pith. sign in

arxiv: 1907.09218 · v2 · pith:HJBSDNL5new · submitted 2019-07-22 · 💱 q-fin.MF · q-fin.ST· q-fin.TR

Generalized statistical arbitrage concepts and related gain strategies

Christian Rein , Ludger R\"uschendorf , Thorsten Schmidt This is my paper

Pith reviewed 2026-05-24 17:57 UTC · model grok-4.3

classification 💱 q-fin.MF q-fin.STq-fin.TR
keywords generalized statistical arbitragesigma-algebratrading strategiesno-arbitragebinomial strategiesfollow-the-trendmarket data
0
0 comments X

The pith

Generalized statistical arbitrage allows strategies to yield positive average gains in sigma-algebra-specified scenarios rather than almost surely.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper defines generalized statistical arbitrage as trading strategies that produce positive expected gains under a class of scenarios given by an information system, instead of requiring gains almost surely. This framework treats classical arbitrage as a special case and also encompasses earlier statistical arbitrage notions. The authors characterize the corresponding no-arbitrage conditions and demonstrate that generalized gain strategies can still exist when standard no-arbitrage holds. They then construct explicit examples such as embedded binomial, follow-the-trend, and partition-type strategies, and report their performance both on simulated paths and on observed market data.

Core claim

Generalized statistical arbitrage notions are introduced and characterized using an information system given by a sigma-algebra; under standard no-arbitrage, generalized gain strategies exist that deliver positive average gains precisely under the scenarios selected by this sigma-algebra. Concrete constructions of such strategies, including embedded binomial, follow-the-trend, and partition-type forms, are provided and shown to produce positive performance on both simulated and market data.

What carries the argument

An information system given by a sigma-algebra that selects the relevant market scenarios in which average gains are required.

If this is right

  • Generalized statistical no-arbitrage conditions can be characterized for any chosen sigma-algebra.
  • Embedded binomial strategies, follow-the-trend strategies, and partition-type strategies qualify as generalized profitable strategies for suitable information systems.
  • These strategies exhibit positive performance on simulated data and on real market data, supporting their use in applications.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Different choices of sigma-algebra could let an investor encode specific views about which market states matter for average performance.
  • The framework may open a route to constructing near-arbitrage opportunities inside markets that satisfy classical no-arbitrage.
  • Extending the constructions to include transaction costs or model uncertainty would test how robust the reported performance remains.

Load-bearing premise

The sigma-algebra can be chosen so that the resulting class of scenarios meaningfully extends classical arbitrage while still permitting explicit construction of profitable strategies.

What would settle it

A concrete market model or data set in which, for every non-trivial sigma-algebra, no trading strategy (static or dynamic) produces strictly positive average gains on the selected scenarios.

Figures

Figures reproduced from arXiv: 1907.09218 by Christian Rein, Ludger R\"uschendorf, Thorsten Schmidt.

