{"paper_id":"2401.03393","title":"Modelling and Predicting the Conditional Variance of Bitcoin Daily Returns: Comparsion of Markov Switching GARCH and SV Models","authors":["Dennis Koch","Vahidin Jeleskovic","Zahid I. Younas"],"paper_date":"2024-01-07","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2401.03393v2","page_url":"https://thequant.space/flowcharts/modelling-and-predicting-the-conditional-variance-of-bitcoin/","scores":{"empirical_rigor":8,"math_complexity":7.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a two‑stage estimation approach separating mean and variance equation coefficients and compares Markov Switching‑GARCH with Stochastic Autoregressive Volatility models for Bitcoin volatility forecasting.","quote":"The paper provides insights into the rationale behind the recommendation for a two-stage estimation approach, emphasizing the separate estimation of coefficients in the mean and variance equations.","summary":"This paper introduces a unique and valuable research design aimed at analyzing Bitcoin price volatility. To achieve this, a range of models from the Markov Switching-GARCH and Stochastic Autoregressive Volatility (SARV) model classes are considered and their out-of-sample forecasting performance is…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The mean process must be free from serial correlation to obtain unbiased MS‑GARCH results.","type":"model","stated":"explicit","quote":"If researchers intend to employ MS‑GARCH models with the aforementioned specifications, it is imperative to have a mean process that is free from serial correlation in order to obtain unbiased results."},{"assumption":"Simple GARCH models are assumed to have stationary variance with persistence close to unity.","type":"statistical","stated":"explicit","quote":"a simple GARCH model with stationary variance and persistence close to unity (the sum of the parameters in the conditional variance equation) could serve as a second‑best solution for risk management purposes."},{"assumption":"Model parameters are re‑estimated daily, implying a rolling‑window forecasting setup.","type":"execution","stated":"implied","quote":"The parameters of the best in‑sample models are re‑estimated in order to obtain more efficient model parameters. The rolling window inherits 800 observations."}]},"datasets":{"status":"available","items":[{"name":"Bitcoin daily closing prices","provider":"coinmarketcap.com","kind":"market-prices","access":"public","period":"29 April 2013 to 9 June 2019","universe":"Bitcoin (USD/BTC)","quote":"The daily Bitcoin closing prices denoted in USD are transformed into daily log returns, resulting in a total of 2233 observations. The period under review begins on 29th April 2013 and extends to 9th June 2019.","known_id":"coinmarketcap"}],"known":[{"id":"coinmarketcap","name":"CoinMarketCap","provider":"CoinMarketCap","kind":"market-prices","access":"public","quote":"3.3 Data and Data Sources The data set is obtained from the website coinmarketcap.comXI.","mentions":1,"used_in":1}],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"forecast-accuracy","findings":[{"finding":"SARV models outperform MS‑GARCH models in one‑step‑ahead variance prediction according to the quasi‑likelihood loss measure.","kind":"predictive-accuracy","quote":"According to the QL loss measure, it is found that SARV models outperform MS‑GARCH models in terms of their ability to predict variance."},{"finding":"In certain situations, a simple GARCH model may even outperform Markov‑Switching GARCH models for predicting Bitcoin log‑return variance.","kind":"performance","quote":"the study suggests that in certain situations, persistent simple GARCH models may even outperform Markov‑Switching GARCH models in predicting the variance of Bitcoin log returns."},{"finding":"All information criteria consistently indicate that the simple GARCH model is the most suitable for describing Bitcoin log‑return conditional variance across regimes.","kind":"statistical","quote":"The estimation of various MS‑GARCH models indicates that the simple GARCH model is the most suitable model for describing the conditional variance of Bitcoin log returns. To be more precise, all information criteria (AIC, BIC, and HQ) consistently suggest that the simple GARCH model performs best in both volatility reg"},{"finding":"SARV model parameter estimates are highly sensitive to initial values due to the weighted least squares target function with heteroscedastic innovations.","kind":"statistical","quote":"The in‑sample parameter estimation of SARV models exhibits a high degree of sensitivity to initial values"}],"reported_performance":null,"out_of_sample":{"method":"walk-forward","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":18,"citations":{"cited_by_in_corpus":1,"by_independent_authors":1,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":1},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":1,"extracted_at":"2026-11-14T18:16:35Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}