Learning to Predict Short-Term Volatility with Order Flow Image Representation

Introduction: The paper addresses the challenging problem of predicting the short-term realized volatility of the Bitcoin price using order flow information. The inherent stochastic nature and anti-persistence of price pose difficulties in accurate prediction. Methods: To address this, we propose a...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lensky, Artem, Hao, Mingyu
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Lensky, Artem
Hao, Mingyu
description Introduction: The paper addresses the challenging problem of predicting the short-term realized volatility of the Bitcoin price using order flow information. The inherent stochastic nature and anti-persistence of price pose difficulties in accurate prediction. Methods: To address this, we propose a method that transforms order flow data over a fixed time interval (snapshots) into images. The order flow includes trade sizes, trade directions, and limit order book, and is mapped into image colour channels. These images are then used to train both a simple 3-layer Convolutional Neural Network (CNN) and more advanced ResNet-18 and ConvMixer, with additionally supplementing them with hand-crafted features. The models are evaluated against classical GARCH, Multilayer Perceptron trained on raw data, and a naive guess method that considers current volatility as a prediction. Results: The experiments are conducted using price data from January 2021 and evaluate model performance in terms of root mean square error (RMSPE). The results show that our order flow representation with a CNN as a predictive model achieves the best performance, with an RMSPE of 0.85+/-1.1 for the model with aggregated features and 1.0+/-1.4 for the model without feature supplementation. ConvMixer with feature supplementation follows closely. In comparison, the RMSPE for the naive guess method was 1.4+/-3.0.
doi_str_mv 10.48550/arxiv.2304.02472
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2304_02472</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2304_02472</sourcerecordid><originalsourceid>FETCH-arxiv_primary_2304_024723</originalsourceid><addsrcrecordid>eNqFjrsKwkAQAK-xEPUDrNwfSDzzQHsxKIiKhrThMGuycLkLm8OYv1eDvdU0AzNCzFfSjzZxLJeKX_T0g1BGvgyidTAWpyMqNmRKcBYujAXdHdwqy85LkWvIrFaONLkeOnIVnLlAhkTbDg61KhGu2DC2aNxHs2YqRg-lW5z9OBGLZJdu995QzhumWnGffw_y4SD8b7wBaig71A</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Learning to Predict Short-Term Volatility with Order Flow Image Representation</title><source>arXiv.org</source><creator>Lensky, Artem ; Hao, Mingyu</creator><creatorcontrib>Lensky, Artem ; Hao, Mingyu</creatorcontrib><description>Introduction: The paper addresses the challenging problem of predicting the short-term realized volatility of the Bitcoin price using order flow information. The inherent stochastic nature and anti-persistence of price pose difficulties in accurate prediction. Methods: To address this, we propose a method that transforms order flow data over a fixed time interval (snapshots) into images. The order flow includes trade sizes, trade directions, and limit order book, and is mapped into image colour channels. These images are then used to train both a simple 3-layer Convolutional Neural Network (CNN) and more advanced ResNet-18 and ConvMixer, with additionally supplementing them with hand-crafted features. The models are evaluated against classical GARCH, Multilayer Perceptron trained on raw data, and a naive guess method that considers current volatility as a prediction. Results: The experiments are conducted using price data from January 2021 and evaluate model performance in terms of root mean square error (RMSPE). The results show that our order flow representation with a CNN as a predictive model achieves the best performance, with an RMSPE of 0.85+/-1.1 for the model with aggregated features and 1.0+/-1.4 for the model without feature supplementation. ConvMixer with feature supplementation follows closely. In comparison, the RMSPE for the naive guess method was 1.4+/-3.0.</description><identifier>DOI: 10.48550/arxiv.2304.02472</identifier><language>eng</language><subject>Computer Science - Learning ; Quantitative Finance - Risk Management ; Quantitative Finance - Trading and Microstructure</subject><creationdate>2023-04</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2304.02472$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2304.02472$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Lensky, Artem</creatorcontrib><creatorcontrib>Hao, Mingyu</creatorcontrib><title>Learning to Predict Short-Term Volatility with Order Flow Image Representation</title><description>Introduction: The paper addresses the challenging problem of predicting the short-term realized volatility of the Bitcoin price using order flow