PolyHedge

Every stock portfolio is an implicit bet on macro outcomes. PolyHedge makes those bets explicit — mapping your holdings to live Polymarket prediction markets and sizing hedges where your TradFi positions disagree with the crowd.

How it works

1. Factor decomposition Each holding is regressed against 18 economic factors (equity style, rates, inflation, credit, commodities, yield curve) via OLS. This produces a beta — sensitivity to each factor — per holding.

2. Portfolio factor profile Per-holding betas are dollar-weighted into a single portfolio factor profile. The output is a dollar exposure: how many dollars you gain or lose per 1-sigma move in each factor.

3. Polymarket mapping Each factor maps to Polymarket categories through a static keyword taxonomy. For each matched market, an implied probability is derived from how your TradFi holdings have moved since a baseline date:

p_implied = tanh(sigma_move × 0.5) × 0.48 + 0.50

4. Signal classification Gaps between the TradFi-implied probability and the current Polymarket price are classified:

Gap	Label
< 8pp	Noise — no action
8–15pp	Hedge signal
> 15pp	Strong signal

5. Position sizing Quarter-Kelly criterion with a 3% single-position cap and 10% total cap. Positions on correlated factors are reduced to avoid double-counting the same exposure.

Stack

Layer	Tools
Market data	yfinance, fredapi
Factor regression	statsmodels OLS
Prediction markets	Polymarket REST API
Numerics	numpy, pandas, scipy, scikit-learn

Quick start

pip install -r requirements.txt
python3 -m polyhedge.pipeline --portfolio portfolio_example.json

Set FRED_API_KEY in your environment (free at fred.stlouisfed.org).

Output

{
  "total_implicit_pm_exposure": 142000,
  "largest_single_factor_bet": "Will the Fed cut rates before July?",
  "net_directional_lean": "long YES",
  "markets": [
    {
      "question": "Will the Fed cut rates before July?",
      "implicit_direction": "long YES",
      "gross_notional": 54000,
      "pm_price": 0.42,
      "tradfi_implied": 0.61,
      "divergence": 0.19,
      "hedge": {
        "effective_edge": 0.14,
        "insurance_cost_pct": 2.1
      }
    }
  ]
}

Key fields

Field	Meaning
implicit_direction	Whether your portfolio benefits if the market resolves YES
divergence	tradfi_implied − pm_price — positive means TradFi thinks YES is more likely than Polymarket does
effective_edge	Confidence-discounted divergence used for sizing
insurance_cost_pct	Hedge position cost as % of the factor exposure being covered
correlated_factors	Other factors measuring the same risk — positions are reduced accordingly