{"paper_id":"2512.04099","title":"Partial multivariate transformer as a tool for cryptocurrencies time series prediction","authors":["Andrzej Tokajuk","Jarosław A. Chudziak"],"paper_date":"2025-11-22","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2512.04099v1","page_url":"https://thequant.space/flowcharts/partial-multivariate-transformer-as-a-tool-for-cryptocurrenc/","scores":{"empirical_rigor":6,"math_complexity":7,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes the Partial-Multivariate Transformer (PMformer) for cryptocurrency price forecasting, balancing information scarcity and noise.","quote":"We apply the Partial-Multivariate Transformer (PMformer) to forecast daily returns for BTCUSDT and ETHUSDT, benchmarking it against eleven classical and deep learning models.","summary":"Forecasting cryptocurrency prices is hindered by extreme volatility and a methodological dilemma between information-scarce univariate models and noise-prone full-multivariate models. This paper investigates a partial-multivariate approach to balance this trade-off, hypothesizing that a strategic…","strategy":"Partial-Multivariate Transformer (PMformer)"},"assumptions":{"status":"available","items":[{"assumption":"The prediction target is the next‑day logarithmic return.","type":"data","stated":"explicit","quote":"The prediction target is the next-day logarithmic return, rt = ln(Pt/Pt−1), a standard variance-stabilizing transform for financial series [22]."},{"assumption":"All numeric features are scaled to [0,1] by Min–Max normalization fitted on the training split.","type":"data","stated":"explicit","quote":"All numeric features are scaled to [0, 1] by Min–Max normalization [25]; scaling parameters are fit on the training split and applied to the validation/test set to prevent leakage."},{"assumption":"The dataset is split chronologically into 70% train, 20% validation, and 10% test.","type":"data","stated":"explicit","quote":"The dataset is split chronologically into 70% train, 20% validation, and 10% test."},{"assumption":"Trading follows a simple sign‑based rule: long if predicted return > 0, short otherwise.","type":"execution","stated":"explicit","quote":"The trading strategy is a simple sign-based rule: go long if predicted return ˆrt+1 > 0 and short otherwise."},{"assumption":"PMformer trains on sampled feature subsets, acting as an implicit adaptive feature selection step that curbs variance.","type":"model","stated":"explicit","quote":"Because PMformer trains on sampled feature subsets, this acts as an implicit, adaptive feature selection step that curbs variance without exhaustive multivariate inputs [19]."},{"assumption":"Technical indicators (SMA, EMA, RSI, CCI, ATR, MACD) capture trend, momentum, and volatility sufficiently for the partial‑multivariate approach.","type":"model","stated":"implied","quote":"This compact set captures trend, momentum, and volatility while remaining compatible with partial-multivariate training."}]},"datasets":{"status":"available","items":[{"name":"BTCUSDT and ETHUSDT daily data","provider":"Binance","kind":"market-prices","access":"public","period":"Oct. 5, 2017 to May 20, 2025","universe":"Bitcoin (BTCUSDT) and Ethereum (ETHUSDT)","quote":"We use daily Bitcoin (BTCUSDT) and Ethereum (ETHUSDT) data from Binance from Oct. 5, 2017 to May 20, 2025.","known_id":"binance"}],"known":[{"id":"binance","name":"Binance","provider":"Binance","kind":"market-prices","access":"public","quote":"DATASETS AND FEATURE ENGINEERING We use daily Bitcoin (BTCUSDT) and Ethereum (ETHUSDT) data from Binance from Oct.","mentions":1,"used_in":1}],"sample":{"start":null,"end":null,"market":"BTC/USDT","asset_class":"crypto","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"no","headline_net_of_costs":false,"detail":null,"quote":null,"source":"front_matter","rule":"R1","candidates":{"front_matter":"no","evidence_card":"unclear","evidence_card_quoted":true},"card_detail":null,"card_quote":"especially in a market where even accurate forecasts might be too small relative to transaction costs or volatility to exploit.","meaning":"a trading or execution result is reported without costs"},"results":{"status":"available","evaluation_type":"unclear","findings":[{"finding":"PMformer achieved the lowest MSE on BTCUSDT (6.5798 × 10⁻⁴), confirming its superior statistical accuracy.","kind":"predictive-accuracy","quote":"PMformer achieved the highest statistical accuracy, posting the lowest MSE (6.5798 × 10−4) among all tested models."},{"finding":"On BTCUSDT, PMformer attained the highest directional accuracy (59.8%) and the lowest maximum drawdown (‑13.8%).","kind":"performance","quote":"PMformer provided superior risk control, securing the lowest maximum drawdown (-13.8%) and the highest directional accuracy (59.8%)."},{"finding":"FEDformer yielded a higher total ROI (+38.1%) and Sharpe Ratio (4.54) than PMformer, highlighting a trade‑off between return and risk.","kind":"performance","quote":"While FEDformer yielded a higher Total ROI (+38.1%) and Sharpe Ratio (4.54), PMformer provided superior risk control..."},{"finding":"For ETHUSDT, PMformer had the lowest forecasting errors but all models, including PMformer, generated negative total ROI, indicating prediction accuracy did not translate to profit.","kind":"negative","quote":"PMformer confirmed expectations on the forecasting front (yielding the lowest errors for ETHUSDT as expected), it also provided a cautionary confirmation that prediction accuracy alone is not enough for successful trading..."},{"finding":"The study reveals a pronounced disconnect between statistical accuracy and practical trading utility across both assets.","kind":"statistical","quote":"Our analysis reveals a pronounced disconnect between statistical accuracy and practical trading utility. For both assets, the models with the lowest prediction errors were frequently not the most profitable."}],"reported_performance":{"strategy":"Partial-Multivariate Transformer (PMformer)","sharpe":3.83,"sharpe_benchmark":null,"annual_return_pct":null,"total_return_pct":20.62,"max_drawdown_pct":13.8},"out_of_sample":{"method":"holdout","headline_out_of_sample":true}},"independent_evidence":{"status":"available","supports":[{"paper":{"paper_id":"2604.12082","title":"When Forecast Accuracy Fails: Rank Correlation and Decision Quality in Multi-Market Battery Storage Optimization","paper_date":"2026-04-13","authors":["Alessandro Falezza"],"empirical_rigor":8.5,"page_url":"https://thequant.space/flowcharts/when-forecast-accuracy-fails-rank-correlation-and-decision-quality-in-multi/"},"question":"Do lower forecasting errors guarantee higher economic profit in market decisions?","this_paper_finding":"For ETHUSDT, PMformer had the lowest forecasting errors but all models, including PMformer, generated negative total ROI, indicating prediction accuracy did not translate to profit.","other_paper_finding":"Persistence forecasts capture only 32.8% of oracle value despite high MAE, whereas ML‑based forecasts with τ≈0.9 capture >99% of oracle value, showing misalignment between MAE and economic performance.","other_paper_quote":"By contrast, the ML 1-step forecast captures 99.2% of oracle value, and the Hybrid 8h forecast 99.7%.","note":"Both studies show error metrics mislead profit: A's PMformer has lowest errors yet negative ROI; B's persistence MAE captures only 32.8% oracle value, while ML τ≈0.9 captures >99% despite MAE differences.","confidence":0.461}],"contradicts":[],"qualifies":[],"compared_with":64,"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":"2026-11-14T17:21:19Z"},"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/"}