Buying New Lows in a Tech Selloff: A Historical Test of the 'Never Buy a New Low' Rule

The recent plunge in software stocks is a targeted correction, not a broad market panic. It's a fear-driven selloff sparked by a specific technological threat. The immediate trigger was Anthropic's launch of a legal AI plug-in, a move that served as a fresh reminder of how advances in artificial intelligence could directly disrupt the business models of companies in the legal and professional services sectors. This single event accelerated a months-long decline, turning investor anxiety into a full-blown sell-off.

The scope of the decline is global and severe. It's not just a few names; it's a sector-wide rout. In Europe, RELX and Wolters Kluwer hit new lows, while the London Stock Exchange Group fell another 6% after a massive drop the day before. The pressure spread to Asia, where Japanese software and systems developers like NEC and Fujitsu slid between 7% and 11%. Even Indian IT exporters were dragged down. This coordinated global move shows the threat is perceived as systemic, not isolated.

This pattern echoes past tech corrections where a specific business model was seen as vulnerable. The dot-com bubble, for instance, saw outsized declines in companies built on internet infrastructure, as the market reassessed their long-term viability. Today, the threat is AI disruption to software and IT services, creating a similar dynamic of "guilty until proven innocent." As one analyst noted, investors are choosing to shun the software market altogether, leaving no place to hide. The core question for investors now is whether this fear-driven sell-off creates a buying opportunity or if the historical rule against "buying new lows" holds true in the face of such a fundamental threat.

The Historical Rule: Why 'Never Buy a New Low' is a Wall Street Career-Saver

The instinct to buy when prices hit new lows is a powerful one, but it is a trap that has cost many careers. The academic research backs this caution. A study cited by DataTrek's Nicholas Colas and Jessica Rabe analyzed momentum across U.S. and international stocks from 1990 to 2024. It found a clear pattern: being long names with positive price momentum and short those with weak momentum yielded positive average annual returns. In other words, stocks that are falling tend to keep falling. This momentum effect is a dangerous force to ignore.

The warning is not just academic; it is a hard-earned lesson from the front lines. Colas, drawing on a plus-30-year Wall Street career, stated bluntly that the temptation to pick a market bottom carries serious career risk. He noted that he has seen more friends fired or lose clients because they tried to pick a bottom than any other single cause. For professionals, breaking the "never buy a new low" rule is a career-ending mistake. This isn't about missing a single trade; it's about the long-term damage of being wrong when the market is in a strong downtrend.

This historical rule gains added weight against the backdrop of a broader market shift. The current "Great Rotation" is a structural move away from the concentrated, vulnerable tech names now hitting lows. In January, small-caps dramatically outperformed mega-caps, with the Russell 2000 surging while the S&P 500 lagged. Value stocks also left growth in the dust. This rotation signals a market-wide flight from the richly valued, high-momentum names that are now under pressure. Attempting to buy the absolute lows in this sector, while the broader market is rotating out, is fighting the tide of momentum and capital flows.

The bottom line is that the historical data and expert warnings form a powerful consensus. Momentum is a persistent force, and the career risk of fighting it is high. In a market where the rotation is clear and the threat to software business models is real, the safer play is often to wait for price stability and a shift in the momentum narrative before stepping in.

Valuation, Catalysts, and the Path Forward

The question of mispricing is not a simple yes or no. The sector-wide selloff has created a broad discount, but the historical pattern of momentum suggests that discounting is not the same as value. The key is to look past the headline lows and identify companies where the fundamental threat is lower. As one analysis notes, companies with a low risk of AI displacement are being lumped together with firms that may face genuine disruption. This is where the historical test of the "never buy a new low" rule becomes a filter. The rule is a career-saver for the sector as a whole, but it may be broken for specific names with durable moats. Infrastructure and cybersecurity firms, for example, are seen as having strong pricing power and even a tailwind from AI, making them candidates for selective buying.

The immediate catalyst to watch is earnings season. Any stumble by enterprise software companies will directly amplify the AI threat narrative and likely trigger another wave of selling. The setup is clear: Any stumble by enterprise software companies this earnings season will make the "AI is eating software" argument even louder. This is the next test of the market's fear. For companies already down sharply, a weak quarter could accelerate the decline, while a resilient one might begin to challenge the narrative. The path forward hinges on this data.

A broader reversal signal would be a rotation back into tech if the overall market stabilizes. The recent divergence is telling. While the Nasdaq 100 fell to a 1.5-week low, the Dow Jones Industrials posted a new record high. This split shows that the broader market can be strong even as specific, vulnerable tech names are punished. A sustained move higher in the Dow and S&P 500, driven by economic strength or a shift in Fed expectations, could eventually pull capital back toward the tech sector. That rotation would be the clearest sign that the sector-specific fear is being priced out, validating a contrarian entry for the right companies. Until then, the historical momentum and the current catalysts point to continued volatility and a need for extreme selectivity.