Testing Intuition-Based and Prediction Market-Based Forecasting: A Preregistered Study Using Financial Prediction Tasks

This preregistered study is a replication and extension of prior exploratory work comparing an Intuition Market (IM) with a traditional Prediction Market (PM) in forecasting binary financial outcomes. A previous exploratory study suggested that IM performance may equal or exceed that of PM under specific conditions, particularly when predictions are expressed with moderate-to-high confidence and when both methods converge on the same outcome. However, these findings require confirmation using independent data and preregistered analyses. The present study evaluates these hypotheses using a new set of 30 future financial events. Primary objectives are to test whether (i) IM predictions exceed chance accuracy when confidence is ≥60%, (ii) concordant IM–PM predictions yield higher-than-chance accuracy. All hypotheses, inclusion criteria, outcome measures, and analyses were specified a priori. Primary outcomes were predictive accuracy percentages. Observed accuracy rates were directionally consistent with the confirmatory hypotheses but did not reach the preregistered effect-size benchmarks. Specifically, IM predictions at confidence ≥60% yielded 58.8% accuracy (H1); concordant IM–PM predictions yielded 63.6% accuracy (H2); and concordant predictions at IM confidence ≥70% yielded 100.0% accuracy (H3). However, subgroup sample sizes were small (n = 17, 11, and 3, respectively) and confidence intervals were wide, precluding firm inferential conclusions. These findings while ensuring full transparency, reproducibility, and protection against analytical flexibility provide preliminary, directionally consistent support for intuition-based forecasting as a complement to standard prediction markets, but the limited event set and wide confidence intervals require cautious interpretation.