Dynamics of Periodic Bubbles and Crashes: Modeling Market Overheating and Panic Selling via Cubic Momentum

2026

Read Dynamics of Periodic Bubbles and Crashes: Modeling Market Overheating and Panic Selling via Cubic Momentum (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-2605-00854.txt.

Introduction

Naohiro Yoshida models the repeated rise and collapse of financial bubbles with a deliberately simple discrete-time simulation. Earlier work described explosive bubble statistics well, but capturing herd buying and panic selling in one framework has proved harder. Yoshida lets a cubic function of market momentum decide the balance of trading directions, so trend-following dominates as momentum grows.

Once momentum passes a critical level, the same mechanism flips the market into sudden crashes. Trading frequency also rises as that push builds, echoing the self-exciting nature of a Hawkes process. Even this minimal setup recreates surges in both cash flow and price, then sharp collapses, giving a compact account of cycles of overheating and panic.

Catalog record

YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE18 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.00854

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Dynamics of Periodic Bubbles and Crashes: Modeling Market Overheating and Panic Selling via Cubic Momentum Abstract This paper proposes a simple and parsimonious discrete-time simulation model to describe the endogenous formation and periodic collapse of financial bubbles. While existing literature has extensively explored the statistical properties of locally explosive bubble dynamics, capturing the micro-level interplay of investor herd behavior and panic selling within a unified framework remains a challenge. Our model addresses this by introducing a cubic function of market momentum to determine the balance of trading directions. This mechanism drives both trend-following behavior during the bubble phase and sudden market crashes when the momentum exceeds a critical threshold. Furthermore, inspired by the self-exciting nature of the Hawkes process, the model endogenizes“market frenzy” by linking trading frequency directly to the accumulated momentum. Simulation results demonstrate that this minimal setup successfully replicates the complex, nonlinear dynamics of bubbles, including simultaneous surges in liquidity and price, followed by dramatic crashes. Naohiro Yoshida Department of Economics, Keiai University 1-5-21, Anagawa, Inage, 263-8588, Chiba, Japan E-mail: [email protected] MSC 2020: 91G15 ; 91G80 ; 37N40 ; 60G55 JEL Classification: G01 ; G12 ; C63 Keywords: Financial bubbles ; Market crashes ; Nonlinear dynamics ; Hawkes process ; Momentum 1 Introduction Understanding the mechanisms behind the formation and subsequent rapid collapse (crash) of periodically occurring bubbles in financial markets is a critical issue in economics and financial engineering. To date, various mathematical and time-series models have been proposed to describe the dynamics of rational bubbles, wherein market prices deviate from economic fundamentals, e.g., [ 6 , 14 , 4 , 5 , 8 ] . A prominent early approach is the discrete-time model of stochastic bubbles proposed by [ 2 ] . This model captures the states of whether a bubble continues to grow or collapses, offering a more realistic approach compared to deterministic bubbles. However, stochastic bubbles suffered from a theoretical limitation: once a bubble completely collapses, it cannot restart (or grow) again. To overcome this limitation, [ 9 ] introduced the concept of “incompletely bursting bubbles.” By constructing a simple and tractable discrete-time model in which a bubble does not drop to absolute zero, [ 9 ] demonstrated that bubbles can resume growth following an incomplete collapse, bridging the gap between discrete-time models and continuous-time intrinsic bubbles. Furthermore, empirical literature on bubble detection—most notably the right-tailed unit root tests developed by [ 13 ] and [ 12 ] —has robustly confirmed that many economic and financial time series periodically exhibit locally explosive behavior (bubble formation) followed by crashes. Motivated by these empirical facts, nonlinear time-series models have gained attention for directly capturing such dynamics. Recently, [ 15 ] proposed a novel and simple time-series model known as the stochastic nonlinear autoregressive (SNAR) model. Their work demonstrates that, despite operating within a strictly stationary process framework, this model can accurately portray bubble dynamics characterized by periodic local explosions and subsequent collapses. These preceding studies have made significant contributions to the mathematical and statistical description