With　the　economic　development　of　countries　in　recent　years,　the　influence　of　the　stock　market　on　the　global　economy　has　increased.　The　economic　trends　of　stocks　are　influenced　by　many　factors,　such　as　information　on　the　financial　results　of　companies,　news　reports,　social　media　forums,　and　so　on.　Stock　predictions　have　also　become　a　popular　topic.　To　date,　there　has　been　a　great　deal　of　research　done　by　professionals,　most　of　which　has　used　methods　related　to　deep　learning.　Computer　deep　learning　methods　such　as　Generative　Adversarial　Networks　(GAN)　[1],　Convolutional　Neural　Networks　(CNN),　Deep　Reinforcement　Learning　(DRL),　and　LSTM　(a　recurrent　neural　network)　[2]　are　all　highly　efficient,　adaptable,　and　portable.　However,　while　many　researchers　have　addressed　many　of　the　problems　of　stock　prediction　through　a　variety　of　deep　learning　models,　such　as　the　barriers　to　predicting　stocks　based　on　the　characteristics　of　the　raw　data,　ignoring　the　correlation　between　similar　stocks　across　the　stock　market　[3],　and　the　inefficiency　of　information.　However,　a　detailed　summary　and　consideration　of　these　areas　are　still　lacking　at　the　moment.　This　paper　examines　and　summarizes　stock　prediction　algorithms　based　on　deep　learning　methods　and　their　development　process　and　in　the　latter　part　of　the　paper　analyses　the　future　trends　of　stock　prediction　algorithms.　©　2022　IEEE.