Historical test
Builder and Discover replay recorded markets. A rule is evaluated from information available at each simulated decision point. Dates and data gaps matter; missing evidence is not a hidden winning trade.
Methodology
A result is only as honest as its assumptions.
PaperMantis helps investigate prediction-market ideas with recorded data and virtual money. This page puts the price, timing and execution boundaries next to the numbers.
What the line means
Three boundaries to keep visible
Price and execution
Historical results use observed quotes under stated assumptions. Standard Discover results are gross and based on the quoted ask; fees, latency penalties, extra slippage and default order-book depth or VWAP effects are not modeled. A quote does not guarantee a real fill.
Later virtual observation
A compatible frozen rule can follow new markets in Portfolio with virtual money. Forward paper entry checks the recorded order book and may skip a signal when that book is stale or insufficient. Its later paper trades remain separate from historical backtests. Neither record is real-money performance.
Use the evidence
Questions to ask of any result
Which markets? Record asset, duration, UTC dates and any gaps.
Which price? Distinguish displayed odds, quoted ask and what a real order might fill.
Which period? Keep discovery, later validation and new virtual observation labeled separately.
Which conclusion? A historical line answers what the simulation produced under its inputs. It cannot promise future returns.