{"paper_id":"2512.12727","title":"EXFormer: A Multi-Scale Trend-Aware Transformer with Dynamic Variable Selection for Foreign Exchange Returns Prediction","authors":["Dinggao Liu","Robert Ślepaczuk","Zhenpeng Tang"],"paper_date":"2025-12-14","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2512.12727v1","page_url":"https://thequant.space/flowcharts/exformer-a-multi-scale-trend-aware-transformer-with-dynamic/","scores":{"empirical_rigor":8.5,"math_complexity":8,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes EXFormer, a novel Transformer-based architecture for forecasting daily exchange rate returns.","quote":"EXFormer, a novel Transformer-based architecture specifically designed for forecasting the daily exchange rate returns.","summary":"Accurately forecasting daily exchange rate returns represents a longstanding challenge in international finance, as the exchange rate returns are driven by a multitude of correlated market factors and exhibit high-frequency fluctuations. This paper proposes EXFormer, a novel Transformer-based…","strategy":"EXFormer"},"assumptions":{"status":"available","items":[{"assumption":"Daily exchange rate returns are stationary.","type":"statistical","stated":"explicit","quote":"are stationary by construction and confirmed by the ADF tests above"},{"assumption":"No strong cointegrating relation among the three spot exchange rates.","type":"statistical","stated":"explicit","quote":"The results indicate that no strong cointegrating relation is present among the three series."},{"assumption":"Transaction costs are a fixed 5 bps plus 2 bps slippage per trade.","type":"execution","stated":"explicit","quote":"We model frictions as a fixed per-trade deduction equal to the sum of a 5 bps transaction cost and a 2 bps slippage cost"},{"assumption":"A 15‑day sliding window provides the best trade‑off between adaptability and noise resilience.","type":"model","stated":"implied","quote":"Only the 15-day configuration yields both immediate and sustained trading profits, balancing adaptability to fresh information with resilience against noise."},{"assumption":"Heteroskedasticity and autocorrelation consistent (HAC) standard errors are appropriate for forecast evaluation.","type":"statistical","stated":"explicit","quote":"using heteroskedasticity and autocorrelation consistent (HAC)"}]},"datasets":{"status":"available","items":[{"name":"Daily EUR/USD, USD/JPY, GBP/USD exchange rate returns","provider":null,"kind":"market-prices","access":"unknown","period":null,"universe":"Three G10 currency pairs","quote":"Using the daily data for EUR/USD, USD/JPY, and GBP/USD, we conduct out-of-sample evaluations across five different sliding windows."}],"known":[{"id":"wind","name":"Wind","provider":"Wind Information","kind":"market-prices","access":"commercial","quote":"All series spot exchange rates, equity indices, treasury yields, commodity indices, and macroeconomic indicators were retrieved from the Wind Database. 4.2.","mentions":1,"used_in":1}],"sample":{"start":null,"end":null,"market":"EUR/USD, USD/JPY, GBP/USD","asset_class":"fx","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"yes","headline_net_of_costs":true,"detail":"Fixed per‑trade deduction of 5 bps transaction cost plus 2 bps slippage, applied whenever the position changes sign or moves between {-1,0,1}.","quote":"We model frictions as a fixed per-trade deduction equal to the sum of a 5 bps transaction cost and a 2 bps slippage cost, applied whenever the position changes sign or moves between {-1,0,1}.","source":"front_matter","rule":"R3","candidates":{"front_matter":"yes","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"Fixed per‑trade deduction of 5 bps transaction cost plus 2 bps slippage, applied whenever the position changes sign or moves between {-1,0,1}.","card_quote":"We model frictions as a fixed per-trade deduction equal to the sum of a 5 bps transaction cost and a 2 bps slippage cost, applied whenever the position changes sign or moves between {-1,0,1}.","meaning":"costs are part of the reported headline results"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"EXFormer improves directional accuracy over the random walk by 8.5–22.8% in out‑of‑sample tests.","kind":"predictive-accuracy","quote":"EXFormer consistently outperforms the random walk and other baselines, improving directional accuracy by a statistically significant margin of up to 8.5–22.8%."},{"finding":"Cumulative returns of 18%, 25% and 18% for EUR/USD, USD/JPY and GBP/USD respectively in near‑one‑year backtests, with Sharpe ratios exceeding 1.8.","kind":"performance","quote":"In nearly one year of trading backtests, the model converts these gains into cumulative returns of 18%, 25%, and 18% for the three pairs, with Sharpe ratios exceeding 1.8."},{"finding":"After accounting for conservative transaction costs and slippage, EXFormer still yields positive cumulative returns of 7%, 19% and 9%, while baselines turn negative.","kind":"performance","quote":"When conservative transaction costs and slippage are accounted for, EXFormer retains cumulative returns of 7%, 19%, and 9%, while other baselines achieve negative."},{"finding":"Statistical superiority confirmed by Clark–West test and dynamic regression tests for both MSFE and directional accuracy.","kind":"statistical","quote":"and directional accuracy, with superiority confirmed by the Clark–West test and dynamic regression tests."},{"finding":"The dynamic variable selector provides time‑varying importance weights for 28 exogenous covariates, offering pre‑hoc interpretability.","kind":"other","quote":"A dynamic variable selector assigns time‑varying importance weights to 28 exogenous covariates related to exchange rate returns, providing pre‑hoc interpretability."}],"reported_performance":{"strategy":"EXFormer","sharpe":2.087,"sharpe_benchmark":-0.047,"annual_return_pct":null,"total_return_pct":25.782,"max_drawdown_pct":7.105},"out_of_sample":{"method":"walk-forward","headline_out_of_sample":true}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":44,"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":"2026-11-14T17:18:52Z"},"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/"}