Standard

Noise fit, estimation error and a Sharpe information criterion. / Paulsen, Dirk; Söhl, Jakob.

In: Quantitative Finance, Vol. 20, No. 6, 2020, p. 1027-1043.

Research output: Contribution to journalArticleScientificpeer-review

Harvard

APA

Vancouver

Author

Paulsen, Dirk ; Söhl, Jakob. / Noise fit, estimation error and a Sharpe information criterion. In: Quantitative Finance. 2020 ; Vol. 20, No. 6. pp. 1027-1043.

BibTeX

@article{9dc7e5c74dc740e193ae76d822a1d6c5,
title = "Noise fit, estimation error and a Sharpe information criterion",
abstract = "When the in-sample Sharpe ratio is obtained by optimizing over a k-dimensional parameter space, it is a biased estimator for what can be expected on unseen data (out-of-sample). We derive (1) an unbiased estimator adjusting for both sources of bias: noise fit and estimation error. We then show (2) how to use the adjusted Sharpe ratio as model selection criterion analogously to the Akaike Information Criterion (AIC). Selecting a model with the highest adjusted Sharpe ratio selects the model with the highest estimated out-of-sample Sharpe ratio in the same way as selection by AIC does for the log-likelihood as a measure of fit.",
keywords = "AIC, Akaike information criterion, Backtesting, Estimation error, Model selection, Noise fit, Overfit, Sharpe ratio, Sharpe ratio information criterion, SRIC",
author = "Dirk Paulsen and Jakob S{\"o}hl",
year = "2020",
doi = "10.1080/14697688.2020.1718746",
language = "English",
volume = "20",
pages = "1027--1043",
journal = "Quantitative Finance",
issn = "1469-7688",
publisher = "Routledge - Taylor & Francis Group",
number = "6",

}

RIS

TY - JOUR

T1 - Noise fit, estimation error and a Sharpe information criterion

AU - Paulsen, Dirk

AU - Söhl, Jakob

PY - 2020

Y1 - 2020

N2 - When the in-sample Sharpe ratio is obtained by optimizing over a k-dimensional parameter space, it is a biased estimator for what can be expected on unseen data (out-of-sample). We derive (1) an unbiased estimator adjusting for both sources of bias: noise fit and estimation error. We then show (2) how to use the adjusted Sharpe ratio as model selection criterion analogously to the Akaike Information Criterion (AIC). Selecting a model with the highest adjusted Sharpe ratio selects the model with the highest estimated out-of-sample Sharpe ratio in the same way as selection by AIC does for the log-likelihood as a measure of fit.

AB - When the in-sample Sharpe ratio is obtained by optimizing over a k-dimensional parameter space, it is a biased estimator for what can be expected on unseen data (out-of-sample). We derive (1) an unbiased estimator adjusting for both sources of bias: noise fit and estimation error. We then show (2) how to use the adjusted Sharpe ratio as model selection criterion analogously to the Akaike Information Criterion (AIC). Selecting a model with the highest adjusted Sharpe ratio selects the model with the highest estimated out-of-sample Sharpe ratio in the same way as selection by AIC does for the log-likelihood as a measure of fit.

KW - AIC

KW - Akaike information criterion

KW - Backtesting

KW - Estimation error

KW - Model selection

KW - Noise fit

KW - Overfit

KW - Sharpe ratio

KW - Sharpe ratio information criterion

KW - SRIC

UR - http://www.scopus.com/inward/record.url?scp=85079817133&partnerID=8YFLogxK

U2 - 10.1080/14697688.2020.1718746

DO - 10.1080/14697688.2020.1718746

M3 - Article

AN - SCOPUS:85079817133

VL - 20

SP - 1027

EP - 1043

JO - Quantitative Finance

JF - Quantitative Finance

SN - 1469-7688

IS - 6

ER -

ID: 71197519