Data engineer building side hustles in public, with real metrics.
Prediction Almanac is an open directory of every prediction market worth trading, the data you can backtest on, and the tools people have actually shipped. It rebuilds itself from scratch every day and fetches its own numbers - volume, repo health, blog freshness - so nothing on the page is a figure I typed once and left to go stale.
A quieter update. The retired sports bot's Rust replacement made $650 last month, why I won't open-source the one that actually works, what two months of blogging in the open returned, and the weather market-making model coming next.
The market-making bot that made me about $5k on Polymarket is now on GitHub, MIT-licensed, wallet still public. Here's what's in it, what I stripped out, and why giving away a bot that made money costs me nothing - the code was never the edge.
Last post I left a question hanging - the directional bets lost $3,184 when every one was supposed to have at least 7% edge. Analytics scripts I built gave me a clear picture but there is not a single simple answer that explains why my ROI decayed. The main reason was that I was using stale odds (up to 30 minutes fresh) which resulted in adverse selection fills a lot more often than I thought…
A free, open dataset: nearly 89,000 of Polymarket's 5-minute crypto up/down markets (BTC, ETH, SOL, XRP, DOGE, HYPE, BNB), captured second by second - about 26.8 million top-of-book order-book observations from March to May 2026. I collected it to backtest a bot. You can have it.
In early 2026 my bot netted ~$5k on Polymarket. The guaranteed arbitrage made +$8.3k - but capturing it forced directional bets that lost $3.2k. The whole story, with every number.
Why kacho.io exists: a logbook of the side hustles I'm actually building, with the real numbers behind them - the losing months included, and a public wallet so you can check.