This　paper　studies　the　identification　of　financial　fraud　behaviors　of　listed　companies　by　innovatively　considering　investors＇　heterogeneous　beliefs.　Firstly,　the　relationship　between　them　is　investigated　through　a　logistic　regression　model　and　the　results　show　that　investors＇　heterogeneous　beliefs　have　a　significantly　positive　correlation　with　financial　fraud.　Moreover,　after　considering　the　indicators　of　investors＇　heterogeneous　beliefs,　the　financial　fraud　identification　accuracies　through　six　machine　learning　models　have　been　improved,　implying　that　the　consideration　of　investors＇　heterogeneous　beliefs　is　meaningful　for　fraud　identification.　This　gives　hints　that　the　regulators　and　investors　can　take　utilization　of　investors＇　heterogeneous　beliefs　when　detecting　financial　fraud　behaviors.　©　2022　The　Authors.　Published　by　Elsevier　B.V.