Figure 1
Figure 1. Figure 1: The considered trinomial model with T = 2 time steps. The first step is binomial, the second step is also (recombining) binomial with an additional top state {ω1, ω4}. the top state is reached by S2(ω1) = S2(ω4) = s ◦ 2 . We illustrate the scheme in [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: An explicit trinomial model with T = 2 time steps Lemma 3.3. Let ν1 := P (ω1) P (ω4) and ν2 := P (ω3) P (ω5) . In the trinomial model there is no statistical arbitrage if ν1 = − ∆S2(ω3) ∆S2(ω1) ν2 and if it holds that Γ1 < ν2 ≤ Γ2. (11) The proof is relegated to the appendix. 3.2. A counter example. In the following we use Lemma 3.3 to show that Propo￾sition 1 in Bondarenko (2003) is not valid without addi… view at source ↗
Figure 3
Figure 3. Figure 3: The considered recombining binomial model with two periods. Z(ω3) = Z(ω5), there would exist an equivalent martingale measure Q fulfilling the conditions q1 q4 = p1 p4 = 3 and q3 q5 = p3 p5 = 3. (12) But the only q ≥ 0 fulfilling (12) is q = ( 1 4 , 0, 1 4 , 1 12 , 1 12 , 1 3 ) which is not an element of Q. This example shows that Proposition 1 in Bondarenko (2003) needs additional assumptions: indeed, we … view at source ↗
Figure 4
Figure 4. Figure 4: The embedding of a binomial model: at the hitting times t1 and t2 of the diffusion the steps of the embedded binomial model take place. The hitting levels are given by s0(1 ± 0.15). procedure afresh by letting t i+1 0 = t i 2 . Generally, we assume that the time horizon T is sufficiently large such that the (typically small) levels s i 0 (1 − 2c), . . . , si 0 (1 + 2c) are reached at least once. Example 5.… view at source ↗
Figure 5
Figure 5. Figure 5: The embedded multi-period binomial trading model with trading points t 1 1 , t 1 2 , t 2 1 and t 2 2 . The statistical arbitrage in this case corresponds to repeated trading strategies from Lemma 3.7: we buy φ1 entities at t 1 0 = 0, change the position to φ + 2 at t 1 1 and equalize the position at t 1 2 . With the new starting time t 2 0 = t 1 2 the strategy will be started again and adjusted at the stop… view at source ↗
Figure 6
Figure 6. Figure 6: Histogram of the profits and losses from the embedded binomial trading strategy used in [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Illustration of the stopping times defined in (33), (34) resp. (35). The first stopping takes place when the process reaches either the first upper or lower boundary s i 0 (1 ± c). Starting from the upper boundary the next stopping takes place if the process increases to the level s i 0 (1 + 2c), decreases to the level s i 0 (1 − 2c) or crosses the level s0. In case the process reached the upper level a th… view at source ↗
Figure 8
Figure 8. Figure 8: The embedded binomial model for the follow-the-trend strategy with positive drift. The filtration generated by the final states is generated by each {ωi} for i = 1, 4, 5 and {ω2, ω3}. We also denote the resulting outcomes by s = s0, s +, s −, . . . and indicate this notation at some places. Trading will be executed at times τ i 1 to τ i 3 when the process reaches one of the predefined boundaries (or tradin… view at source ↗
Figure 9
Figure 9. Figure 9: The embedded binomial model for the follow-the-trend strategy with negative drift. The filtration generated by the final states is generated by each {ωi} for i = 1, 4, 5 and {ω2, ω3}. We also denote the resulting outcomes by s = s0, s +, s −, . . . and indicate this notation at some places. Proposition 6.1. In the follow-the-trend model with s −−+ > s0 there is G fin - arbitrage if  ψ1∆S1(ω1) + ψ + 2 ∆S2(… view at source ↗
Figure 10
Figure 10. Figure 10: Daily closing prices of the shares of the Kellogg Com￾pany during January 1, 2000 and December 31, 2017. Prices are presented in US-Dollar. and at least one inequality is strict. In this regard, define the matrix A by A =   ∆S1(ω1) ∆S2(ω1) 0 0 ∆S1(ω3) + r∆S1(ω2) r∆S2(ω2) ∆S2(ω3) 0 ∆S1(ω4) 0 ∆S2(ω4) ∆S3(ω4) ∆S1(ω5) 0 ∆S2(ω5) ∆S3(ω5)   with r = P (ω2) P (ω3) . If A is invertible, for any α ≥ 0, the … view at source ↗
Figure 11
Figure 11. Figure 11: Daily closing prices of the shares of the Deutsche Bank AG during January 1, 2000 and December 31, 2017. Prices are presented in Euro. a sliding-window approach with a window length of 3 years) [PITH_FULL_IMAGE:figures/full_fig_p029_11.png] view at source ↗
read the original abstract

Generalized statistical arbitrage concepts are introduced corresponding to trading strategies which yield positive gains on average in a class of scenarios rather than almost surely. The relevant scenarios or market states are specified via an information system given by a $\sigma$-algebra and so this notion contains classical arbitrage as a special case. It also covers the notion of statistical arbitrage introduced in Bondarenko (2003). Relaxing these notions further we introduce generalized profitable strategies which include also static or semi-static strategies. Under standard no-arbitrage there may exist generalized gain strategies yielding positive gains on average under the specified scenarios. In the first part of the paper we characterize these generalized statistical no-arbitrage notions. In the second part of the paper we construct several profitable generalized strategies with respect to various choices of the information system. In particular, we consider several forms of embedded binomial strategies and follow-the-trend strategies as well as partition-type strategies. We study and compare their behaviour on simulated data. Additionally, we find good performance on market data of these simple strategies which makes them profitable candidates for real applications.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper introduces generalized statistical arbitrage concepts tied to trading strategies that produce positive expected gains conditional on scenarios specified by a sigma-algebra (rather than almost surely). This framework is claimed to recover classical arbitrage (when the sigma-algebra is the full one) and to contain Bondarenko’s statistical arbitrage as a special case. The first part characterizes the associated generalized no-arbitrage conditions; the second part constructs explicit strategies (embedded binomial, follow-the-trend, partition-type) and reports their performance on simulated and real market data.

