{"paper_id":"2601.02310","title":"Temporal Kolmogorov-Arnold Networks (T-KAN) for High-Frequency Limit Order Book Forecasting: Efficiency, Interpretability, and Alpha Decay","authors":["Ahmad Makinde"],"paper_date":"2026-01-05","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2601.02310v1","page_url":"https://thequant.space/flowcharts/temporal-kolmogorov-arnold-networks-t-kan-for-high-frequen/","scores":{"empirical_rigor":9,"math_complexity":8.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes Temporal Kolmogorov-Arnold Networks (T-KAN) that replace fixed linear LSTM weights with learnable B-spline activation functions for LOB forecasting.","quote":"we introduce Temporal Kolmogorov-Arnold Networks (T-KAN) to replace the fixed, linear weights of standard LSTMs with learnable B-spline activation functions.","summary":"High-Frequency trading (HFT) environments are characterised by large volumes of limit order book (LOB) data, which is notoriously noisy and non-linear. Alpha decay represents a significant challenge, with traditional models such as DeepLOB losing predictive power as the time horizon (k) increases.…","strategy":"Temporal Kolmogorov-Arnold Networks (T-KAN)"},"assumptions":{"status":"available","items":[{"assumption":"The universal approximation approach with fixed activations is parameter‑inefficient for capturing localized oscillations in microstructure data.","type":"model","stated":"explicit","quote":"We assume that this ’universal approximation’ approach is parameter-inefficient for capturing localized oscillations found in microstructure data."},{"assumption":"Mean and standard deviation are computed over a rolling window to keep the auction environment stationary.","type":"data","stated":"explicit","quote":"where µ and σ are the mean and standard deviation calculated over a rolling window to maintain a stationary auction environment."},{"assumption":"Class imbalance is addressed by inverse frequency weighting in the loss function.","type":"model","stated":"explicit","quote":"To address the significant class imbalance in the FI-2010 Auction dataset, we utilize Inverse Frequency Weighting"},{"assumption":"A flat 1.0 basis‑point transaction cost is applied to the mid‑price trading simulation.","type":"execution","stated":"explicit","quote":"A mid-price trading simulation was conducted using a 1.0 bps transaction cost to evaluate the economic significance of the model’s predictions."},{"assumption":"The FI‑2010 dataset provides a representative benchmark for high‑frequency LOB forecasting.","type":"data","stated":"implied","quote":"The FI-2010 benchmark dataset [4] was used to evaluate the performance of Temporal Kolmogorov-Arnold Networks against the DeepLOB baseline."}]},"datasets":{"status":"available","items":[{"name":"FI-2010","provider":"Ntakaris et al.","kind":"order-book","access":"public","period":null,"universe":null,"quote":"using data from the FI-2010 dataset","known_id":"fi-2010"}],"known":[{"id":"fi-2010","name":"FI-2010","provider":"Ntakaris et al.","kind":"order-book","access":"public","quote":"equency Weighting To address the significant class imbalance in the FI-2010 Auction dataset, we utilize Inverse Frequency Weighting [7] in our loss function.","mentions":10,"used_in":4}],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":"intraday"},"data_resolution":"l2-book"},"transaction_costs":{"status":"available","modeled":"yes","headline_net_of_costs":true,"detail":"A flat 1.0 bps transaction cost is applied in the backtest.","quote":"A mid-price trading simulation was conducted using a 1.0 bps transaction cost to evaluate the economic significance of the model’s predictions.","source":"front_matter","rule":"R3","candidates":{"front_matter":"yes","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"A flat 1.0 bps transaction cost is applied in the backtest.","card_quote":"A mid-price trading simulation was conducted using a 1.0 bps transaction cost to evaluate the economic significance of the model’s predictions.","meaning":"costs are part of the reported headline results"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"T‑KAN improves F1‑score by 19.1% relative to DeepLOB at a 100‑tick horizon (0.3995 vs 0.3354).","kind":"performance","quote":"T-KAN achieved an F1-Score of 0.3995, representing a relative improvement of 19.1% over the baseline of 0.3354."},{"finding":"In a transaction‑cost‑adjusted backtest, T‑KAN yields a terminal return of 132.48% versus DeepLOB’s –82.76%.","kind":"performance","quote":"the T-KAN model resulted in a terminal return of 132.48%"},{"finding":"Learned B‑spline activations create a dead‑zone near zero‑mean inputs, filtering micro‑structural noise.","kind":"other","quote":"This learned function effectively creates a \"dead-zone\" near zero-mean inputs, filtering out micro-structural noise while non-linearly amplifying high-conviction signals."},{"finding":"T‑KAN uses more parameters (104,451) than DeepLOB (58,211) but achieves higher profitability density per parameter.","kind":"other","quote":"While T-KAN uses a far higher parameter footprint (104,451 vs. 58,211), the ’profitability density’ per parameter is significantly higher."}],"reported_performance":{"strategy":"Temporal Kolmogorov-Arnold Networks (T-KAN)","sharpe":null,"sharpe_benchmark":null,"annual_return_pct":null,"total_return_pct":132.48,"max_drawdown_pct":null},"out_of_sample":{"method":"holdout","headline_out_of_sample":true}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":45,"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":true,"code_url":"https://github.com/AhmadMak/Temporal-Kolmogorov-Arnold-Networks-T-KAN-for-High-Frequency-Limit-Order-Book-Forecasting","note":null},"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":"2026-11-14T17:17:21Z"},"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/"}