NEW YORK — Saying traditional valuation tools "no longer capture the lived experience of the American investor," a consortium of Wall Street analysts on Thursday unveiled Tweets Per Percentage Point, a metric that measures how many presidential social media posts it takes to move the S&P 500 by one percent, and which its authors called "the first honest pricing model since the abacus."
"For a hundred years we priced stocks on earnings, interest rates and forward guidance," said Corinne Vasquez-Hale, head of narrative quant at Pemberton-Ostrowski Capital, standing in front of a slide that showed none of those things. "Then we looked at what actually moved the market, and it was a phone. So we built a model around the phone."
The formula is simple. Analysts count the posts published during market hours, divide by the absolute change in the index, and adjust for capital letters. A low TPP means efficient posting. A typical week runs between 9 and 12. The consortium's benchmark is Oct. 10, 2025, when a single post saying the administration was considering "a massive increase in tariffs" on Chinese goods helped send the S&P 500 down 2.7%, its worst day since April, and the Dow down 878 points. "One post, 2.7 points," Vasquez-Hale said. "A TPP of 0.37. That's the Babe Ruth of posting. We had it framed."
The metric has a pedigree. In 2019, analysts at JPMorgan Chase built the "Volfefe Index," which found that 146 of roughly 4,000 presidential tweets sent during market hours had moved Treasury yields, with "China," "billion" and "products" among the words that did the most damage. TPP improves on that work, its authors said, by "dropping the pretense that the other 3,854 were noise. They were inventory."
The posts now go out on Truth Social, not Twitter, but the consortium kept the name. "Nobody on a trading desk is going to say 'Truths Per Percentage Point,'" said Vasquez-Hale. "Compliance won't let us put the word 'truth' in a model." Analysts also noted that the acronym gives the United States a TPP it is willing to stay in.
The Post Impact Terminal
Build a hypothetical presidential post from the words the market fears most, set the caps lock and the punctuation, pick a time, and see what the model says it does to the S&P 500.
KeywordsThe Post Impact Terminal is a toy model. Its weights are loosely inspired by the 2019 Volfefe Index's list of market-moving words and are otherwise invented. It does not predict anything, which puts it level with most models. Not financial advice.
The consortium has published a family of supporting measures. The Caps Lock Coefficient adjusts for how much of a post is shouted. Exclamation Elasticity tracks the diminishing returns of the third exclamation point. The Weekend Gap measures posts published on Saturday morning, which cannot move anything until Monday and therefore "accrue interest." The most closely watched is the Lunch Window. JPMorgan's 2019 study found that the most market-moving tweets usually landed between noon and 2 p.m. Traders now eat at their desks while staring at their phones, which analysts noted is also how they ate before.
The industry has adapted. At Hollis Gate Research, every analyst was issued a second monitor for the posts in January. By February the posts had been moved to the main monitor, and the Bloomberg terminal to the second. "The Bloomberg tells you what happened," said one Hollis Gate associate, who asked not to be named because he has not looked at it since March. "The phone tells you what's about to have happened."
The model's one known flaw is what its authors call "negative TPP." In March, traders placed roughly half a billion dollars in oil bets about 15 minutes before a presidential post about delaying strikes on Iran, Reuters reported, and oil prices then plunged. No evidence of insider trading has been established. Analysts say the episode breaks the formula anyway, because the percentage point arrived before the post. "You can't divide by a post that hasn't happened yet," said Vasquez-Hale. "Well, you can. Somebody clearly did. We just can't figure out who."
The consortium handled the problem by adding a variable, the Clairvoyance Premium, which captures the share of each move that happens before the news. It has declined to ask who is collecting it. "That's an enforcement question," Vasquez-Hale said. "We're a metrics shop. We measure the money. We don't follow it."
The Front-Run Simulator
You are the poster. Watch the volume bars tick by, then publish whenever you decide to. Nobody can know when that will be. Not even you.
Simulation. The 15-minute lead and the scale of the bets are modeled on press reports of oil trading before a March 2026 post; no wrongdoing has been established in that case. In this simulator, the spike is always 15 minutes early, because the simulator has a source.
Academics remain skeptical. "The metric is ridiculous," said Dr. Abner Lusk, who studies market fundamentals at the Institute for Things That Used to Matter. "Unfortunately, it outperforms mine." Lusk noted that Bank of America analysts reported in 2019 that on days with more than 35 presidential tweets, the market fell slightly, and on days with fewer than five, it rose. "So we have known the cure for seven years," he said. "Nobody has been able to administer it."
Retail investors are already on board. Several brokerage apps are testing TPP push alerts, which notify users that a post has moved their portfolio, and then a second alert when the post is deleted, and then a third alert when the deleted post is reposted with more capital letters. A TPP-weighted exchange-traded fund is expected to list next quarter under the ticker POST. Its prospectus lists a single holding, which is not a company.
The consortium is already planning ahead. Its next paper will introduce Percentage Points Per Shrug, a companion metric for the day posts stop moving markets. Analysts said they do not expect to need it soon.
At press time, the S&P 500 had dipped 0.4% after a post about Tweets Per Percentage Point, giving the metric a TPP of 2.5 on its first day. Analysts called it "a strong debut."
Reader Comments (2,704)