Hodookim

I build data tools that answer one question well.

Most of what I ship wraps a public API and returns something the source does not: a value that only appears when two datasets are matched, or a contradiction that is wrong no matter what anyone believes about the market.

What I care about in a tool

• 🔍 Real prices, not midpoints. A midpoint always sums to par and shows arbitrage that vanishes when you try to fill it. I quote the order book.
• 🧮 Fees modelled, not buffered. Subtracting a flat cushion is too small in the middle of the range and too large in the tails.
• 🚫 Incomplete inputs disqualified, not scored. A partial outcome set guarantees nothing and looks best when it is most broken.
• 📉 Honest empty results. Most scans find nothing, and the run says so. A scanner that always finds an opportunity is measuring the wrong number.

Actors

• Prediction Market Mispricing Scanner  — implied probability distributions, monotonicity violations, and complete-set costs on Polymarket.

Ideas, bugs, or a dataset you want joined to another one — open an issue on the Actor and I will read it.