Conditional Path Decomposition at the Infimum and Maximum Drawdowns for Spectrally Negative Lévy Processes
C. Vardar-Acar, M. Çağlar · 2026
Read Conditional Path Decomposition at the Infimum and Maximum Drawdowns for Spectrally Negative Lévy Processes by C. Vardar-Acar, M. Çağlar (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-27573.txt.
Introduction
C Vardar-Acar and M Çağlar study processes that jump only downward. These are called spectrally negative Lévy processes. The authors ask how deep the worst falls can be. A drawdown is the drop from a past peak to a later low. The infimum is the lowest value reached. Their 2026 work watches the process up to a random exponential time limit.
The path is split at its lowest point. One split gives a part before the low and a part after it. A second split adds a middle section when the low comes before the high. The authors describe the early part with scale functions, special functions that encode these processes. The laws they derive are Doob h-transforms, reweighted versions of killed processes. The formulas give the largest drawdown on each separate piece. They also confirm the classical results known for Brownian motion.
Catalog record
| CREATOR | C. Vardar-Acar, M. Çağlar |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.27573
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Conditional Path Decomposition at the Infimum and Maximum Drawdowns for Spectrally Negative Lévy Processes C. Vardar-Acar, M. Çağlar Abstract We study maximum-drawdown laws conditioned on extremes for a spectrally negative Lévy process and observed up to an independent exponential time. The main contribution is a set of scale-function characterizations of the pre-infimum path arising from two decompositions of the process. The first is the decomposition at the infimum into pre-infimum and post-infimum components. The second, under the ordering in which the infimum is attained before the supremum, decomposes the path into pre-infimum, intermediate, and post-supremum components. We also identify the distribution of the supremum for the pre-infimum process in the first decomposition. The resulting conditional laws are expressed as Doob h h -transforms of killed spectrally negative Lévy processes and they yield explicit formulas for the maximum drawdown on each independent path component. The results confirm the classical decompositions for Brownian motion. Keywords: Spectrally negative Lévy process; Maximum drawdown; Scale functions; Doob h -transform; Path decomposition at extrema; Exponential time horizon. 1 Introduction Maximum drawdown is considered as a measure of the largest realized loss/risk in the financial instruments. Its conditional distribution is relevant for risk assessment, timing rules, drawdown-linked contracts, and path-dependent derivatives, ( [ 24 ] , [ 12 ] , [ 33 ] , [ 8 ] , [ 9 ] , [ 11 ] , [ 8 ] , [ 19 ] , [ 18 ] , [ 7 ] , [ 32 ] , [ 14 ] , [ 25 ] , [ 28 ] , [ 31 ] , [ 4 ] , [ 21 ] , [ 17 ] and [ 5 ] ). The distributional properties of maximum drawdowns are studied in ( [ 10 ] , [ 19 ] , [ 26 ] , [ 30 ] , [ 29 ] , [ 23 ] , [ 27 ] , [ 1 ] , [ 13 ] , [ 20 ] ). The distributions of maximum drawdown are closely connected to first-passage problems. Scale functions provide tractable identities for such passage problems. In the spectrally negative Lévy setting, see [ 3 ] , [ 6 ] , [ 16 ] and [ 15 ] . For standard Brownian motion, [ 10 ] and [ 26 ] have obtained laws of maximum drawdown and maximum drawup. Explicitly in Brownian and diffusion models, distributions are found up to a passage time as well as exponential time. Maximum drawdown and maximum drawup have been analyzed for spectrally negative Lévy processes until the passage time of a given level in [ 30 ] . Path decompositions at extrema for spectrally negative Lévy processes together with applications to maximum drawdown and drawdown duration were developed by [ 29 ] . However, the law of the pre-infimum component was left as an open question in a setting where the process was split first into two at its global infimum, and next into three pieces under the condition that infimum is reached before the supremum. The present paper completes the missing part by deriving the law of the pre-infimum component and obtaining the distribution of its maximum-drawdown. In this paper, we consider a spectrally negative Lévy process X observed up to an independent exponential time T with rate \gamma>0 . We assume that X is not the negative of a subordinator and hence it is of unbounded variation. Let S_{t}:=\sup_{0\leq s\leq t}X_{s},\qquad I_{t}:=\inf_{0\leq s\leq t}X_{s} denote its running supremum and infimum processes, respectively, for 0\leq t\leq T . The drawdown and drawup processes are defined as D_{t}:=S_{t}-X_{t},\qquad U_{t}:=X_{t}-I_{t}, and the corresponding maximum drawdown and maximum drawup are \displaystyle M_{t}^{-} \displaystyle: \displaystyle=\sup_{0\leq u\leq v\leq t}(X_{u}-X_{v})=\sup_{0\leq s\leq t}D_{s}, \displaystyle M_{t}^{+} \displaystyle: \displaystyle=\sup_{0\leq u\leq v\leq t}(X_{v}-X_{u})=\sup_{0\leq s\leq