Boom-Busty Marketron as a Model of Market Dynamics: How Investor-Market Reflexivity Amplifies Non-Equilibrium
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Quantitative Finance. The day in quantitative finance: models of markets, risk and prices.
Ruth: We're looking at the paper: "Boom-Busty Marketron as a Model of Market Dynamics".
Warren: Here's the gist of the paper. The Boom-Busty Marketron (BBM) model introduces a multi-asset market framework where the demand of a single collective investor influences prices, creating an investor-market reflexive loop analogous to Soros’s theory.
Ruth: This structure results in a Non-Equilibrium Steady State (NESS), characterized by broken detailed balance and a circulating probability current between mispricing, demand, and volatility.
Warren: The model is calibrated to the S&P 500 from 1996 to 2024 using simulated moments, successfully reproducing stylized facts of returns alongside quantitative metrics, including cross-sectional ones.
Ruth: A central emergent feature is a boom-bust cycle whose frequency aligns with the S&P’s within one-sigma ranges.
Paper summary: Ruth: So wrapping up this paper, "Boom-Busty Marketron as a Model of Market Dynamics: How Investor-Market Reflexivity Amplifies Non-Equilibrium," the main thing is that they've introduced a model where investor demand moves prices and itself depends on market performance, creating an investor-market reflexive loop.
Warren: They claim this pushes the market into a non-equilibrium steady state with broken detailed balance, characterized by a probability current circulating between mispricing, demand, and volatility. This is amplified by the feedback loop itself.
Iris: The model’s construction is detailed across several sections, covering everything from the reflexivity mechanism to how they calibrated it against S andP five hundred moments over two periods <ref:2611.01185#pg1>. They also test excitability directly and measure circulating current and entropy production to confirm this non-equilibrium nature.
Ruth: What this means in simpler terms is that market perceptions and actions shape the market, and those changes feed back into those perceptions, creating a system that isn't just settled at a point but is constantly moving in a specific way, driven by noise pushing it over thresholds.
Warren: It provides a framework for seeing why we see boom-bust cycles align with specific statistical ranges and how individual firm dynamics get complicated by these memory effects. It moves beyond simple equilibrium assumptions.
Iris: The authors are essentially showing that even if participants are assumed to be identical, the way their beliefs and actions interact creates a persistent dynamic structure in the market, one that is measurable through things like entropy production.
Ruth: So it’s less about predicting an exact price point and more about understanding the underlying mechanism of how market dynamics are driven by this reflexive loop and how that leads to these specific, non-equilibrium behaviors we observe.
Conclusion: Ruth: So we’re wrapping up on this "Boom-Busty Marketron" paper, focusing on how investor demand feeding back into prices creates this whole reflexive loop that pushes markets out of their normal steady state.
Warren: Yeah, I mean the math shows this isn't just some smooth equilibrium thing; it’s a non-equilibrium steady state with a probability current circulating between mispricing and demand.
Iris: The authors used simulated moments from the S andP five hundred over two periods to calibrate this model, which is pretty interesting because they managed to reproduce the stylized facts of returns and even some cross-sectional metrics.
Ruth: They also showed that this structure produces cycles whose frequency matches what we see in the S andP’s within one sigma ranges, and they even generated things like default rates similar to investment grade issuers.
Warren: And they proved that the model has a measurable arrow of time, which is a big thing because it means the system has an inherent directionality to it.
Iris: The key finding for me is how they frame this as an excitable system; the index doesn't just oscillate forever, but it returns to rest after one big excursion because of certain accounting rules.
Ruth: So what does this mean for us? It suggests that the market isn't just reacting to news in a straightforward way; it’s driven by these deep structural feedback mechanisms that create cycles and persistence.
Warren: It implies that strategies built on equilibrium assumptions might miss the real drivers of risk, especially when you consider how investor behavior amplifies volatility through this loop.
