Part 4: The Great Disconnect: Why We Need Hidden Markov Models to Navigate Our Post-Market Economy

In Part 3 of this series, we examined how shifts in technology and market behavior are reshaping financial strategy within an increasingly unpredictable economic landscape. Today, we turn to the deeper structural question beneath that transformation — how we model an economy that no longer behaves as a single, coherent system, but as a set of overlapping and often conflicting regimes. If the old models can no longer explain the world we’re in, what tools will?

Last week, I received two messages that perfectly encapsulate the economic schizophrenia of our times. The first: a father of young children, bewildered by grocery prices that have doubled, spring break flights that have tripled, and prescription costs that have quintupled. The second: a seasoned investment banker who matter-of-factly explains that we no longer live in a market-based economy—we live in a liquidity-drugged simulation where equity markets exist primarily to generate return streams, not capitalize companies.

Both are right. And both point to why our traditional economic models are failing us spectacularly.

The Regime We’re Actually In

At the FDIC, we spent considerable time modeling systemic risks that existed outside conventional frameworks. Now, as CEO of Frontier Foundry, I’m applying those same principles to a much larger question: How do we model an economy where the observable data (market performance, employment statistics, inflation indices) tells a completely different story than lived experience (or how to analyze historic, terrible data that was not captured in a way that allows for logical analysis)?

The answer lies in understanding we’re not in one regime—we’re operating in multiple simultaneously. Traditional models are unable to account for this complexity, treating each regime as mutually exclusive and failing to recognize their dynamic and constantly changing applications to different markets. For example, semiconductors, as an industry, is behaving radically different than healthcare. The former is in, charitably, a bubble based on over-the-top capital expenditures, contributing to over 90% of GDP growth so far in 2025, while the latter is a money printing machine based on an older recurring revenue model with a thinner bottom line. According to traditional models, both exist in the same regime, yet are affected in vastly different ways.

The impacts of a multi-regime system act as hidden currents beneath the surface of today’s economy — forces we can sense but not directly observe. Hidden Markov Models (HMMs) offer a way to map those unseen dynamics, translating what appears to be chaos into probabilistic structure.

Unlike conventional economic models that assume linear relationships and observable states, HMMs acknowledge that the true economic “state” is hidden from direct observation. What we see — market prices, economic indicators, policy announcements, employment data — are merely emissions from underlying structural realities we can only infer probabilistically.

Three Hidden States, One Visible Chaos

Consider our current economic reality through an HMM lens with three primary hidden states:

State 1: The Liquidity Regime

• Characterized by asset inflation disconnected from productive capacity

• Observable emissions: Record stock prices, low unemployment, “transitory” inflation

• Probability transitions: High persistence, low exit probability without external shock

State 2: The Scarcity Regime

• Characterized by supply-demand imbalances in essential goods and services

• Observable emissions: Explosive costs in housing, healthcare, education, food

• Probability transitions: Self-reinforcing through hoarding behaviors and supply chain concentration

State 3: The Debasement Regime

• Characterized by currency manipulation to manage unsustainable debt loads

• Observable emissions: Divergence between official inflation metrics and purchasing power reality

• Probability transitions: Accelerating as debt-to-GDP ratios approach mathematical limits

HMMs don’t require us to pretend these states are mutually exclusive. They understand our economy operates in all three simultaneously, with different transition probabilities affecting different sectors and demographics.

The AI Multiplier Effect

Here’s where it gets interesting from an AI perspective. Artificial intelligence isn’t just another technological advancement—it’s a regime-change accelerator that’s amplifying the hidden state transitions in unpredictable ways.

AI is simultaneously:

• Deflationary in information processing, customer service, and certain manufacturing

• Inflationary in energy consumption, specialized hardware, and human expertise that can’t be automated

• Disruptive to labor markets in ways that don’t show up in employment statistics for 18-24 months

Traditional economic models treat technological change as an exogenous variable. HMMs allow us to model AI as both a state-transition catalyst and an emission-pattern disruptor. The result? We can begin to predict when AI advancement will shift us between the three regimes described above.

Geopolitical State Dependencies

The Ukraine conflict, U.S.-China tensions, and Middle East instability aren’t separate from our economic modeling—they’re state-dependent variables affecting transition probabilities between regimes.

In the Liquidity Regime, geopolitical instability drives “flight to quality” which reinforces asset bubbles.

In the Scarcity Regime, it creates supply shock multipliers.

In the Debasement Regime, it accelerates the weaponization of currency systems.

An HMM framework allows us to model these dependencies without assuming they’re linear or predictable. Instead, we can assign probability distributions to different geopolitical scenarios and their economic regime impacts.

Leave a comment

Building the Model: From Theory to Implementation

At Frontier Foundry, we’ve developed the HMM-based economic modeling tools that financial institutions and policy makers need to navigate regime uncertainty. The key components:

Observable Variables (Emissions):

• Traditional economic indicators (CPI, unemployment, GDP)

• Alternative data streams (satellite imagery, credit card spending, social media sentiment)

• Geopolitical event frequencies and intensities

• AI adoption and displacement metrics

Hidden States:

• Liquidity dependency levels

• Resource scarcity intensities

• Currency debasement velocities

• Geopolitical stability coefficients

Transition Probabilities:

• Policy intervention likelihoods

• External shock probabilities

• Technological disruption rates

• Social stability thresholds

The Path Forward

The father concerned about his family budget and the investment banker analyzing market distortions are both observing emissions from the same underlying reality. Traditional economics tells them they’re experiencing different phenomena. HMMs tell us they’re seeing different aspects of the same multi-regime system.

This isn’t academic theorizing. Banks need these models to manage systemic risk. Businesses need them for strategic planning. Individuals need them to make rational decisions about everything from career choices to retirement planning. Frontier Foundry understands that traditional models can no longer be relied upon in today’s complex environment, and we’re building the HMM solutions needed to navigate it. Those who fail to take advantage of these tools are at the mercy of an unpredictable economy, exposing themselves not only to financial risk, but potentially disastrous outcomes.

The era of pretending we live in a market-based economy with predictable relationships between inputs and outputs is over. The era of probabilistic regime modeling—enhanced by AI capabilities—is just beginning.

The question isn’t whether our economic reality will become more complex. It’s whether we’ll use the tools needed to navigate that complexity intelligently.

Or whether we’ll keep pretending that $1,800 spring break flights in a “low inflation” environment make sense.

This article was written by Sultan Meghji, CEO of Frontier Foundry. Visit his LinkedIn here.

To stay up to date with our work, visit our website, or follow us on LinkedIn, X, and Bluesky. To learn more about the services we offer, please visit our product page.

© 2025 Frontier Foundry. All rights reserved. This article and its contents are the exclusive intellectual property of Frontier Foundry. Unauthorized reproduction, distribution, republication, or derivative use — in whole or in part — without express written permission is strictly prohibited and may result in civil and criminal penalties under U.S. and international copyright law.

Frontier Foundry builds deterministic, secured AI for financial services, life sciences, and U.S. federal law enforcement. Our work spans AI governance aligned to the NIST AI RMF and the EU AI Act, post-quantum cryptographic agility, and privacy-first deployment patterns for organizations where getting the answer wrong carries regulatory, safety, or reputational consequences. Founded and led by Sultan Meghji — former inaugural Chief Innovation Officer of the FDIC.