information. The inherent stochastic nature and anti-persistence of price pose difficulties in accurate prediction. Methods: To address this, we propose a method that transforms order flow data over a fixed time interval (snapshots) into images. The order flow includes trade sizes, trade directions, and limit order book, and is mapped into image colour channels. These images are then used to train both a simple 3-layer Convolutional Neural Network (CNN) and more advanced ResNet-18 and ConvMixer, with additionally supplementing them with hand-crafted features. The models are evaluated against classical GARCH, Multilayer Perceptron trained on raw data, and a naive guess method that considers current volatility as a prediction. Results: The experiments are conducted using price data from January 2021 and evaluate model performance in terms of root mean square error (RMSPE). The results show that our order flow representation with a CNN as a predictive model achieves the best performance, with an RMSPE of 0.85+/-1.1 for the model with aggregated features and 1.0+/-1.4 for the model without feature supplementation. ConvMixer with feature supplementation follows closely. In comparison, the RMSPE for the naive guess method was 1.4+/-3.0.</description><subject>Computer Science - Learning</subject><subject>Quantitative Finance - Risk Management</subject><subject>Quantitative Finance - Trading and Microstructure</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNqFjrsKwkAQAK-xEPUDrNwfSDzzQHsxKIiKhrThMGuycLkLm8OYv1eDvdU0AzNCzFfSjzZxLJeKX_T0g1BGvgyidTAWpyMqNmRKcBYujAXdHdwqy85LkWvIrFaONLkeOnIVnLlAhkTbDg61KhGu2DC2aNxHs2YqRg-lW5z9OBGLZJdu995QzhumWnGffw_y4SD8b7wBaig71A</recordid><startdate>20230404</startdate><enddate>20230404</enddate><creator>Lensky, Artem</creator><creator>Hao, Mingyu</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230404</creationdate><title>Learning to Predict Short-Term Volatility with Order Flow Image Representation</title><author>Lensky, Artem ; Hao, Mingyu</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-arxiv_primary_2304_024723</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Learning</topic><topic>Quantitative Finance - Risk Management</topic><topic>Quantitative Finance - Trading and Microstructure</topic><toplevel>online_resources</toplevel><creatorcontrib>Lensky, Artem</creatorcontrib><creatorcontrib>Hao, Mingyu</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Lensky, Artem</au><au>Hao, Mingyu</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Learning to Predict Short-Term Volatility with Order Flow Image Representation</atitle><date>2023-04-04</date><risdate>2023</risdate><abstract>Introduction: The paper addresses the challenging problem of predicting the short-term realized volatility of the Bitcoin price using order flow information. The inherent stochastic nature and anti-persistence of price pose difficulties in accurate prediction. Methods: To address this, we propose a method that transforms order flow data over a fixed time interval (snapshots) into images. The order flow includes trade sizes, trade directions, and limit order book, and is mapped into image colour channels. These images are then used to train both a simple 3-layer Convolutional Neural Network (CNN) and more advanced ResNet-18 and ConvMixer, with additionally supplementing them with hand-crafted features. The models are evaluated against classical GARCH, Multilayer Perceptron trained on raw data, and a naive guess method that considers current volatility as a prediction. Results: The experiments are conducted using price data from January 2021 and evaluate model performance in terms of root mean square error (RMSPE). The results show that our order flow representation with a CNN as a predictive model achieves the best performance, with an RMSPE of 0.85+/-1.1 for the model with aggregated features and 1.0+/-1.4 for the model without feature supplementation. ConvMixer with feature supplementation follows closely. In comparison, the RMSPE for the naive guess method was 1.4+/-3.0.</abstract><doi>10.48550/arxiv.2304.02472</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2304.02472
ispartof
issn
language eng
recordid cdi_arxiv_primary_2304_02472
source arXiv.org
subjects Computer Science - Learning
Quantitative Finance - Risk Management
Quantitative Finance - Trading and Microstructure
title Learning to Predict Short-Term Volatility with Order Flow Image Representation
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-01T00%3A53%3A48IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Learning%20to%20Predict%20Short-Term%20Volatility%20with%20Order%20Flow%20Image%20Representation&rft.au=Lensky,%20Artem&rft.date=2023-04-04&rft_id=info:doi/10.48550/arxiv.2304.02472&rft_dat=%3Carxiv_GOX%3E2304_02472%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true