Igor Halperin
q-fin.ST, nlin.AO, q-fin.GN
Submitted: 2026-10-31
Updated: 2026-10-31
Comments: 99 pages, 39 figures, 42 tables
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
Importance score: 81/100
The gist: The Boom-Busty Marketron (BBM) model introduces a multi-asset market framework where the demand of a single collective investor influences prices…
Key concepts
- Investor-Market Reflexive Loop
- This describes a feedback mechanism where collective investor demand affects market prices, and these prices in turn influence future investor demand. This loop is central to the model's dynamics, leading to non-equilibrium behavior rather than simple equilibrium.
- Non-Equilibrium Steady State (NESS)
- The market settles into a NESS where detailed balance is broken. Instead of a stable, unchanging state, the system maintains a circulating probability current between mispricing, investor demand, and volatility. This allows for continuous dynamics like boom-bust cycles.
- Structural Excitability
- A single firm exhibits structural excitability meaning its mispricing and demand memory move down a potential gradient. Recovery follows this fall for several months before noise pushes it past a saddle point into default, mimicking an excitable neural system.
Terminology
Summary
The Boom-Busty Marketron (BBM) model introduces a multi-asset market framework where the demand of a single collective investor influences prices, creating an investor-market reflexive loop analogous to Soros’s theory. p. 0 com
Abstract
We introduce the Boom-Busty Marketron, a model of a multi-asset market in which the
demand of a single collective investor moves prices and itself depends on market performance. This creates an investor-market reflexive loop in the spirit of Soros’s theory. This creates an investor-market reflexive loop in the spirit of Soros’s theory. The
market state is stationary without being at equilibrium: a non-equilibrium steady state
(NESS), with broken detailed balance and a probability current circulating between mispricing, demand and volatility. While stochastic volatility with the leverage effect already
breaks detailed balance and makes the market dynamics irreversible, the investor-market
loop pushes the market further away from equilibrium. Calibrated to the S&P 500 from
1996 to 2024 by simulated moments, the model reproduces the stylized facts of returns together with quantitative metrics of the market, including cross-sectional ones, and its return
and cycle statistics stay stationary for a century and a half on the 1996–2010 inputs. Calibrated to the S&P 500 from
1996 to 2024 by simulated moments, the model reproduces the stylized facts of returns together with quantitative metrics of the market, including cross-sectional ones, and its return
and cycle statistics stay stationary for a century and a half on the 1996–2010 inputs. Its
central emergent behavior is a boom-bust cycle, whose frequency meets the S&P’s within
one-sigma ranges. Its
central emergent behavior is a boom-bust cycle, whose frequency meets the S&P’s within
one-sigma ranges. The model also produces a default rate of the order of that of investmentgrade issuers, prices inside Fisher Black’s factor of two of value, the dumb-money effect of
Frazzini and Lamont, and a measurable arrow of time. The model also produces a default rate of the order of that of investmentgrade issuers, prices inside Fisher Black’s factor of two of value, the dumb-money effect of
Frazzini and Lamont, and a measurable arrow of time. Standard long-short factors built
from its cross section rotate through the cycle, and so do its industry groups along the
defensive axis.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
The BBM model extends Halperin and Itkin's Marketron by incorporating an investor-market reflexive loop. Each stock's log price (S t) is modeled as three components separated by the survival time of a shock: S t = F t + x t + eta t. This structure provides a hierarchy of timescales—a diffusion process for mispricing reversion over months, decay of transitory errors over days, and permanence of value—while allowing the dynamics to be written in stationary or nearly-stationary processes. The system is fundamentally bipartite: the capital market prices respond to latent investor demand (y t), and this demand is derived from observing past market performance. p. 4 stationary in law but with a probability current that never vanishes, and its entropy production
stays positive, giving it a measurable arrow of time. While irreversibility also arises in stochastic volatility models with a leverage effect, the investor-market reflexive loop amplifies it at the
calibrated parameters by up to two orders of magnitude: its entropy production, estimated in
a linear approximation, rises forty- to a hundredfold against a demand that does not respond
to prices, and four- to sixfold against a market without impact (Section 9.Where the paper says it
The model achieves stationarity through excitable dynamics returning to a NESS, manifesting as a boom-bust cycle that ends in one large excursion rather than continuous oscillation. p. 5 As a physics model, the BBM reaches stationarity by a route we have not seen proposed
in the financial literature: excitable dynamics returning to a non-equilibrium steady state, a
boom and the bust that ends it being one large excursion. A single firm has a threshold in its
own stochastic differential equations, but a firm that crosses it in earnest does not come back,
