Would you take a coin flip as an outcome?
I got to know a successful growth fund manager in the 1990s. Well. More than one! Louis Navellier was crushing it for technology investors using “modern portfolio theory” principles such as risk-adjusted return and how volatility presages tops and bottoms.
That by the way is not true in markets dominated by ETFs. When arbitrage is the principal price-setting mechanism, the absence of volatility often predicts turns because spreads between two assets expected to behave similarly have closed.
Anyway, this guy, not Louis but the other one, I can’t remember his name now, said, “I only have to be right 51% of the time.”
One percent is an edge.
So public companies and investors alike, what are the odds that a company beating the consensus expectation for topline and bottomline (revenue and earnings) results in the quarter will see its stock rise?
Sifma represents broker-dealers and asset managers controlling about 90% of the brokerage business in the US and about $75 trillion of managed assets. Every quarter, Sifma measures outcomes around earnings in the S&P 500. Sure, others do too.
Of the 305 components of the S&P 500 reporting thus far for calendar Q2 2026 (as of publication), about eighty percent beat on one or both of earnings and revenue. Yet more than half declined. In fact, 53% beating on revenues fell, 55% beating earnings expectations declined.
Worse than a coin flip.
Statistically, you’d be better off without the consensus benchmark. Because then it would be a coin flip.
The problem isn’t consensus. It’s that outcomes in the stock market around earnings are not determined in a statistically meaningful way by beating consensus.
Sure, maybe your guidance was weak or you came short of the so-called “whisper number” if such a thing exists. That’s for another blog, different day.
Right now, this is the point: Maybe we need a new measure.
I’d say the outcome reflects how “consensus” is something 80% of firms beat. How is that valuable? In Consumer Discretionary and Communications Services, firms more than doubled earnings expectations. Well then, how is that helpful for decisions?
It’s like a jobs report from the government. The variance is likely to be so great as to render the data almost meaningless.
Sifma says that in Q2 2026, 97% of Healthcare sector members of the S&P 500 had positive earnings surprises. How could it be a surprise? That’s like, “Everybody is a winner and gets a blue ribbon!”
Seems to me the surprise is the 3% that didn’t.
I’m aiming to make this entertaining! But should not a profession take a step back and examine measures meant to signal success that perform worse than random chance?
We run Prediction Models with earnings. Our models are 90% accurate at predicting volatility range and more than 75% correct on direction. All we need is 51% to be better than a coin flip.
Clearly, our models are superior to sellside consensus for outcome-prediction. I submit it’s because other factors drive outcomes. Demand, Supply, volatility. Long/short bets. The presence or absence of buy/sell patterns. Variable behaviors.
Earnings are reliable as a determinant of outcomes to the degree the market is motivated by them.
ETFs are by factors beyond anything else the largest form of fund-flows. And ETFs don’t care about earnings. Not because earnings don’t matter. Because ETFs are an arbitrage trade predicated on the exchange of collateral with predictable values. If anything, big moves with earnings are bad news for ETFs. It introduces unpredictability.
It’s why size matters most to ETFs, followed by liquidity and volatility.
For investors, we can predict short-term return probabilities in quantitatively curated large cap and liquid stocks to better than 90% routinely. Now, we target discrete outcomes – a gain of 2% or more in 1-5 days. What’s that probability?
The lowest probability yesterday before the open in that curated list was KKR at 57.1%. Well, it delivered in the first 30 minutes.
It’s measurable, predictable, mathematical. Ceteris paribus. Data, not advice.
We can nearly always outperform SPY with focused quantitative portfolios designed for do-it-yourself investors that rebalance with Demand/Supply divergences.
I’m not flogging our quantitative analytics. I’m questioning tactics.
A stock market where earnings outcomes are not the central determinant of flows should be navigated by public companies with other measures, including volatility and the percentage of volume driven by Passive Investment. We use both.
If the market is telling you by rendering a measure inoperable that you need a new measure, well. Maybe you need a new measure.
In May in beta tests on a new product, we scored companies on 15 metrics. Among them were JNJ, INTC. Which do you think scored higher? INTC, 42. JNJ, 84.
If you want to foster a product Blackrock will buy, you’re 2-to-1 better off as JNJ.
Whether you’re investing in stocks or trying to drive shareholder value at a public company, it’s important to measure the market the way it works. Anybody can flip a coin. We’ve got to be better than that.
