Nvidia and other tech stocks have a wide "dispersion" of price targets. Why that could be a problem for investors. By Elijah Nicholson-Messmer
Nvidia may no longer be the hottest artificial-intelligence trade, but Wall Street analysts are still fans. Among dozens of analysts tracked by FactSet, the average price target is $314, about 43% above recent prices around $219, according to consensus estimates.
But Nvidia's average price target obscures the fact that there's wide disagreement on the stock's fair value, ranging from a low of $180 to a high of $743.
That disparity isn't unique to Nvidia: It's a broader market phenomenon, and it has implications for investors. The S&P 500 index is seeing its highest level of price target dispersion -- a measure of disagreement among analysts on a stock's value -- in a decade, a Barron's analysis found.
The average gap between analysts' highest and lowest price targets now stands at 62.5% of the consensus target. For a stock with a $100 target, for instance, the spread between the highest and lowest forecasts would be $62.50 -- roughly $131 on the high end and $69 on the low end.
The S&P 500's 62.5% dispersion is well above the average of 43% since 2016.
Much of the S&P 500's dispersion is in tech, led by companies like Nvidia, Microsoft, and Broadcom. Dispersion for those stocks is well above average since the start of the year.
But this isn't just an AI story. On an equal-weighted basis, the S&P 500 had one of the highest dispersion levels in July of any month over the past decade. The typical S&P 500 stock had a dispersion of 44% in July, about four percentage points higher than average.
High dispersion can be a red flag. A 2024 study by researchers at the Yale School of Management and Indiana University found that consensus price targets become less reliable when analysts strongly disagree. Stocks with highly dispersed targets have historically delivered weaker returns, even when the average looks attractive.
That finding could be the basis for a trading strategy, says Frank Zhang, an accounting professor at Yale who co-authored the research. By shorting (betting against) stocks with high dispersion and owning stocks with low dispersion, a hypothetical portfolio generated an 11% annualized return from 1999 to 2020. That increased to a 19%-28% risk-adjusted return when using monthly rebalancing.
Dispersion can stem from real differences in opinion -- or from analysts' incentives. Some analysts genuinely disagree on a stock's value, while others hold off on cutting price targets to maintain good relationships with the companies they cover. High dispersion may also reflect the fact that consensus targets don't immediately incorporate changing views on a stock. Consider the case of Paycom Software, a payroll and human-resource software provider.
For the S&P 500, Zhang says, the index's high dispersion reflects a deep divide over AI. "Some people...are very optimistic about these AI stocks, " Zhang says. "Other people think that we may have a bubble, and you would expect the stock price to go down."
Regardless of who turns out to be correct, the disagreement is bearish, Zhang says. "If dispersion is very high, the stock is likely to be overvalued."
Consensus estimates that may be artificially high can also have a snowball effect, pushing stocks into even more overvalued territory.
When analysts are sharply divided, investors who agree with the bullish forecasts can readily buy shares and push the price higher. But investors who think the stock is overvalued face greater costs and constraints when shorting, or betting against it. That imbalance can give the bullish view more influence on the stock's price, helping to explain why highly dispersed stocks are prone to overvaluation.
Some sectors don't have high dispersion. Utilities and financials are relatively low, reflecting more agreement in their stock values.
Tracking dispersion isn't easy. It's rarely used as an investment indicator, and brokerage platforms typically don't provide it.
Zhang and his co-authors calculate dispersion using every individual analyst target for a stock. Most investors don't have access to that level of detail. Zhang says a practical proxy is to compare the spread between the highest and lowest targets relative to consensus. The wider the spread, the greater the disagreement among analysts.
As a general rule, Zhang says dispersion levels above 30% start to raise red flags. In July, the 10 least divided stocks in the S&P 500 had dispersion levels of 7% to 13%. Dispersion among the most divided stocks ranged from 117% to 273%.
Since the start of the year, the typical stock in the least divided group returned 16% through the end of July, and all 10 stocks rose over that time. Within the most divided group, the typical stock declined 10%.
Trading on dispersion remains a niche approach, but some strategists argue it can be fruitful.
Jack Gang, a former quantitative trading researcher and founder of the Alpha Engine Report, developed a strategy based on his own formula, which favors recent price targets while penalizing stale ones. In back tests against the S&P 500 and the Nasdaq 100, Gang says his strategy delivered returns of 17% to 20% a year, compared with the 10% to 12% returns from major indexes.
Gang says his approach doesn't require shorting stocks. It's fundamentally a "value strategy," he says. The hitch: While the strategy may outperform, it does so with slightly more volatility than passive index investing.
Methodology for S&P 500 Dispersion Data
For each company each month, dispersion is the difference between the highest and lowest analyst price target, divided by the consensus target: (High - Low) ÷ Consensus.
Index membership is set annually. Each year uses that year's list of S&P 500 constituents, so the history isn't distorted by which companies happen to be in the index today. Averages are weighted by index weight, not by company count. Each company's dispersion counts in proportion to its share of the S&P 500, so the figures describe the index as investors actually hold it rather than the typical constituent.
The 10 most- and least-divided companies are ranked on the most recent month's reading, ignoring whether the disagreement is new or longstanding.
Comments