so it is a threshold unit with an absorbing far side rather than a spiking one. The index has the
behavior without the geometry, never approaching that threshold, its booms and busts being
large noise-driven excursions about one rest point while a probability current circulates in the
plane of mispricing against demand. What returns the index to rest is its own accounting,
the membership rule retiring the names that failed and seating replacements, so the index’s
excitability is behavioral rather than the dynamics of a system pushed past a critical point. A single name has the geometry: a rest state, a threshold that is a saddle’s stable
manifold, an all-or-none excursion past it, a short refractory period of a few months after its
own fall, and far past the threshold a region it does not come back from. The index has the
behavior without the geometry, never approaching that threshold, its booms and busts being
large noise-driven excursions about one rest point while a probability current circulates in the
plane of mispricing against demand. An index built from its names. The model simulates five hundred individual stocks, the
ten largest and the rest, and the index is built from them as the S&P is: each name has
a capitalization that compounds with its own return, the index return is the capitalizationweighted average of the names’ returns, and a name whose capitalization falls far enough behind
is removed and replaced, the removed firm being followed until it is sold or defaults.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
The irreversibility of the system stems from two sources: stochastic volatility with a leverage effect, which breaks detailed balance, and the coupling between investor demand and price, which amplifies this effect. p. 6 Stationarity is achieved not through an equilibrium but
through a non-equilibrium steady state (NESS), with probability currents that never vanish. The irreversibility has two sources, stochastic volatility with a leverage effect, which breaks
detailed balance on its own, and the coupling of the investor’s demand to the price, which
multiplies it by the factors given above (Section 9. While irreversibility also arises in stochastic volatility models with a leverage effect, the investor-market reflexive loop amplifies it at the
calibrated parameters by up to two orders of magnitude: its entropy production, estimated in
a linear approximation, rises forty- to a hundredfold against a demand that does not respond
to prices, and four- to sixfold against a market without impact (Section 9. The market produced by our model is noise-driven and excitable, in different senses at
its two levels. This creates an investor-market reflexive loop in the spirit of Soros’s theory. The
market state is stationary without being at equilibrium: a non-equilibrium steady state
(NESS), with broken detailed balance and a probability current circulating between mispricing, demand and volatility. Section 6 follows the
same cycle through money and firms, and Section 7 tests the market’s excitability directly, one
experiment at a time. Section 8 measures the circulating current and the entropy production
that show it to be a non-equilibrium steady state.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
The model's construction is detailed across several sections: Section 1.1 provides an overview; Section 2 elaborates on reflexivity and causal ambiguity; Section 3 details the model build; Section 4 covers calibration to the S&P 500 over two periods against alternatives; Sections 5 and 6 present results through the cycle, money flows, and firm dynamics respectively; Section 7 tests excitability directly; Section 8 measures circulating current and entropy production; Section 9 synthesizes findings and notes model failures; and Section 10 concludes. p. 5 Section 6 follows the
same cycle through money and firms, and Section 7 tests the market’s excitability directly, one
experiment at a time. Section 8 measures the circulating current and the entropy production
that show it to be a non-equilibrium steady state. The closer structural relative
is the growth-with-killing of Reed and Hughes [29], where killing proportionally growing units
turns lognormal sizes into a power law. We note that the BBM is not a fundamental model of the market at the level of a limit
order book with tick data, but a long-horizon phenomenological model of market dynamics at
the scale of a day and beyond. A model calibrated on twenty-nine years that can be run for
centuries, its slow relaxation to a settled state measured rather than assumed away, supplies
many independent histories with the observed moments, correlation structure and cadence of
busts. What such a simulator has to get right is not the next return but the frequency, depth
and spacing of the events a long-dated allocation must survive, which is why the cycle statistics
and the tail quantities of this paper are the ones we hold the model to.Where the paper says it
Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improve predictive modeling by incorporating a reflexive loop where
the investor-market loop pushes the market further away from equilibrium,
allowing for simulations that reproduce stylized facts and cycle statistics for extended periods without needing exogenous macroeconomic drivers. -
Enhance crisis forecasting capabilities by utilizing the model's structure as an excitable system, enabling the identification of critical points where
small perturbations decay into small fluctuations, and responds to any perturbation past a critical size with a sudden large excursion.
