March 2020: What the System Saw

January 2020. Markets were calm. The S&P had just closed at another all-time high. Volatility was low. The financial press was writing about whether the bull market could continue into its twelfth year. The mood was relaxed, almost lazy. People were planning vacations.

The system didn't have a mood. It read the data the same way it reads data every night: correlation structure, regime indicators, flow signals. On January 15th, 2020, the readings were unremarkable. Markets were behaving normally. Stocks moved on their own fundamentals. Sectors rotated at the pace sectors rotate. Nothing interesting.

By the first week of February, something shifted. Not in the headlines. In the correlations.

Correlation is a quiet signal. Nobody puts it on cable news. But it's one of the most reliable early warnings in markets. When stocks that normally move independently start moving together, it means something has changed beneath the surface. Individual stories stop mattering. The market starts trading as one organism.

The system noticed it the way a thermometer notices a fever. No interpretation. Just measurement. Cross-asset correlations were ticking up. Stocks that had no business moving in lockstep were beginning to drift in the same direction. Not dramatically. Just enough to register.

At this point, human beings were still arguing about whether a respiratory illness in Wuhan would affect quarterly earnings. The system didn't know about Wuhan. It doesn't read news. It reads prices, volumes, and the relationships between them.

By mid-February, the correlation signal had strengthened. And something else appeared: gold was catching a bid.

The early bid in gold is always the tell.

When sophisticated capital gets nervous, it doesn't sell stocks first. Selling stocks is loud. It moves prices. It attracts attention. Instead, the first move is quiet accumulation of safe havens. Gold. Long-dated treasuries. The Swiss franc. These are the positions you build before the storm, while you can still get in at reasonable prices.

Retail investors watch the stock market for signs of trouble. Professional money watches the safe haven markets for signs of quiet accumulation. Gold was moving before any headline justified it. That's the point. By the time the headline arrives, the positioning is already done.

The system doesn't think in terms of "nervous" or "cautious." It tracks relative flows. Capital was rotating, slowly but measurably, from risk assets toward defensive positions. The regime indicators were shifting from what we call "calm trending" toward "elevated uncertainty."

Late February. The S&P dropped a few percent in a single day. Headlines finally caught up: virus fears spreading, cases outside China, markets rattled. For most observers, this was the beginning. For the system, this was confirmation of something it had been tracking for weeks.

By this point, the regime had officially changed. The system's classification moved from the environment where it runs full equity exposure to the environment where it begins reallocating. Not all at once. Gradually. Reducing positions that perform poorly in stressed markets. Increasing allocation to strategies that perform well when correlations spike and volatility rises.

The first week of March was surreal. Markets swung wildly. Up three percent, down four percent, up two percent. The news cycle was accelerating: Italy was locking down, the WHO was issuing warnings, companies were canceling conferences. Every day brought a new reason to panic.

The system had been moving capital for two weeks.

Not because it predicted a pandemic. It had no opinion about pandemics. It saw correlations rising, safe havens catching bids, and regime indicators deteriorating. These are the same signals it reads in every drawdown, whether the cause is a virus, a credit crisis, or a trade war. The cause doesn't matter. The market's behavior does.

March 9th. The S&P fell seven percent in a single session. Oil crashed. Circuit breakers tripped for the first time since 1997. Trading desks were in chaos. Phone lines jammed. The VIX, Wall Street's fear gauge, spiked to levels not seen since the 2008 financial crisis.

By March 9th, the system's allocation had shifted substantially toward gold, bonds, and the mean-reversion strategies that thrive in high-volatility environments. Not because anyone told it to. Because regime change triggers reallocation, and regime change had been confirmed weeks earlier.

The thing about systematic trading is that the hard decisions happen before the crisis, not during it.

When March 9th arrived, there was nothing to decide. The decisions had already been made by the data in February. The system was simply executing the allocation that the regime dictated. No meetings. No debates about whether this was the bottom. No waiting to see what the Fed would do. Just the plan, running as designed.

March 12th. Another massive drop. Markets were now in free fall. The WHO declared a global pandemic. Travel bans. School closures. The financial system was being stress-tested in real time.

March 16th. The worst day. The S&P fell another twelve percent at the open. Circuit breakers halted trading for fifteen minutes. When trading resumed, the selling continued. Professional traders who had lived through 2008 were saying this felt worse. The speed of the decline was unlike anything in modern markets.

The system was positioned in gold, in bonds, and in strategies designed for exactly this kind of environment. It was running its mean-reversion engine at elevated allocation, because mean reversion performs best when fear is highest. When every stock is being sold regardless of quality, the gap between price and value widens. The system lives in that gap.

There's something important to note here. The system didn't avoid all losses. No system does. The equity portion of the portfolio was reduced, but it wasn't zero. Some positions were still exposed. The goal was never to perfectly time the exit and re-entry. The goal was to be structurally different in a crisis than in calm markets. Less exposed where exposure hurts. More exposed where dislocation creates opportunity.

By late March, something else began to show up in the data. Correlations, which had spiked to extreme levels, started to ease. Not back to normal. But the rate of change shifted. Gold was still elevated, but it had stopped accelerating. Bond yields had stabilized. The VIX was still high, but it was no longer making new highs.

The regime wasn't calm. But it was no longer deteriorating.

For human investors, late March 2020 was terrifying. The news was getting worse every day. Case counts climbing. Economies shutting down. Nobody knew when it would end. The emotional signal was overwhelmingly negative.

The data signal was different. Markets had priced in a catastrophe. The question was whether conditions were still getting worse at the same rate. They weren't. The system began nudging allocation back toward risk. Not aggressively. Not all at once. But the direction shifted.

This is the part that's hardest to explain. The system started buying equities in late March 2020, while the world was still in lockdown, while case counts were still rising, while the news was still apocalyptic. It bought because the data said the worst of the market's reaction had passed. Not the worst of the pandemic. The worst of the pricing.

Markets are not the economy. Markets are a discounting mechanism. They price in the future before the future arrives. By late March, the market had priced in a very dark scenario. And the rate of deterioration in market signals had slowed. That was enough.

April arrived. The market rallied. Then May. The rally continued. By June, the S&P had recovered a remarkable portion of its losses, even as the pandemic raged on. Investors who had sold in March were sitting on the sidelines, waiting for a clarity that would never come. The system had been adding exposure for weeks.

Looking back, the timeline is clean. Correlations shifted in early February. Gold caught a bid. Regime changed. Capital moved to defensive positions. Crisis hit. System was positioned. Crisis peaked. Data showed deceleration. Capital moved back to risk. Recovery unfolded.

But that clean narrative hides the most important truth about the process. None of it felt clean at the time. In real time, every day brought conflicting signals. The system doesn't experience conflict the way a person does. It weighs the evidence, updates its estimates, and acts. But the actions it took, buying equities in late March while the world was falling apart, would have been nearly impossible for a human to execute. Not intellectually. Emotionally.

This is the real argument for systematic investing. Not that it's smarter than human judgment. It often isn't. But it's consistent. It reads the same signals the same way every day, regardless of what the news says, regardless of what it feels like. It doesn't get scared. It doesn't get greedy. It doesn't call an emergency meeting. It just reads the data and acts.

There's a line we come back to when people ask us about March 2020. They want to know if the system predicted the crash. It didn't. They want to know if it predicted the recovery. It didn't do that either.

It read the signals. Correlations, regime, flow. The same signals it reads every night. In January, those signals said nothing interesting. In February, they said something was changing. In March, they said the change was severe. In late March, they said the severity was peaking.

The data was never panicking. Only the people were.