Significance. If the characterizations are rigorous and the constructed strategies demonstrably deliver positive conditional expected gains without introducing inconsistencies with standard no-arbitrage, the work could supply a flexible bridge between theoretical no-arbitrage theory and practical, scenario-conditioned trading rules. The explicit constructions and empirical comparisons constitute a concrete strength.

major comments (3)
  1. [Characterization of generalized no-arbitrage notions (first part)] The central extension via an arbitrary sigma-algebra is load-bearing for all subsequent claims. The manuscript must supply the precise measure-theoretic condition replacing “no positive gain a.s.” (likely in the characterization section) and prove that it is equivalent to the existence of an equivalent measure on the atoms; without this, it is unclear whether the notion collapses or fails to preserve strategy-construction properties for typical filtrations used in continuous-time models.
  2. [Introduction / Abstract] The claim that the generalized notion properly contains Bondarenko (2003) requires an explicit choice of sigma-algebra and a proof of inclusion; the abstract states containment but does not indicate the functional relationship that guarantees it.
  3. [Second part / Empirical study] For the constructed strategies (binomial, follow-the-trend, partition-type), the paper must verify that the reported positive average gains are indeed conditional on the chosen sigma-algebra atoms and are not artifacts of in-sample fitting or data snooping; otherwise the empirical support does not substantiate the theoretical claims.
minor comments (2)
  1. Notation for the information system (sigma-algebra) and the associated conditional expectations should be introduced once and used consistently; multiple ad-hoc symbols for the same object reduce readability.
  2. [Abstract] The abstract asserts “good performance on market data” without reporting quantitative metrics (Sharpe ratios, drawdowns, out-of-sample periods); these should be added for reproducibility.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thorough review and insightful comments on our manuscript. We address each of the major comments below and outline the revisions we plan to make.

read point-by-point responses

  1. Referee: [Characterization of generalized no-arbitrage notions (first part)] The central extension via an arbitrary sigma-algebra is load-bearing for all subsequent claims. The manuscript must supply the precise measure-theoretic condition replacing “no positive gain a.s.” (likely in the characterization section) and prove that it is equivalent to the existence of an equivalent measure on the atoms; without this, it is unclear whether the notion collapses or fails to preserve strategy-construction properties for typical filtrations used in continuous-time models.

    Authors: We acknowledge the importance of a rigorous measure-theoretic foundation. The manuscript's characterization section introduces the generalized no-arbitrage condition as the non-existence of a trading strategy with positive expected gain conditional on the sigma-algebra. We will revise to include the precise statement replacing the almost-sure condition and provide a proof of equivalence to the existence of an equivalent measure on the atoms of the sigma-algebra. Additionally, we will include a discussion on the applicability to continuous-time models to address potential concerns about the notion's behavior under typical filtrations. revision: yes

  2. Referee: [Introduction / Abstract] The claim that the generalized notion properly contains Bondarenko (2003) requires an explicit choice of sigma-algebra and a proof of inclusion; the abstract states containment but does not indicate the functional relationship that guarantees it.

    Authors: The abstract and introduction state that the generalized notion contains Bondarenko's statistical arbitrage as a special case. To make this explicit, we will specify the sigma-algebra (namely, the one generated by the relevant market variables at maturity) and add a short proof of the inclusion in a revised version of the introduction. This will clarify the functional relationship. revision: yes

  3. Referee: [Second part / Empirical study] For the constructed strategies (binomial, follow-the-trend, partition-type), the paper must verify that the reported positive average gains are indeed conditional on the chosen sigma-algebra atoms and are not artifacts of in-sample fitting or data snooping; otherwise the empirical support does not substantiate the theoretical claims.

    Authors: We agree that it is essential to confirm the conditional nature of the gains in the empirical section. The strategies are designed so that the positive expected gains hold conditionally on the atoms by construction. In the revised manuscript, we will include additional tables or figures showing the conditional expected gains on the sigma-algebra atoms for both simulated and market data. We will also emphasize the use of out-of-sample testing and robustness checks to rule out data snooping effects. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained

full rationale

The paper introduces definitions of generalized statistical arbitrage via an arbitrary sigma-algebra on scenarios, explicitly recovering classical arbitrage as the special case of the full sigma-algebra and extending the external Bondarenko (2003) notion. Characterization of the associated no-arbitrage conditions and explicit construction of strategies (embedded binomial, follow-the-trend, partition-type) proceed from standard measure-theoretic inequalities on conditional expectations without reducing any claimed prediction or uniqueness result to a fitted parameter, self-citation chain, or definitional tautology. No load-bearing self-citation, ansatz smuggling, or renaming of known results is exhibited.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit free parameters, axioms, or invented entities; the work rests on standard mathematical finance background such as no-arbitrage assumptions whose details are not visible.

