Every December, Wall Street strategists publish their year-end S&P 500 targets for the coming year.
Since 2000, the consensus has been off by an average of 14%.
If a weatherman missed the high temperature by 14% every year for 25 years, you'd ignore them. Why don't we ignore the strategists?
The 25-year scoreboard
Each year, Bloomberg surveys roughly 20 chief strategists from the major sell-side banks — Goldman, Morgan Stanley, JPMorgan, BofA, Citi, Barclays, UBS, Deutsche, and so on. The median target is published. It is then forgotten by January, when the next year's targets come out.
Here is the scoreboard that almost nobody publishes:
2008: Consensus target 1,650. Actual close: 903. Miss: 45%.
2020: Consensus target 3,280. Actual close: 3,756. Miss: 14%.
2022: Consensus target 4,825. Actual close: 3,840. Miss: 20%.
2023: Consensus target 4,075. Actual close: 4,770. Miss: 17%.
2024: Consensus target 4,861. Actual close: 5,882. Miss: 21%.
The average miss across the 25 years between 2000 and 2024 is roughly 14%. In years with meaningful market moves — recessions, COVID, the 2023 AI rally, the 2024 melt-up — the consensus misses by 20% or more.
The only years the consensus is "right" are years where the market does almost nothing.
The herd problem in one chart
Plot every strategist's year-end target against the consensus, and a striking pattern emerges. Targets cluster tightly around the consensus, almost always within ±5%.
This is not because the strategists agree. It is because being wrong with the herd is safe; being right alone is dangerous. If your number is far from the median and the market moves the other way, you lose your job. If your number is near the median and the market moves the other way, everyone was wrong together.
So the targets cluster. And because they cluster, the consensus is structurally biased toward small moves — the one outcome the market rarely delivers.
What the targets are actually built from
The official methodology at every major bank is roughly the same. Take a forward earnings estimate. Apply a target multiple. Multiply.
The catch: the forward earnings estimate comes from the same equity strategist team. The target multiple is whatever multiple makes the resulting number land within shouting distance of the consensus.
Goldman has admitted as much in published research. JPMorgan strategists have written entire notes about why their previous year's target was wrong. The forecasts are not the output of a model. They are the output of a committee that doesn't want to be the outlier.
The three years that broke the consensus
2008: The consensus expected 1,650 by year-end. Lehman fell in September. The S&P closed at 903.
2020: The consensus expected 3,280, set in December 2019. Then a pandemic. The market fell 34% in five weeks and then rallied 65% to close at 3,756. A target set six weeks before the largest economic dislocation of the century could not have been right by any process.
2023: After the brutal 2022, the consensus was bearish — 4,075. The market closed at 4,770. Every strategist missed the AI rally. The companies that drove the rally — Nvidia, Microsoft, Meta — were in the S&P, knowable, public. The strategists missed the signal because their model said "earnings recession" and the market said "spend $50 billion on GPUs."
What a market-implied target looks like
S&P 500 options markets give you a real-time, money-backed estimate of where the index could land by any future date. The implied distribution updates continuously, narrows as the date approaches, and reflects the actual money on the table.
In December 2023, options markets implied a 65% probability that the S&P would end 2024 between 4,400 and 5,200. The actual close was 5,882 — outside the range, but the upper tail of the distribution had a non-trivial weight. Strategist consensus had zero weight there.
Prediction markets do the same thing for any binary event. "Will S&P close above 6,000 by year-end?" becomes a contract. The price is the probability. Aggregated across thousands of traders, that probability is calibrated.
The three things strategists do that markets don't
Anchor on the previous year. Targets quietly start from "last year's close plus the long-run average return" and adjust from there.
Pretend earnings can be forecast. Forward earnings estimates have their own large error bars. Building one number forecast on top of another doesn't reduce the uncertainty — it hides it.
Round to the nearest 100. Targets cluster at 4,800 or 5,000 or 5,200, never at 4,837. The fake precision is exactly that — fake.
How Juno fits in
Juno is a prediction market built for global retail traders. Instead of reading one strategist's target — anchored, hedged, and 80% likely to be wrong — you can see what thousands of traders are willing to pay for specific S&P outcomes.
"S&P closes above 6,500 by end-2026" becomes a real contract with a real price. You can buy, sell, or just read it. The market does the calibration work for you.
The 14% lesson
If the consensus has missed by 14% on average for 25 years, the next "consensus target" you read is probably wrong by about that much.
You won't know which direction. But you know it's not 5,200. Or 5,800. Or whatever round number they're publishing this December.
You can keep reading the targets. Or you can read the market.