)
├── n1: Iran retaliates beyond proxies (p = 45%)
│   ├── n1.1: Missile/drone strike on Israeli territory (p = 60%)
│   └── n1.2: Naval confrontation in Strait of Hormuz (p = 30%)
├── n2: US military involvement (p = 25%)
│   ├── n2.1: US bases in region attacked (p = 40%)
│   └── n2.2: Congressional authorization (p = 15%)
└── n3: Diplomatic channels fail (p = 55%)
    ├── n3.1: UN Security Council deadlocked (p = 70%)
    └── n3.2: Back-channel negotiations collapse (p = 45%)

Why Trees Beat Single Estimates

  1. Decomposition makes estimation easier: Estimating "probability of WWIII" is impossible. Estimating "probability of UN Security Council deadlock on Iran resolution" is tractable.
  2. Information maps to nodes: A headline about Iranian missile tests maps directly to node n1.1. Without the tree, you'd have to reason about how it affects the overall probability — with the tree, you update one node and the math propagates.
  3. Confidence tracking: Each node has both a probability and a confidence level. If n3.2 has low confidence, you know where to focus research.

The Agent Uses Trees for Everything

When the heartbeat service runs, it:

  1. Scans for new information (news, price changes)
  2. Maps each piece to a causal node
  3. Updates node probabilities
  4. Recomputes the root probability
  5. Compares against market prices to find edge
  6. Triggers strategy actions if conditions are met

Example

sf thesis view --tree

Thesis: "US enters recession by end of 2026" Root probability: 42%

Causal tree: n1: Labor market deterioration p=55% conf=72% n1.1: Unemployment > 5% p=35% conf=68% n1.2: Non-farm payrolls < 100K/mo p=40% conf=65% n2: Consumer spending decline p=45% conf=70% n2.1: Real retail sales negative p=38% conf=75% n2.2: Consumer confidence < 80 p=52% conf=60% n3: Fed policy error p=30% conf=55% n3.1: Overtightening (rates > 5.5%) p=20% conf=50% n3.2: Too late to cut p=40% conf=58%

Root = weighted combination of branches Edge on KXRECSSNBER-26: 42% thesis - 28% market = +14pt

CLI

sf thesis view --tree

Related

thesis, node-probability, confidence, thesis-implied-price, edge-detection