{"paper_id":"2505.24831","title":"Optimising cryptocurrency portfolios through stable clustering of price correlation networks","authors":["Ruixue Jing","Ryota Kobayashi","Luis Enrique Correa Rocha"],"paper_date":"2025-05-30","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2505.24831v1","page_url":"https://thequant.space/flowcharts/optimising-cryptocurrency-portfolios-through-stable-clusteri/","scores":{"empirical_rigor":6,"math_complexity":6.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes an integrated framework that combines network analysis, ARIMA price forecasts, consensus clustering, and modern portfolio theory to build profitable cryptocurrency portfolios.","quote":"This paper introduces an integrated framework that combines network analysis, price forecasting, and portfolio theory to identify stable groups of highly correlated cryptocurrencies for profitable portfolio construction.","summary":"The emerging cryptocurrency market presents unique challenges for investment due to its unregulated nature and inherent volatility. However, collective price movements can be explored to maximise profits with minimal risk using investment portfolios. In this paper, we develop a technical framework…","strategy":"P(ARIMA) predictive clustering portfolio"},"assumptions":{"status":"available","items":[{"assumption":"Cryptocurrency daily closing prices are stationary over the sample period.","type":"statistical","stated":"explicit","quote":"Augmented Dickey‑Fuller (ADF) test results confirm that the log returns ri,t of all examined cryptocurrencies are stationary, rejecting the null hypothesis at p < .001."},{"assumption":"Missing price data for more than ten days leads to exclusion of a cryptocurrency.","type":"data","stated":"explicit","quote":"Cryptocurrencies with missing prices for more than ten days are excluded."},{"assumption":"ARIMA models provide useful forecasts of future log returns.","type":"model","stated":"explicit","quote":"By incorporating predicted returns into the correlation network, the proposed P(ARIMA) strategy improves the detection of forward‑looking interdependencies and provides better‑performing portfolios over extended holding horizons."},{"assumption":"Trading can be executed at daily closing prices with manageable slippage and bid‑ask spreads.","type":"execution","stated":"implied","quote":"These returns also provide a buffer against typical cryptocurrency trading costs (approximately 0.10%–0.20% per side on major exchanges), such that at shorter horizons the strategies can remain profitable after transaction fees, provided that execution slippage and bid–ask spreads are well managed."},{"assumption":"The cryptocurrency market exhibits high volatility but the identified clusters remain persistent across regimes.","type":"market","stated":"implied","quote":"Central to our methodology is the identification of persistent clusters of highly interdependent cryptocurrencies via consensus clustering. These clusters capture structural co‑movement patterns that remain stable across market regimes."}]},"datasets":{"status":"available","items":[{"name":"Cryptocurrency price data","provider":"various online sources","kind":"market-prices","access":"public","period":"15.11.2017 to 15.04.2022","universe":"157 cryptocurrencies","quote":"The dataset comprises daily closing market prices (at midnight) of 5,450 cryptocurrencies"}],"known":[],"sample":{"start":"2017-11","end":"2022-04","market":"cryptocurrency market","asset_class":"crypto","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"unclear","headline_net_of_costs":false,"detail":null,"quote":null,"source":"conflict","rule":"R4b","candidates":{"front_matter":"no","evidence_card":"partial","evidence_card_quoted":true},"conflict_kind":"graded","card_detail":"Typical trading costs are about 0.10%–0.20% per side; the paper discusses that returns can remain profitable after these fees but does not embed them in headline performance numbers.","card_quote":"These returns also provide a buffer against typical cryptocurrency trading costs (approximately 0.10%–0.20% per side on major exchanges), such that at shorter horizons the strategies can remain profitable after transaction fees, provided that execution slippage and bid–ask spreads are well managed.","meaning":"the platform classifier and the paper-text extraction disagree, so no value is asserted"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"Predictive consensus‑clustering portfolios deliver consistently positive and stable performance up to a 14‑day holding horizon, with favourable gain‑loss asymmetry and tighter tail‑risk control.","kind":"performance","quote":"Our empirical results show that predictive consensus‑clustering portfolios maintain consistently positive and stable performance up to a 14‑day horizon, exhibit favourable gain‑loss asymmetry, and achieve tighter tail‑risk control."},{"finding":"The P(ARIMA) strategy outperforms the baseline historical‑only strategy, especially for longer holding periods.","kind":"performance","quote":"Our portfolio strategies that incorporate time‑series prediction of log returns outperform the Baseline strategy relying only on historical data, particularly over longer horizons."},{"finding":"Consensus clustering frequently identifies Bitcoin as part of the largest and most persistent clusters, reflecting its role as a market‑wide reference asset.","kind":"statistical","quote":"Our consensus clustering frequently identifies Bitcoin (BTC) as part of the largest and most persistent clusters, reflecting its role as a dominant asset and market‑wide reference point."},{"finding":"Incorporating predicted returns strengthens gain‑loss asymmetry, reduces tail‑risk exposure, and maintains profitability as horizons increase.","kind":"performance","quote":"prediction‑enhanced clustering systematically strengthens gain‑loss asymmetry, reduces tail‑risk exposure, and maintains profitability as horizons increase."},{"finding":"High computational costs of ARIMA forecasting may not justify adoption compared to simpler strategies.","kind":"negative","quote":"On the other hand, the high computational costs of forecasting models may not justify their adoption compared to our simpler strategies."}],"reported_performance":{"strategy":"P(ARIMA) predictive clustering portfolio","sharpe":null,"sharpe_benchmark":null,"annual_return_pct":null,"total_return_pct":null,"max_drawdown_pct":null},"out_of_sample":{"method":"walk-forward","headline_out_of_sample":true}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":37,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"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":"2026-11-14T17:36:54Z"},"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/"}