-
Develop robust risk management systems that account for non-equilibrium steady states by measuring
the circulating current and the entropy production that show it to be a non-equilibrium steady state,
providing metrics beyond traditional equilibrium concepts. -
Improve asset allocation strategies by using the investor's belief function, which is
linear in the momentum signal s(i)t and in the demand memory y(i)t,
to derive optimal holdings that areaiming at
a target allocation rather than just following a static mean-variance rule. -
Create more accurate correlation modeling by incorporating the
two-scale volatility
structure of market factors, which accounts for howthe common factor scales both the market and the industry terms of Eq. (37).
-
Refine index construction methods by using the I-BBM specification, which treats
the index as one marketron,
to generate untargeted diagnostics that test against multiple windows without relying on a single fitted asset model.
Abstract
We introduce the Boom-Busty Marketron, a model of a multi-asset market in which the demand of a single collective investor moves prices and itself depends on market performance. This creates an investor-market reflexive loop in the spirit of Soros's theory. The market state is stationary without being at equilibrium: a non-equilibrium steady state (NESS), with broken detailed balance and a probability current circulating between mispricing, demand and volatility. While stochastic volatility with the leverage effect already breaks detailed balance and makes the market dynamics irreversible, the investor-market loop pushes the market further away from equilibrium. Calibrated to the S&P 500 from 1996 to 2024 by simulated moments, the model reproduces the stylised facts of returns together with quantitative metrics of the market, including cross-sectional ones, and its return and cycle statistics stay stationary for a century and a half on the 1996--2010 inputs. Its central emergent behaviour is a boom-bust cycle, whose frequency meets the S&P's within one-sigma ranges. The model also produces a default rate of the order of that of investment-grade issuers, prices inside Fisher Black's factor of two of value, the dumb-money effect of Frazzini and Lamont, and a measurable arrow of time. Standard long-short factors built from its cross section rotate through the cycle, and so do its industry groups along the defensive axis.
Sources
- Six decades of the FitzHugh-Nagumo model: A guide through its spatio-temporal dynamics and influence across disciplines
- Observable Matrix Dynamics of Stocks
- Black was right: Price is within a factor 2 of Value
- SciPhy Reinforcement Learning for Portfolio Optimization
- Spectral properties of distance matrices
- Learning as Observable Matrix Dynamics: Diffusive Relaxations versus Phase Transitions
- Frustrated Dynamics of Distance Matrices
- The entropy production of stationary diffusions
- Sig-Wasserstein GANs for Time Series Generation
Related papers
- PanelGP: A Probabilistic Factor Model for Asset-Return Panels
- Physics-Informed Singular-Value Learning for Cross-Covariances Forecasting in Financial Markets
- A general technique for explicit quantile representations
- A Controlled Comparison of Deep Learning Architectures for Multi-Horizon Financial Forecasting: Evidence from 918 Experiments
- Who Are a Firm's Economic Peers? Mapping a Latent Information Field with Language Models
- Neural Realized Measures: Forecast-Supervised Intraday Aggregation for Volatility Forecasting