Modeling Maximum drawdown Records with Piecewise Deterministic Markov Processe in Capital Markets
Rolando Rubilar-Torrealba, Lisandro Fermin, Soledad Torres · 2025
Read Modeling Maximum drawdown Records with Piecewise Deterministic Markov Processe in Capital Markets by Rolando Rubilar-Torrealba, Lisandro Fermin, Soledad Torres (2025) 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-2503-23221.txt.
Introduction
Rolando Rubilar-Torrealba, Lisandro Fermin, and Soledad Torres modeled stock market drops in a 2025 paper. A drawdown is the fall from a past peak to a later low, and the largest such fall is a key risk gauge for traders. The authors track the chain of record values of that largest fall over time.
They model the chain with a process that flows in a smooth way but jumps at random times. From that model they derive the mean and spread of the record chain. They also build ways to fit the model values from data and test them in computer runs. A real market case shows the full steps, from data to fitted risk figures.
Catalog record
| CREATOR | Rolando Rubilar-Torrealba, Lisandro Fermin, Soledad Torres |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-23221
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Modeling Maximum drawdown Records with Piecewise Deterministic Markov Processe in Capital Markets Rolando Rubilar-Torrealba, Lisandro Fermin, Soledad Torres Abstract We propose to model the records of the maximum Drawdown in capital markets by means a Piecewise Deterministic Markov Process (PDMP). We derive statistical results such as the mean and variance that describes the sequence of maximum Drawdown records. In addition, we developed a simulation study and techniques for estimating the parameters governing the stochastic process, using a practical example in the capital market to illustrate the procedure. August 24, 2026 K eywords Records process \cdot financial time series \cdot Piecewise Deterministic Markov Models \cdot risk management 1 Introduction One of the financial risk management strategies is to regulate the maximum losses that can occur over a fixed time horizon or maximum drawdown. As the level of risk is dynamic, the incentive structure of managers also changes in the face of the new maximum drawdown ( [ Pospisil and Vecer(2010) ] ), leading to changes in investment decisions. The management of financial risk by fund managers typically involves the application of a set of rules defining the behaviour of investments in the context of events that shape financial market prices (cited in references such as [ Leal and de Melo Mendes(2005) , Zhang(2009) , Power(2004) ] ). Such events can be conditioned by observed losses in the capital markets. The strategies associated with the maximum Drawdown of a financial asset as a basic element for risk management generate a direct link between the price structure of financial assets with the statistical theory of extreme values, more specifically with the theory of records ( [ Embrechts et al.(2013)Embrechts, Klüppelberg and Mikosch , Gomes and Guillou(2015) , Landriault et al.(2021)Landriault, Li and Lkabous , Li and Zhou(2022) ] ), assuming that the financial information of the financial asset is public knowledge for all members of the capital market. In this article we link Piecewise Deterministic Markov Process (PDMP) processes with record theory for the particular case of a risk management process known as maximum Drawdown, which is widely used in the financial literature, but can be easily extended to other disciplines. This research aims to carry out a statistical and mathematical analysis of the characteristics of the limit distribution of the process that defines the maximum Drawdown records, based on the PDMP methodology, and to propose methods for estimating the parameters that determine the process. The PDMP are processes which was originally developed by [ Davis(1993) ] and have had an important relevance in determining phenomena in different areas of science. These processes are characterised by having a certain number of states that change randomly and by having a deterministic part that evolves in each of the states of the process. The general characteristics of PDMP processes allow to approach areas that have not been deeply developed in the literature and to generate new research opportunities. In particular, record theory allows us to understand processes that evolve over time and to understand the asymptotic behaviour of certain phenomena, where record is the largest (smallest) value in a sequence of values that are observed over time. Among recent developments, we can cite [ Rudnicki and Tyran-Kamińska(2015) ] who uses PDMP techniques to set up an analysis framework for biological systems, cell life cycle and other biological applications. Another interesting development is by [ Kouretas et al.(2006)Kouretas, Koutoumpas and Lygeros ] who model biological networks for the study of antibiotic production. Similarly we can see developments in other areas of science such as physics, economics, among others (see [ Lin and Buchler(2018) , Schäl(1998) , Schmidli(2010) ] ). The results of the paper show a description of the analytical mean and variance of records obtained from the PDMP process in the context of the maximum Drawdown approach. In addition, simulation techniques and estimation of the parameters of the processes that describes the model were developed, providing important tools for potential applications in different areas of science. The use of a variety of analytical tools allows for a comprehensive understanding of the stochastic process that governs the sequence of jumps that define the maximum drawdown. Markov chains are a fundamental tool for analysing the evolution of financial series, which have been extensively employed in economic and financial literature ( [ Mendoza-Arriaga and Linetsky(2016) , Cui et al.(2018)Cui, Lee and Liu , Cui et al.(2019)Cui, Kirkby and Nguyen ] ). This paper is organized as follows. Section 1 presents the Introduction, the Model and preliminaries is presented in the Section 2. The Characterisation of the process is show in Section 3. The Section 4 corresponds to the Simulation and estimation of the process. In Section 5 we apply the methodology developed for an applied case, and the Section 6 corresponds to the Conclusions of this research. 2 Model and preliminaries The classical theory of records consider a time series X_{0},X_{1},\ldots,X_{n} of random variables that can reach a maximum (minimum) value in a period of time. We define an upper and lower record as X_{n}>\text{max}\left\{X_{0},X_{1},\ldots,X_{n-1}\right\}, X_{n}<\text{min}\left\{X_{0},X_{1},\ldots,X_{n-1}\right\}. Maximum Drawdown in T time can be informally defined as the largest drop from a peak of the time series to the smallest value of a valley, which implies reaching a new record in that time. Following [ Magdon-Ismail and Atiya(2004) ] , we define the maximum Drawdown process as a process driven by a Brownian process that indicates the maximum drop observed in a period in [0,T] . Let X(t) , 0\leq t\leq T , a Brownian motion with drift given by X(t)=\sigma W(t)+\mu t , where \mu\in\mathbb{R} is the trend and \sigma\geq 0 is the diffusion parameter. The mathematical definition of maximum Drawdown ( D ) is D(T;\mu,\sigma)=\sup_{t\in[0,T]}\left[\sup_{s\in[0,t]}X(s)-X(t)\right], (1) with 0\leq s\leq t\leq T . This definition allows to characterise the maximum drop of a time series when considering a fixed value of T>0 . However, we are interested in the process for each time instant over the records that it can reach at the maximum Drawdown . For this stochastic process we consider the following definition of the occurrence times of a records in the process, such that T_{i}=\text{inf}\{t>0:D(t)>D(T_{i-1})\wedge\exists s>t,D(s)-D(t)=0\}, (2) with T_{0}=0 and i\in\mathbb{N} , where T_{i} represents the ocurrence of the i-th record and D(\cdot) corresponds to the maximum drawdown random variable defined in ( 1 ), which present a new record in the time series and, therefore, D(T_{i}) and T_{i} are random variables that define a stochastic process of drawdown records that must be characterised in terms of their time evolution and distribution. The observation of the occurrence of new records allows modeling a stochastic process that alternates in different states and thus observe the long-term behaviour of the records that are achieved. In the following subsection a model is developed that allows to characterise the evolution of the records as a stochastic process. 2.1 Piecewise Deterministic Markov Process In our model we consider a Piecewise Deterministic Markov Process (PDMP) as the main tool that allows us to characterise the stochastic process. PDMP processes were first introduced by [ Davis(1993) ] and which has been popularized in recent years [ Azais and Bouguet(2018) ] , becoming a powerful tool for working with stochastic differential equations, specifically