A displayed probability is the result of activity under a particular market structure. It can summarize information and disagreement, but the number alone cannot reveal how many independent judgments contributed to it or whether participants had strong incentives to examine the evidence. Treating it as an infallible consensus would ask it to do more than the data establish.
Begin with the observation
A last trade, a bid and an ask are different measures. A price also has a timestamp. Before attributing meaning to a change from 40% to 60%, confirm that the observations concern the same contract and comparable price measures. If one observation is a stale last trade and the other is a current offer, the comparison needs qualification.
Judge a sequence, not a winning example
Forecast quality is difficult to assess from one successful prediction. A useful record includes all forecasts in the chosen sample, captured before their outcomes were known. Keep unresolved and unsuccessful forecasts as well as the successes. Selecting only memorable winners creates a misleading impression of reliability.
Calibration asks whether groups of forecasts resolve at approximately the frequencies their probabilities imply. That requires an adequate sample and attention to dependence among events. Ten forecasts about closely related outcomes are not necessarily ten independent tests of a method.
Separate forecast quality from profitability
A reasonably calibrated market can still be expensive to trade when spreads and fees are included. Conversely, a profitable episode does not establish that a method is consistently accurate. Evaluating probabilities and evaluating a trading strategy are different research tasks with different inputs.
Our editorial standard is to document the question, price measure, time and resolution before drawing conclusions. That record supports a more useful discussion than treating every movement as new wisdom or every surprise as proof that the market failed.
Sources & context
CME explanation of binary event contracts ↗Factual references checked October 8, 2026. Interpretive discussion is editorial analysis. Examples are hypothetical unless explicitly identified otherwise.