Could We Predict the Financial Crises of 1987, 2008 and 2025?

Market crashes often seem to strike like lightning, leaving devastation in their wake. From Black Monday in 1987 to the 2008 financial crisis, these events remind us how fragile economies can be. But what if the next crisis could be predicted, not by hedge fund managers in corner offices, but by algorithms sifting through data?

Crisis Detection Beyond Traditional Metrics

Traditional predictors of market crashes include metrics like high debt-to-income ratios, overleveraged banks, and soaring asset prices. AI has the potential to analyse these and more significant indicators in real time.

What about features that aren’t in the textbooks? Research suggests that unconventional data, like social media sentiment, credit card delinquencies, and even nighttime satellite images of urban lighting, can serve as leading indicators of economic distress. For instance:

Social Media Sentiment: Platforms like Twitter and Reddit can offer real-time insights into consumer confidence. Sudden surges in terms like "job loss" or "default" might indicate at a forthcoming downturn.

Electricity Consumption: In times of economic struggle, urban areas often cut back on energy usage. AI can analyse changes in nighttime satellite imagery to detect economic contractions, especially in developing regions. In fact, in my neighbourhood in Belgium, they have cut back on city lights during the summer at night and some hours in the winter as well, possibly reflecting either financial responsibility by the government or signs of tighter budgeting amidst a crisis.

Retail Spending Trends: By analysing anonymised transaction data, AI can track shifts in discretionary spending, a key signal of consumer confidence.

Lessons from Past Crashes: 1987 and 2008

Financial crashes often reveal systemic weaknesses, and the 1987 and 2008 crises provide some important lessons.

In 1987, Black Monday saw global stock markets plummet, with the Dow Jones dropping 22.6% in a single day. The cause remains debated, but automated trading likely played a role. Early algorithmic programs, designed to limit losses, triggered a feedback loop of panic selling. While no clear bubble preceded the crash, today’s AI might have flagged the cascading sell orders and comprehensive vulnerabilities, potentially averting this disaster.

The 2008 financial crisis, by contrast, was rooted in clear structural issues. The collapse of the U.S. housing market exposed reckless subprime lending and the overvaluation of mortgage-backed securities. As homeowners defaulted, financial institutions heavily invested in these assets faced ruin, leading to a global recession. If AI had been more advanced at the time, it might have detected the unsustainable rise in defaults and simulated the broader impact, providing earlier warnings.

Michael Burry vs Machine Learning

Michael Burry, famously portrayed by Christian Bale in the movie The Big Short, is best known for predicting the 2008 financial crisis. A hedge fund manager with an uncanny ability to see what others overlooked, Burry accurately analysed the subprime mortgage market, examining thousands of mortgage bond tranches to uncover hidden vulnerabilities. While most investors dismissed early signs of trouble, Burry recognised the unsustainable foundation of these loans and made a bold bet against the housing market, earning his investors billions when the crisis hit.

Back in 2008, AI wasn’t advanced enough to replicate this kind of deep analysis. Burry’s work involved an almost obsessive attention to detail, coupled with a human ability to interpret not just numbers, but context, something machines struggled with at the time.

Fast-forward to today, as AI’s analytical capabilities have dramatically improved. Machine learning models can now sift through massive datasets, analysing millions of variables at lightning speed. What took Burry weeks or months to uncover in mortgage bond tranches, modern AI might process in minutes. This technological leap means AI could potentially predict vulnerabilities not only in real estate but also in corporate debt, stock markets, or even emerging fields like cryptocurrencies.

However, even with all its computational power, AI has a critical blind spot, which is analysing context. Burry’s brilliance wasn’t just in identifying patterns; it was in understanding their broader implications. For instance, if a machine learning algorithm detected a rise in mortgage defaults in 2006, would it have grasped the systemic risks that Burry foresaw? Likely not, at least not without human oversight to guide the interpretation of those patterns.

Can AI Avoid the Next Crisis, or Make It Worse?

The question isn’t just whether AI can predict the next crash but whether it can prevent one. And here’s where the debate heats up.

The Optimistic View

If AI systems were widely adopted, they could provide early warnings to governments, financial institutions, and individuals. Central banks could take preemptive action, companies could strengthen their balance sheets, and households could tighten budgets. AI could even enforce accountability by exposing risky financial practices before they spiral out of control.

