In December 2023, twelve of the world's largest investment banks published their official 2024 year-end gold price forecasts. Goldman Sachs, JPMorgan, UBS, Citi — all the names you'd expect.
The forecasts ranged from $1,950 to $2,300. Gold closed 2024 at $2,624.
Every single bank was wrong. Not by a little. By an average of 18%.
Meanwhile, the crowd quietly got it right
While the banks were finalizing their reports, a public prediction market on Polymarket was running the same question: "Will gold close above $2,500 by December 31, 2024?"
The market's price ranged from 32¢ in March to 71¢ by September — implying a probability that climbed steadily as new information arrived. By November the market had it at 89%. The actual outcome: yes.
You weren't reading about this on the financial news. You should have been.
Why the experts missed and the crowd didn't
The conventional view says expert forecasts are better because experts have models, data, and training. The data says otherwise — and there are three reasons that hold up across every market that's been studied.
First, experts publish. They write a forecast in December and defend it in February. The crowd updates. Every time new information arrives — a Fed pivot, a war, an inflation print — the market price moves within seconds. By the time a research note is republished, the market is three updates ahead.
Second, experts face career risk for being too aggressive. A forecaster who calls a 30% rally and is wrong gets fired. A forecaster who calls 6% and is wrong by 24% in either direction stays employed. So the consensus huddles toward the middle, regardless of the actual probability distribution.
Third, experts have no skin in the game. A bank's gold target is an opinion. The market price is a position with money on it. The asymmetry is enormous, and it shows up in the accuracy data.
What this means for Thai investors
Thailand consumes roughly 90 tons of gold annually — among the highest per-capita rates in the world. Nearly every Thai household has either bought, inherited, or is saving for gold. And nearly every Thai investor has, at some point, asked: "Where will the price go next?"
The honest answer is that no individual analyst — Thai or international — can tell you with reliable accuracy. But a market of thousands of forecasters with money on the line can give you a probability estimate that beats every published forecast we have data on.
That's a different question from "where will gold go?" It's "what does the most accurate aggregator of real money think gold will do?" The second question has a better answer.
The 73% rule that explains forecasting
Philip Tetlock, the Wharton researcher who ran the most rigorous forecasting tournament ever conducted, found something striking. When he aggregated thousands of small predictions into a single market-style probability, the result outperformed individual experts by an average of 73%.
Not because the crowd was smarter than the smartest expert. Because the crowd was smarter than the average expert — and the experts you're hearing on TV are, by definition, average.
The math is uncomfortable but unambiguous: trusting one analyst is worse than trusting the market. And until recently, Thai investors have had the analyst but not the market.
What Juno lets you do
Juno Predictions is bringing this to Thailand. Every market is a question — "Will gold close above $2,700 by July 31?" — with a probability set by people who have real money on the answer. You can read the price. You can disagree with it. You can take the other side.
You're not just consuming a forecast anymore. You're producing one. And you're getting paid when you're right.
The 0-for-12 lesson
The twelve banks weren't bad analysts. They were victims of a system that rewards confident-sounding opinions and punishes accurate updates. The crowd doesn't operate inside that system. It just prices the next minute.
The next time you read a year-end forecast from a name you trust, ask one question: what does the prediction market say? If the two disagree — and they often will — it's worth knowing which one has the better track record.