{"paper_id":"2409.19386","title":"Multi-Factor Polynomial Diffusion Models and Inter-Temporal Futures Dynamics","authors":["Peilun He","Nino Kordzakhia","Gareth W. Peters","Pavel V. Shevchenko"],"paper_date":"2024-09-28","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2409.19386v1","page_url":"https://thequant.space/flowcharts/multi-factor-polynomial-diffusion-models-and-inter-temporal/","scores":{"empirical_rigor":4,"math_complexity":9,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes a family of multi‑factor stochastic commodity models based on polynomial diffusions, extending the Schwartz‑Smith two‑factor framework and providing methods for estimating latent spot prices and futures curve dynamics.","quote":"In this work, we develop the family of stochastic models using polynomial diffusion to obtain the unobservable spot price to be used for modelling futures curve dynamics. The polynomial family of diffusion models allows one to incorporate a variety of non-linear, higher-order effects, into a multi-factor stochastic model, which is a generalisation of Schwartz and Smith (2000) two-factor model.","summary":"In stochastic multi-factor commodity models, it is often the case that futures prices are explained by two latent state variables which represent the short and long term stochastic factors. In this work, we develop the family of stochastic models using polynomial diffusion to obtain the…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The logarithm of the spot price is the sum of a short‑term factor χₜ and a long‑term factor ξₜ.","type":"model","stated":"explicit","quote":"This approach models the logarithm of spot price St as the sum of two unobserv-able factors χt and ξt, log(St) = χt +ξt, (2) where χt represents the short-term fluctuation and ξt is the long-term equilibrium price level.","support":"full"},{"assumption":"Both χₜ and ξₜ follow Ornstein‑Uhlenbeck processes with mean‑reversion speeds κ and γ.","type":"model","stated":"explicit","quote":"We assume both χt and ξt follow an OU process, dχt = −κχtdt +σχdW χ t , (3) and dξt = (µξ −γξt)dt +σξdW ξ t , (4) while in [17] only one factor follows the OU process.","support":"full"},{"assumption":"Changes in the short‑term factor χₜ are temporary and converge to zero as t → ∞.","type":"model","stated":"explicit","quote":"We assume the changes in the short-term factor χt are temporary and converging to 0 as t →∞.","support":"full"},{"assumption":"The Brownian motions driving χₜ and ξₜ are correlated with coefficient ρ.","type":"model","stated":"explicit","quote":"The processes (W χ t )t≥0 and (W ξ t )t≥0 are correlated standard Brownian Motions with correlation coefficient ρ.","support":"full"},{"assumption":"Risk premiums λχ and λξ are constant, leading to risk‑neutral dynamics for χₜ and ξₜ.","type":"model","stated":"explicit","quote":"By assuming a constant risk premium λχ and λξ, the risk-neutral processes of χt and ξt are given by dχt = (−κχt −λχ)dt +σχdW χ∗ t , (5) and dξt = (µξ −γξt −λξ)dt +σξdW ξ∗ t , (6) where W χ∗ t and W ξ∗ t are correlated standard Brownian Motions with correlation coefficient ρ.","support":"full"},{"assumption":"Interest rates are assumed non‑stochastic when deriving futures prices.","type":"market","stated":"explicit","quote":"Let Ft,T be the market price of a futures contract at time t with maturity time T. For eliminating arbitrage, given all information until time t, the futures price must be equal to the expected spot price at the maturity time T. Therefore, under the risk-neutral measure, we have (assuming the interest rate is not stochastic) log(Ft,T) = log[E∗(ST|Ft)] = e−κ(T−t)χt +e−γ(T−t)ξt +A(T −t).","support":"full"}]},"datasets":{"status":"available","items":[{"name":"Synthetic simulation data (two datasets)","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"To evaluate the model, we conducted a simulation study on two datasets. The first data has 1000 observations and 13 contracts, with maturity time from T1 = 1 month for the first available con-tract to T13 = 13 months for the last available contract.","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"synthetic"},"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":"none","findings":[{"finding":"The eigen‑decomposition method provides the most accurate approximation of the matrix exponential (accuracy 0.0094) and is among the fastest methods.","kind":"performance","quote":"Eigen-decomposition method has the best accuracy which is only 0.0094. The Eigen-decomposition and Scaling and Squaring methods have the fastest computing time in ascending order.","support":"full"},{"finding":"Futures contracts can be accurately estimated by the polynomial diffusion model.","kind":"pricing-calibration","quote":"Overall, we found that the futures contracts can be accurately estimated.","support":"full"},{"finding":"Parameter estimation remains challenging and often fails to converge to true values, indicating identification problems.","kind":"negative","quote":"While the futures contracts can be estimated accurately, parameter estimation remains chal-lenging, even when separating parameters in the state equation and the coordinate representations in the measurement equation. However, parameter estimation remains challenging, even when we impose constrains on the model parameters and estimate coordinate repre-sentations of the polynomial of spot price.","support":"full"},{"finding":"Comparing seven matrix‑exponential methods shows similar stability across methods, but large differences in accuracy and efficiency.","kind":"performance","quote":"Methods Stability Accuracy Efficency Taylor series 1.7184 10.0214 0.1038 Pade approximation 1.7182 7.1447e+22 0.2415 Scaling and squaring 1.7183 4.3527 0.0657 Lagrange 1.7183 0.5412 0.1030 Newton 1.7183 0.0938 0.2098 Vandermonde 1.7183 4.0583e+10 0.1153 Eigen-decomposition 1.7183 0.0094 0.0340 The mean stability, mean accuracy and efficiency (in second) of seven meth-ods are given in Table 1. All methods have","support":"full"},{"finding":"The polynomial diffusion model generalises the Schwartz‑Smith two‑factor model and overcomes limitations of linear Gaussian spot‑price specifications.","kind":"theoretical","quote":"To overcome these two limitations, we introduced the polynomial diffusion model in this paper, which serves as a generalisation of the Schwartz-Smith two-factor model. In other words, while the state variables χt and ξt can not be estimated correctly, however the futures prices were recovered reasonably well.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":19,"citations":{"cited_by_in_corpus":2,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":2},"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":2,"extracted_at":"2026-11-14T19:55:06Z"},"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/"}