It took four versions to get to PaperMantisPaperMantis is roughly my fourth attempt to answer one question: can I test a prediction-market idea without risking real money or building a trading operation?Blog
Notes from the sandbox
Field notes on building PaperMantis, testing prediction market ideas, and learning what the odds leave out.
It took four versions to get to PaperMantisPaperMantis is roughly my fourth attempt to answer one question: can I test a prediction-market idea without risking real money or building a trading operation?
A 60% market is not offering you a 40% profitA 60% prediction-market headline can hide a 64¢ ask, a thin order book and a very ordinary amount of risk.
The spread is a toll booth, even when nothing happensA prediction-market bid–ask spread can cost money even when the odds do not move. The bid, the ask and the fill tell that small story.
The title is not the contractA prediction-market title invites a quick answer. The settlement rule decides which answer gets paid.
The backtest ends before the experiment doesA prediction-market backtest can be the beginning of an experiment. It becomes much less trustworthy when I let later paper results rewrite its past.
The election market that looked like a classroom experimentOne influential modern prediction market began as a university question, long before event odds became a shareable internet card.