← Posts

Thread · 3 posts

World Modeling in Markets

Applying anomaly-detection ideas to market data. The posts move from defining the evaluation to repairing price-based backtests, then to alternative-data pipelines and early prediction-market experiments, with costs and benchmarks included throughout.

  1. Started
  2. Last updated

Posts follow their reading order. Each shows its publication date and latest revision.

Reading path

The double ring marks the latest post.

  1. Post 01

    Why Markets

    Using public market data to test anomaly detection, with transaction costs and benchmark returns as part of the evaluation.

    Published
    Updated

    Implement and audit the backtest

    Public data and evaluation criteria become an implemented backtest. Rebalancing, universe selection, and signal timing are checked before interpreting strategy returns.

  2. Post 02

    Twenty Strategies, No Alpha

    Testing about 20 strategy variants across 20 years of ETF and equity data, correcting the backtest, and comparing the surviving momentum and trend strategies across three periods.

    Published
    Updated

    Try new data and contracts

    The price-based tests found no significant alpha. The search extends to filings, positioning, macro data, and Kalshi contracts, with publication timing, leakage, and fees part of the evaluation.

  3. Post 03Latest post

    Alternative Data and Prediction Markets

    SEC filing sentiment, CFTC positioning, and early experiments with Kalshi data, with attention to publication times, evaluation leakage, and trading costs.

    Published
    Updated