pith-pipeline@v0.9.0 · 5722 in / 1067 out tokens · 24005 ms · 2026-05-24T17:57:58.678440+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

14 extracted references · 14 canonical work pages

  1. [1]

    and Lee, J.-H

    Avellaneda, M. and Lee, J.-H. (2010), ‘Statistical arbitrage in the US equities market’, Quantitative Finance 10(7), 761–782

    work page 2010

  2. [2]

    Bernardo, A. E. and Ledoit, O. (2000), ‘Gain, loss, and asset pricing’, Journal of political economy 108(1), 144–172

    work page 2000

  3. [3]

    (2003), ‘Statistical arbitrage and securities prices’, The Review of Financial Studies 16(3), 875–919

    Bondarenko, O. (2003), ‘Statistical arbitrage and securities prices’, The Review of Financial Studies 16(3), 875–919

    work page 2003

  4. [4]

    Borodin, A. N. and Salminen, P. (2012), Handbook of Brownian motion-facts and formulae, Birkh¨ auser. ˇCern` y, A. and Hodges, S. (2002), The theory of good-deal pricing in financial markets, in ‘Mathematical Finance—Bachelier Congress 2000’, Springer, pp. 175–202

    work page 2012

  5. [5]

    Cochrane, J. H. and Saa-Requejo, J. (2000), ‘Beyond arbitrage: Good-deal asset price bounds in incomplete markets’, Journal of political economy 108(1), 79–119

    work page 2000

  6. [6]

    and Schachermayer, W

    Delbaen, F. and Schachermayer, W. (1994), ‘A general version of the fundamental theorem of asset pricing’, Mathematische Annalen 300(1), 463–520

    work page 1994

  7. [7]

    and Schachermayer, W

    Delbaen, F. and Schachermayer, W. (1995 b), ‘The no-arbitrage property under a change of num´ eraire’,Stochastics and Stochastic Reports 53(3-4), 213–226

    work page 1995

  8. [8]

    and Schachermayer, W

    Delbaen, F. and Schachermayer, W. (2006), The mathematics of arbitrage , Springer Science & Business Media

    work page 2006

  9. [9]

    J., Van Der Hoek, J

    Elliott, R. J., Van Der Hoek, J. and Malcolm, W. P. (2005), ‘Pairs trading’, Quanti- tative Finance 5(3), 271–276

    work page 2005

  10. [10]

    Gatev, E., Goetzmann, W. N. and Rouwenhorst, K. G. (2006), ‘Pairs trading: Performance of a relative-value arbitrage rule’, The Review of Financial Studies 19(3), 797–827. G¨ onc¨ u, A. (2015), ‘Statistical arbitrage in the Black–Scholes framework’,Quantitative Finance 15(9), 1489–1499

    work page 2006

  11. [11]

    Hansen, L. P. and Jagannathan, R. (1991), ‘Implications of security market data for models of dynamic economies’, Journal of political economy 99(2), 225–262

    work page 1991

  12. [12]

    and Warachka, M

    Hogan, S., Jarrow, R., Teo, M. and Warachka, M. (2004), ‘Testing market efficiency using statistical arbitrage with applications to momentum and value strategies’, Journal of Financial economics 73(3), 525–565

    work page 2004

  13. [13]

    and Liebmann, T

    Kassberger, S. and Liebmann, T. (2017), ‘Additive portfolio improvement and utility-efficient payoffs’, Mathematics and Financial Economics 11(2), 241–262

    work page 2017

  14. [14]

    (2017), ‘Statistical arbitrage pairs trading strategies: Review and outlook’, Journal of Economic Surveys 31(2), 513–545

    Krauss, C. (2017), ‘Statistical arbitrage pairs trading strategies: Review and outlook’, Journal of Economic Surveys 31(2), 513–545. Freiburg University, Dep. of Mathematics, Ernst-Zermelo Str. 1, 79104 Freiburg, Germany. E-mail address : ch.rein@gmx.net, ruschen@stochastik.uni-freiburg.de Freiburg Institute of Advanced Studies (FRIAS), Germany. Universit...

    work page 2017