Prediction Markets

Can a bot systematically beat prediction-market crowds? This is a running experiment, not a pitch. Below is the live cumulative result on sample data — the shape of the curve is the point, not the size of the number.

This strategy is about edge, not scale — it is capacity-constrained, not capital-constrained, so capacity is limited. Headline is absolute cumulative PnL shown with its denominator (resolved markets + window), never ROI. v1 shows sample data.

Calibration — taking side only

Validates directional skill (did "30%" resolve ~30%?) independent of capital. It says nothing about market-making edge, which has its own metrics. Placeholder — illustrative only.

Making vs. taking attribution and the live daily refresh are deferred (see PRPs 00/07). Placeholder content for v1.