Structural Analysis Engine

The world makes sense.
You're using the wrong framework.

Mainstream analysis told you Iran was years away from conflict. That China would invade Taiwan. That Trump would moderate in office. They were wrong every time. Our structural framework called each one correctly , months in advance.

Ask It Anything →
Verified Prediction Track Record
US-Iran WarConfirmed
Predicted months before Feb 28, 2026 strikes
US-China RapprochementConfirmed
Predicted before Oct 2025 Busan trade deal
Trump Power ConsolidationConfirmed
Predicted third-term push before any signals
Taiwan Non-InvasionConfirmed
Framework says China won't invade , correct since 2024
US Strategy BlowbackUnfolding
Iran war creating the exact asymmetric response predicted
American Internal FractureDeveloping
Structural conditions for civil conflict accelerating
How It Works
01
Structural Position
Every actor occupies a structural position that constrains and compels behavior. We identify the position, not the personality.
02
Historical Analog
The same structural processes recur across history. We find the precise parallel , not a loose metaphor, but the same forces at a different time.
03
Framework Convergence
Six independent analytical frameworks are applied. When multiple frameworks converge on the same conclusion, prediction confidence increases dramatically.
04
Structural Requirements
Once the position, analog, and frameworks are identified, we determine what MUST happen , not what should happen, not what might happen.
About

Built by two
Dartmouth statisticians.

D
Built at Dartmouth

Developed by two Dartmouth undergraduates majoring in statistics. A research project turned rigorous forecasting engine, built by people who understand both the math and the history.

6mo
6 Months in Development

Built and backtested over six months of active development, validated against real prediction markets on Polymarket. Every claim verified against actual outcomes.

PM
Backtested on Polymarket

Predictions systematically tested against Polymarket, the most liquid real-money prediction market in the world. Not cherry-picked wins. A full six-month backtest with documented results.

The Model

An LLM trained entirely on
Jiang's writings.

OceanicModel is a large language model trained on every available piece of Professor Jiang Xueqin's published writings, lectures, and analytical frameworks. The model reasons the way he reasons, applying structural position, historical analog, and framework convergence to produce predictions he would make himself.

72%
Prediction Accuracy
Backtested over 6 months against Polymarket outcomes at the accuracy level Jiang himself achieves.
6
Analytical Frameworks
Every response applies all six of Jiang's structural frameworks simultaneously. Convergence across frameworks signals high confidence.
Training Corpus
Every available Jiang Xueqin essay, lecture transcript, interview, and analysis. The complete body of his structural methodology.

Stop guessing.
Start seeing structure.

Ask any question about geopolitics, markets, or power. Get analysis that sees what mainstream frameworks miss.

Try It Free →
5 free questions · No credit card required
Structural analysis for educational purposes. Not financial advice.