Recognition: no theorem link
On the modeling assumptions of Historical Simulation for Value-at-Risk
Pith reviewed 2026-05-12 04:58 UTC · model grok-4.3
The pith
Defining a parametric model for asset returns and extracting historical innovations reproduces Historical Simulation VaR methods exactly.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
By explicitly defining a parametric model form for the asset returns and extracting the realized increments of the driving innovation process from historical data, the Historical Simulation, filtered Historical Simulation, and displaced Historical Simulation methods are reproduced exactly. This shows that the methods incorporate more underlying assumptions than what is often claimed in the literature.
What carries the argument
The parametric model for asset returns together with extraction of realized increments from the driving innovation process, which substitutes back into the model to produce the simulated returns used for VaR.
If this is right
- Each variant of Historical Simulation corresponds to a distinct choice of parametric return model.
- The methods are not free of parametric assumptions and can be viewed as special cases of a general innovation-based simulation procedure.
- Changes in the chosen parametric form will produce different VaR estimates even when the same historical data are used.
- The framework provides a way to compare and extend the methods by varying the return model specification.
Where Pith is reading between the lines
- Risk managers should document and test the return model they are implicitly using when applying these simulation methods.
- The same extraction approach could be applied to other data-driven risk calculations to reveal their hidden modeling steps.
- Misspecification of the parametric form may lead to VaR estimates that fail to capture current market dynamics even if historical data are recent.
Load-bearing premise
Once a parametric return model is chosen, historical data increments can be treated as direct realizations of the innovation process.
What would settle it
Implement the unified framework using the parametric form implied by a standard Historical Simulation implementation on identical data and verify whether the VaR numbers match; any systematic difference would show the reproduction does not hold.
read the original abstract
Historical Simulation (HS) and its extensions form a popular class of methods for estimating Value-at-Risk for portfolios of financial assets based on historical data. In this note, we seek to unify several ideas and models from throughout the literature into a single modeling framework. By explicitly defining a parametric model form for the asset returns and extracting the realized increments of the driving innovation process from historical data, we are able to reproduce the Historical Simulation, filtered Historical Simulation, and displaced Historical Simulation methods. This shows beyond a doubt that these methods need more underlying assumptions than what is often alluded to.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript claims that Historical Simulation (HS) and its extensions (filtered HS and displaced HS) for Value-at-Risk can be exactly reproduced within a single parametric framework by specifying an explicit return model of the form r_t = f(θ_t, ε_t) and inverting observed historical returns to recover the realized innovation sequence ε_t; resampling these innovations and re-applying the model then recovers the standard HS procedures, thereby demonstrating that these methods rest on more modeling assumptions than is typically stated.
Significance. If the claimed equivalences are rigorously established, the note provides a useful unifying lens on the implicit parametric structure of popular 'non-parametric' risk measures. This could help practitioners and regulators articulate the assumptions embedded in HS-based VaR calculations and clarify the boundary between parametric and historical methods in quantitative risk management.
minor comments (4)
- §2: The inversion step that extracts ε_t from r_t is presented as always feasible, but the manuscript should explicitly state the invertibility condition on f (e.g., monotonicity in ε) and note any cases where multiple ε values could map to the same r_t.
- §3.2 (filtered HS): The reproduction assumes that the time-varying parameters θ_t are estimated exactly as in the original filtered HS implementation; a short numerical example or pseudocode comparing the extracted ε sequence to a standard GARCH-filtered HS run would strengthen the claim of exact recovery.
- Abstract and §4: The phrasing 'shows beyond a doubt' is stronger than the constructive equivalence argument warrants; the text should instead state that the methods are recovered once the parametric form is chosen to match their implicit structure.
- References: The literature review omits several standard references on the original HS method (e.g., the 1990s RiskMetrics documentation and subsequent comparisons in Jorion's Value-at-Risk textbook) that would help situate the unification.
Simulated Author's Rebuttal
We thank the referee for the supportive summary, significance assessment, and recommendation of minor revision. We are pleased that the unifying parametric lens on Historical Simulation methods is viewed as potentially useful for clarifying assumptions in risk management practice.
Circularity Check
No significant circularity; derivation is self-contained unification
full rationale
The paper's central move is to posit an explicit parametric return model r_t = f(theta_t, epsilon_t), back out the historical epsilon sequence from observed returns, and then show that resampling those epsilons recovers the standard HS, filtered HS, and displaced HS procedures. This equivalence is presented as a modeling device to expose implicit assumptions rather than as a statistical prediction or first-principles derivation that must be independently verified. No equation or claim reduces a fitted quantity to itself by construction, no self-citation is load-bearing for the unification, and the framework remains falsifiable by checking whether the chosen parametric form actually matches the implicit structure of each HS variant. The result is therefore a transparent re-expression, not a circular loop.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Asset returns follow a parametric model form from which realized innovation increments can be extracted
Reference graph
Works this paper leans on
-
[1]
Philippe Jorion , title =
-
[2]
Federal Reserve Bank of New York Economic Policy Review , volume =
Darryll Hendricks , title =. Federal Reserve Bank of New York Economic Policy Review , volume =
-
[3]
Thomas J. Linsmeier and Neil D. Pearson , title =. Financial Analysts Journal , volume =
-
[4]
Carol Alexander , title =
-
[5]
Henrik Hult and Filip Lindskog and Ola Hammarlid and Carl Johan Rehn , title =
-
[6]
Attilio Meucci , title =
-
[7]
Journal of Banking & Finance , volume =
Matt Pritsker , title =. Journal of Banking & Finance , volume =
-
[8]
Journal of derivatives , volume=
An overview of value at risk , author=. Journal of derivatives , volume=
-
[9]
Journal of Empirical Finance , volume=
Displaced relative changes in historical simulation: Application to risk measures of interest rates with phases of negative rates , author=. Journal of Empirical Finance , volume=. 2017 , publisher=
work page 2017
-
[10]
Quantitative Finance , volume=
How to choose the return model for market risk? Getting towards a right magnitude of stressed VaR , author=. Quantitative Finance , volume=. 2019 , publisher=
work page 2019
-
[11]
Journal of derivatives , volume=
Value at risk when daily changes in market variables are not normally distributed , author=. Journal of derivatives , volume=. 1998 , publisher=
work page 1998
-
[12]
Journal of Futures Markets , volume =
Giovanni Barone-Adesi and Kostas Giannopoulos and Les Vosper , title =. Journal of Futures Markets , volume =
-
[13]
Jianqing Fan and Qiwei Yao , title =
- [14]
-
[15]
Journal of Econometrics , volume =
Tim Bollerslev , title =. Journal of Econometrics , volume =
- [16]
- [17]
- [18]
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.