---
title: Candlestick Words Improve Intraday Volatility Forecasts Beyond HAR and GARCH
url: https://www.ml-quant.com/papers/ssrn/7621579/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-10-23
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: SSRN 7621579
source_url: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7621579
featured: 2026-10-23
citations: unknown
topic: Derivatives & Volatility
---


# Candlestick Words Improve Intraday Volatility Forecasts Beyond HAR and GARCH

Encoding intraday bars as descriptive words adds 1.7 to 3.7 points of out-of-sample R² to standard volatility models after spikes but shows no edge on daily bars.

- Source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7621579
- Identifier: SSRN 7621579
- Released: 2026-10-20
- First featured: Quant Letter No. 136 (2026-10-23): https://www.ml-quant.com/issues/2026-10-23/
- Citations (Semantic Scholar): not tracked
- Published in: not yet
- Topic: Derivatives & Volatility
- Authors: Jeremia Passmore

## Related

- [Unified GARCH-Recurrent Neural Network in Financial Volatility Forecasting](https://www.ml-quant.com/papers/arxiv/2504.09380/): The paper proposes a new GARCH-GRU model for financial volatility forecasting, showing better computational efficiency and forecasting accuracy than other models.
- [Combining Deep Learning and GARCH Models for Financial Volatility and Risk Forecasting](https://www.ml-quant.com/papers/arxiv/2310.01063/): The research introduces a hybrid method for predicting the volatility and risk of financial tools by merging GARCH time series models with deep learning neural networks, finding that while this approach improves volatility predictions, it doesn't necessarily enhance Value-at-Risk and Expected Shortfall forecasts.
- [Improving Realised Volatility Forecast for Emerging Markets](https://www.ml-quant.com/papers/ssrn/4584573/): A study comparing four models for forecasting volatility in emerging markets found the HAR model best for long-term volatility and realised GARCH models for volatility clustering and persistence.
- [Adaptive LASSO-MGARCH for Multivariate Volatility Forecasting](https://www.ml-quant.com/papers/repec/cdf-wpaper-2026-4/): Introducing coefficient-specific penalization into multivariate GARCH equations reduces complexity and improves out-of-sample covariance forecasts across bonds, equities, and commodities.
- [EPUs Impact](https://www.ml-quant.com/papers/ssrn/4959148/): The Tree-based GARCH-MIDAS model reveals that high economic policy uncertainty weakens the response of asset volatility to macroeconomic variables, improving volatility predictions over longer periods.
- [SP 500 Volatility Forecasting](https://www.ml-quant.com/papers/ssrn/4903194/): The research investigates four techniques to enhance the precision of volatility forecasts for the SP 500, including the GARCH model, an LSTM network, a hybrid LSTM-GARCH model, and an advanced hybrid model incorporating the VIX index.