The Pessimistic View

However, there’s a darker side. High-frequency trading algorithms have already shown how AI can increase volatility. If market players rely too heavily on AI, it could create feedback loops, where small market movements are amplified into larger crises, a bit like on Black Monday. Imagine a scenario where AI predicts a downturn, leading to mass sell-offs that trigger the very crisis it sought to avoid.

From Global Markets to Personal Finance

The promise of AI isn’t just about predicting the next big crash, it’s about helping individuals and institutions prepare for it. Imagine a future where AI tools are seamlessly integrated into our financial lives, constantly monitoring global and personal economic conditions, alerting us to risks, and offering actionable advice tailored to our circumstances.

On a global scale, AI could act as a vigilant watchdog, analysing a vast array of economic signals to identify potential recessions. For example, health data could serve as an early warning system; studies suggest that increased healthcare spending often correlates with economic downturns, as families focus on essentials. Similarly, crime rates, which tend to rise during periods of economic hardship, could be another data point for AI to track and interpret. Even education enrolments could provide insight, rising numbers in vocational or community colleges might signal increasing unemployment as people retrain for new career paths.

On an individual level, AI has the potential to function as a personal financial advisor, helping people become recession-ready. Think of it as a financial Fitbit, offering real-time insights and adjustments to keep your financial health on track. For instance, AI-powered tools could dynamically adjust your budget based on economic forecasts, encouraging you to save more during times of increased risk. Investment advice could also become more proactive, with AI recommending shifts toward safer assets like bonds or commodities when it detects signs of an impending downturn. Furthermore, AI could analyse your spending habits to identify areas of overspending, helping you build a financial buffer for challenging times ahead.

Could 2025 Be the Next Year in the History Books?

The financial world thrives on speculation, and while no one can pinpoint the exact timing of the next crash, several signs suggest that storm clouds are gathering. Some experts predict a major correction in 2025, citing rising corporate debt, tightening monetary policies, and escalating administrative tensions. These factors echo warning signs from past crises and raise questions about the stability of today’s economic landscape.

One of the most concerning areas is corporate debt. During the pandemic, low interest rates encouraged companies to borrow heavily. Now, with central banks raising rates to combat inflation, the cost of servicing that debt has risen dramatically. This is particularly dangerous for highly leveraged sectors like real estate and technology. Commercial real estate, for instance, faces declining demand as remote work reshapes urban office needs, potentially leading to widespread defaults and ripple effects across the economy, a scenario scarily similar to 2008.

Meanwhile, the tech sector shows signs of vulnerability. Valuations for many tech companies, particularly in AI and startups, have surged to unsustainable levels, drawing comparisons to the dot-com bubble of the late 1990s. Back then, overhyped technology stocks collapsed under the weight of inflated expectations, triggering a market crash. Today’s AI boom, while transformative, risks repeating this pattern if investors discover that these companies cannot deliver on their lofty promises.

Unconventional Indicators and Historical Parallels

Beyond traditional metrics, unconventional indicators are also flashing red. The "unclaimed corpses" metric, for example, is a haunting reflection of economic despair: when families can’t afford funeral costs, it often signals widespread financial hardship, as was seen during the Great Recession. Another curious yet telling sign is the "mosquito correlation". During economic downturns, public health budgets often shrink, leading to neglected infrastructure and surges in mosquito-borne diseases. These odd metrics, while unconventional, reflect the human cost of financial instability and could serve as early warning signs that mainstream analyses might miss.

Drawing parallels to historical crashes only deepens the sense of urgency. In 2008, the collapse of the housing market, driven by reckless lending and unchecked speculation, sent shockwaves through the global economy. Today, rising corporate debt, speculative bubbles in technology, and a fragile global economy, strained by inflation, supply chain disruptions, and geopolitical instability, present a dangerous combination of risks.

The sectors most at risk in a potential crash include real estate, particularly commercial real estate, which faces long-term structural challenges; technology, with its soaring valuations; and high-yield corporate bonds, as rising interest rates put pressure on companies with weak financial health. While AI may help identify and mitigate some risks, history reminds us that models alone cannot prevent crises driven by human behaviour, policy mistakes, or unexpected shocks.

AI and the Human Touch to Avoid a Financial Heart Break

AI has the potential to revolutionise financial forecasting, offering unprecedented insights into when and where the next crisis might strike. As we move forward, the challenge will be balancing AI’s capabilities with human intuition. The hope is that together, we can not only predict the next crisis but also build systems resilient enough to withstand it. And perhaps, just perhaps, avoid the heart break that comes with it.