Private Beta
Pick any month.
See why.
A hybrid symbolic-neural macro inference system.
9 influencing factors → 5 macro axes → 3 asset groups (PE / Yields / Gold). LLM scoring + Theil-Goldberger weight learning + 25y FRED-calibrated backtest.
436mo
any month since 1990
+4.2σ
current Gold/M2
3
valuation axes
9
factor decomp
83/89
pair β (OOS)
Time Machine: input any month 1990-now, see P/E, 10Y yield, and Gold/M2 as σ-deviations from 10yr rolling norm, plus the 9-factor macro context explaining why the market thought what it did. Data sourced from Shiller + FRED + yfinance. 89 pair coefficients (83 fitted empirically via HAC Newey-West regression on Δ/Δln monthly series, 6 pending external data).
Model architecture
Block A — Inputs
Economy cycle
Fiscal support
Debt sustainability
QE
Globalization
US dominance
AI investments
AI productivity
Iran escalation
Gemini Flash · Google Search · k=3
Block B — Macro axes
GDP
Inflation
Margins
USDX
Liquidity
Correlation weights — Andrei Guseletov
25yr historical calibration
Block C — Assets
PE
GDP + Margins + Liquidity
Yields
Inflation
Gold
−USDX + Liquidity
direction: bullish / bearish / neutral
POST /api/v1/macro/forecastv1.2
curl -X POST https://ml.expanova.io/api/v1/macro/forecast \
-H "Authorization: Bearer motm-••••••••••••••••" \
-H "Content-Type: application/json" \
-d '{"period": "NOW"}'
{
"period": "NOW",
"factor_scores": { "Economy": -0.1, "QE": 0.4, ... },
"macro_totals": { "GDP": 0.0, "Inflation": 0.5, ... },
"assets": {
"PE": { "score": 0.75, "direction": "strongly bullish" },
"Yields": { "score": 0.50, "direction": "bullish" },
"Gold": { "score": 0.95, "direction": "strongly bullish" }
}
}