of bubble periodicity and collapse mechanisms. However, there remains room for exploration in developing models that integrate the micro-level dynamics of actual market trading behaviors into an extremely simple and endogenous mechanism. In the field of econophysics, approaches such as [ 3 ] have demonstrated the utility of Langevin-type equations to describe stock market fluctuations and crashes driven by collective investor behavior. Yet, synthesizing these micro-behavioral insights—specifically,“investor herd psychology (trend following)” and “rapid panic selling triggered by fear during market overheating”—with the discrete-time periodic bubble literature remains a challenge. Therefore, this paper proposes a new discrete-time simulation model for financial bubbles, eliminating complex assumptions to prioritize simplicity as its defining feature. The primary innovation of this model is that the balance between investors’ buy orders and sell orders is determined by a cubic function of momentum M_{t} , which represents the accumulation of past price fluctuations, as employed in [ 11 ] . When the momentum is within a moderate range (indicative of trend-following behavior), positive feedback drives bubble formation. However, once the momentum exceeds a certain threshold and the market overheats, the properties of the cubic function endogenously generate strong selling pressure, triggering a dramatic bubble collapse. Furthermore, inspired by the Hawkes process [ 10 ] , the model incorporates a mechanism wherein the frequency of trading itself increases in response to momentum. This feature allows the model to naturally replicate the “market frenzy,” during which volatility and liquidity surge as the bubble forms. The remainder of this paper is structured as follows. Section 2 presents the formalization of the proposed simulation model, explaining the price-updating mechanism driven by the momentum-dependent cubic function and Section 3 provides the simulation result of this basic model. Section 4 analyzes the impact of various model parameters—such as the momentum threshold b , the memory decay rate r , and the baseline trading probability \Lambda —on the scale and periodicity of the bubbles. Finally, Section 5 provides the concluding remarks. 2 Model In this section, we present a simple discrete-time simulation model to describe the endogenous formation and collapse of financial bubbles. The model consists of two main mechanisms driven by market momentum: the acceleration of trading frequency (inspired by the Hawkes process) and the nonlinear transition of price directions using a cubic function (inspired by the potential in the Langevin equation). Let P_{t} be the price of an asset at time t for t=1,2,\dots,T . We define the market momentum M_{t} as the exponentially weighted moving average of past log-returns: M_{t}=\sum_{s=1}^{t-1}e^{-r(t-s)}(\log P_{s}-\log P_{s-1}), (1) where r>0 determines the decay rate of past price changes. A smaller r implies that investors have a longer memory of the historical trend. 2.1 Trading Frequency Driven by Momentum During a bubble, market enthusiasm typically leads to a surge in trading volume. To capture this phenomenon, we define the intensity of trading occurrence, \lambda_{t} , as a linear function of the momentum M_{t} : \lambda_{t}=\Lambda+kM_{t}, (2) where \Lambda is the baseline parameter for the minimum trading probability, and k>0 determines the sensitivity of the trading intensity to the momentum. Let \Delta N_{t}\in\{0,1\} be an indicator variable taking the value 1 if a trade occurs at time t , and 0 otherwise. We assume that \Delta N_{t} follows a Bernoulli distribution: \Delta N_{t}\sim Be(\Phi(\lambda_{t})), (3) where \Phi(\cdot) denotes the cumulative distribution function (CDF) of the standard normal distribution. Consequently, the cumulative number of trades up to time t is given by N_{t}=N_{t-1}+\Delta N_{t} . This formulation ensures that as the momentum builds up, the probability of trading increases, endogenously generating market liquidity. 2.2 Nonlinear Price Direction and Bubble Collapse The core innovation of our model lies in the determination of the price direction. We introduce a state variable x_{t} that governs the balance between buying and selling pressures. The evolution of x_{t} is defined by a cubic polynomial of the momentum: x_{t}=x_{t-1}+h(M_{t}-a)(M_{t}-b)(M_{t}-c), (4) where h>0 is a scaling parameter, and a,b,c ( a<b<c ) are the roots of the cubic function. This cubic specification models the nonlinear herd behavior of investors, analogous to the movement within a potential landscape in the Langevin equation (see, e.g., [ 3 ] ). When M_{t} is slightly positive ( 0<M_{t}<b ), the cubic

Read the complete text