{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,20]],"date-time":"2026-10-20T05:38:23Z","timestamp":1792474703973,"version":"4.5.0"},"posted":{"date-parts":[[2026]]},"group-title":"SSRN","reference-count":0,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>We turn each 1-minute bar into a word built from four parts: its shape, its intensity, its range-to-volume balance and its relation to the previous bar. We then ask whether these words improve 15-minute volatility forecasts beyond the standard public intraday models: HAR, HAR with signed semivariances and jumps (HAR-RS-J), multiplicative component GARCH and Realized GARCH. We call the model with the words added HAR-W. The tests use QQQ, TSLA and US small caps on data the models never saw. They were registered in advance and judged against the best benchmark chosen after the fact, on two losses (MSE of log volatility and QLIKE). The words pass 10 of 12 tests. After volatility spikes they add 1.7 to 3.7 points of out-of-sample R\u00b2 and cut QLIKE by 4% to 9%. At every minute the gain is smaller but consistent: 0.5 to 1.3 points and 1.6% to 4.2%. In every model confidence set, a model with the words survives. On daily bars the words add nothing beyond HAR, so the effect is an intraday phenomenon. Against Kronos, a pre-trained foundation model for candles, the words still add to the best benchmark after spikes on QQQ and TSLA; on small caps they add to it but miss the registered R\u00b2 bar by a hair (4 of 6 tests pass). The spike bar and the three bars after it carry most of the gain (+0.74 R\u00b2 points on the hold-out year). The words do not predict direction.<\/jats:p>","DOI":"10.2139\/ssrn.7621579","type":"posted-content","created":{"date-parts":[[2026,10,20]],"date-time":"2026-10-20T04:40:56Z","timestamp":1792471256000},"source":"Crossref","is-referenced-by-count":0,"title":["Candlestick Words Improve Intraday Volatility Forecasts Beyond HAR and GARCH"],"prefix":"10.2139","author":[{"ORCID":"https:\/\/orcid.org\/0009-0005-0652-8513","authenticated-orcid":true,"given":"Jeremia","family":"Passmore","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,10,20]],"date-time":"2026-10-20T04:40:56Z","timestamp":1792471256000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ssrn.com\/abstract=7621579"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":0,"URL":"https:\/\/doi.org\/10.2139\/ssrn.7621579","relation":{},"subject":[],"published":{"date-parts":[[2026]]},"subtype":"preprint"}}