t}U_{s}. (1.1) Conditional on the extrema I_{T}=a<0<S_{T}=b, the path is decomposed according to infimum, and also with respect to both extremes given that the infimum is attained before the supremum. Explicitly, we focus on the ordering H_{I}<H_{S}, where H_{I} and H_{S} denote attainment times of I_{T} and S_{T} , respectively. In this case, the path decomposes into three pieces: the pre-infimum, the intermediate path from the infimum to the supremum, and the post-supremum. The novelty of the present work is the identification of the conditional law of the pre-infimum path component towards the objective of obtaining the conditional distribution of the maximum drawdown. We characterize this component both in the decomposition at the infimum and also, more delicately, under the conditioning H_{I}<H_{S},I_{T}=a,S_{T}=b, where the infimum is attained before the supremum. These conditional pre-infimum laws are obtained using Doob- h transforms and in terms of well known scale functions W^{(\gamma)} , Z^{(\gamma)} . The formulas include the derivative of W^{(\gamma)} , which exists everywhere by our assumption that X is of unbounded variation. Explicitly, we show in Theorem 3.1 that the law of the pre- H_{I} process is an h -transform of the law of the spectrally negative Lévy process killed at \tau_{a}\wedge T with h(x)=\frac{\gamma}{\Phi(\gamma)}W^{(\gamma)\prime}(x-a)-\gamma W^{(\gamma)}(x-a) for x\geq a , obtain the conditional distribution of the supremum as \displaystyle\mathbb{P}_{0}\left\{S_{H_{I}}<b\mid I_{T}=a\right\}=1-\frac{\frac{\gamma}{\Phi(\gamma)}W^{(\gamma)\prime}(b-a)-\gamma W^{(\gamma)}(b-a)}{\frac{\gamma}{\Phi(\gamma)}W^{(\gamma)\prime}(-a)-\gamma W^{(\gamma)}(-a)}\left[\frac{W^{(\gamma)}(-a)}{W^{(\gamma)}(b-a)}\right] for b\geq 0 , and derive the conditional distribution of maximum drawdown M_{0,H_{I}}^{-} of the pre- H_{I} process given by \displaystyle\mathbb{P}\left\{M_{0,H_{I}}^{-}<d\mid I_{T}=a\right\}=1-\frac{\frac{\gamma}{\Phi(\gamma)}W^{(\gamma)\prime}(d)-\gamma W^{(\gamma)}(d)}{\frac{\gamma}{\Phi(\gamma)}W^{(\gamma)\prime}(-a)-\gamma W^{(\gamma)}(-a)}\left[\frac{W^{(\gamma)}(-a)}{W^{(\gamma)}(d)}\right] for d\geq-a . In Proposition 4.1 , the process is split at both infumum, with I_{T}=a , and supremum, with S_{T}=b . In this case, the pre- H_{I} distribution is found as the Doob- h transform of the process killed on leaving (a,b) through the lower boundary, with h_{1}(x)=\frac{W^{(\gamma)\prime}(x-a)W^{(\gamma)}(b-a)-W^{(\gamma)\prime}(b-a)W^{(\gamma)}(x-a)}{\big(W^{(\gamma)}(b-a)\big)^{2}},\qquad a<x<b. Under the same decomposition, the distribution of maximum drawdown is found in Theorem 4.3 as \hskip-28.45274pt\small\mathbb{P}_{a,b}^{<}\{M_{0,H_{I}}^{-}<d\}=\begin{cases}0,&d\leq-a,\\ \displaystyle 1-\left[\frac{W^{(\gamma)\prime}(d)W^{(\gamma)}(r)-W^{(\gamma)\prime}(r)W^{(\gamma)}(d)}{W^{(\gamma)\prime}(-a)W^{(\gamma)}(r)-W^{(\gamma)\prime}(r)W^{(\gamma)}(-a)}\right]\left(\frac{W^{(\gamma)}(-a)}{W^{(\gamma)}(d)}\right),&-a<d<r,\\ 1,&d\geq r.\end{cases} where r=b-a . The paper is organized as follows. The preliminary definitions and our assumptions are given in Section 2 . We perform two path decompositions through the extremes, one at the global infimum in Section 3 and one at both infimum and supremum in Section 4 . In Theorem 3.1 , conditionally on I_{T}=a , the laws of the pre- H_{I} and post- H_{I} processes as well as their supremums and maximum drawdowns are characterized in terms of the scale functions. We identify the law of the pre- H_{I} process under conditioning H_{I}<H_{S},I_{T}=a,S_{T}=b in Proposition 4.1 . Finally in Theorem 4.3 , we find the conditional distribution of maximum drawdown of the pre-infimum process and recall for the intermediate and the post-supremum processes under the conditioning of Proposition 4.1 . 2 Preliminaries Throughout, X=(X_{t})_{t\geq 0} is a spectrally negative Lévy process with unbounded variation on a complete filtered probability space (\Omega,\mathcal{F},(\mathcal{F}_{t})_{t\geq 0},\mathbb{P}) . Its Laplace exponent \psi is defined by \mathbb{E}_{x}\big[e^{\lambda(X_{t}-X_{0})}\big]=e^{t\psi(\lambda)},\qquad\lambda\geq 0, on the set where the expectation is finite where \Psi(\lambda)=\frac{\sigma^{2}}{2}\lambda^{2}+\mu\lambda+\int_{(0,\infty)}\left[e^{-\lambda y}-1+\lambda y\right]\Pi(\mathrm{d}y),\lambda\geq 0 with a Lévy measure \Pi of -X satisfying \int_{(0,\infty)}\left(y\wedge y^{2}\right)\Pi(\mathrm{d}y)<\infty , drift parameter \mu\in\mathbb{R} and diffusion parameter \sigma\geq 0 . For q\geq 0 , let \Phi(q)=\sup\{\lambda\geq 0:\psi(\lambda)=q\}. The q -scale function W^{(q)}:\mathbb{R}\to[0,\infty) is characterized by W^{(q)}(x)=0 for x<0 and \int_{0}^{\infty}e^{-\lambda x}W^{(q)}(x)\,\mathrm{d}x=\frac{1}{\psi(\lambda)-q},\qquad\lambda>\Phi(q). The associated second q -scale function is found as Z^{(q)}(x)=1+q\int_{0}^{x}W^{(q)}(y)\,\mathrm{d}y,\qquad x\geq 0. Since X is assumed to have unbounded variation, the origin is regular for both half-lines and the scale function W^{(q)} is continuously differentiable on (0,\infty) [ 15 , Lem.2.4]