Tom Yeung here with your Sunday Digest.
In the 1940s, mathematician John von Neumann began working with Princeton University meteorologist Jule Charney on a computer model to predict the weather.
The first version… well… wasn’t that useful. It took 24 hours for their ENIAC punch-card machine to model 24 hours of weather… something you could also do in real-time by looking out a window.
But computers eventually got faster, and weather forecasters began using machines to help refine their work. By the early 2000s, computing power had grown powerful enough to run models independently of humans.
Recent advancements in AI have only improved their accuracy.
Now, it’s important to note that these weather “predictions” are not the type made by the Oracle of Delphi or your local tarot card reader. Those are more like guesses about the future.
Instead, computer weather forecasts are digital representations that divide our environment into smaller pieces (usually in a three-dimensional grid, ranging from 10 to 50 kilometers) and attempt to model what will happen next.
If one cube in the grid has a certain temperature, humidity, and wind speed, how would it affect its neighboring cubes? And what does it look like once the grid covers the entire continental U.S.?
It’s essentially creating a simplified version of the real world within a computer program.
The same quantitative principle is now used in financial markets… both on Wall Street and right here where I’m employed.
Over the past several years, our team at TradeSmith has created a detailed system that models the likely next steps in a market. If a retailer like Walmart Inc. (WMT) reports strong financial results, they can use those numbers and resulting share-price moves to see what might come next for rivals like Target Corp. (TGT).
In a new presentation later this week, TradeSmith CEO Keith Kaplan will introduce this system, known as TradeSmithGPT. The AI-powered model uses millions of financial datapoints to pinpoint each stock’s “profit window,” or the ideal time frame to trade a stock, on any given day.
They’re now offering a “lite” version of this system for a limited time to test for yourself. To get access to that “lite” version, just click here to sign up for Keith’s free presentation on Wednesday, June 25, at 10 am. Eastern.
The TradeSmith team is allowing me to reveal three of their system’s top picks this week. However, you may not want to buy or trade any of these stocks until Keith explains how it works.
As you’ll see this week, these stock choices have some serious upside… but they require a system like TradeSmithGPT to know exactly when to get in and out.
A Crude Awakening
Recent events in the Middle East have sent shockwaves through energy markets. Over the past month, crude oil prices have jumped 18% as memories of previous conflicts have resurfaced. Natural gas has risen 27%.
Our quantitative systems believe even more gains are on the way. This week, the TradeSmith model projects a 22% rise in Comstock Resources Inc. (CRK) over the next 30 days. The Dallas area-based natural gas producer now has the largest upside we’ve seen in a while.
The alert is also notable because TradeSmith’s model has an 85.2% historical target accuracy for CRK. In other words, an investor following TradeSmithGPT’s recommendation on CRK would have been right 85.2% of the time.
Now, it’s important to note that CRK is an unabashedly leveraged bet on energy prices. The energy firm generated $2.2 billion in operating income during the 2022 natural gas boom… and saw that figure flip to a $170 million loss last year.
In fact, the company noted in its latest annual report that raising its natural gas reference by 77% to $3.26 per 1,000 cubic feet (Mcf), up from the $1.84 per Mcf the Securities and Exchange Commission mandates for 2024 financial reports, increases its reserves value by 250%.
That means rising gas prices can trigger gains of 50% or more. Comstock has already risen 175% this year on tightening gas markets. A sudden decline in gas prices, however, would trigger an equally sharp selloff in CRK.
And so, a model like TradeSmithGPT is a “must-have” for getting out just as smart money has second thoughts.
The TikTok Tug-of-War
The cosmetics firm e.l.f. Beauty Inc. (ELF) is no stranger to volatility. Shares surged 700% between 2022 and 2024 as young TikTok viewers discovered the brand. A crackdown on the social media app then sent shares straight back down.
Since then, the Trump administration has moved to repair relationships with TikTok, boosting ELF stock. In January, Donald Trump signed an executive order that temporarily paused a “ban” on the video streaming app. Earlier this week, he delayed the ban for another 90 days.
That’s great news for ELF, which has seen its share price recover sharply in recent weeks. The company relies on word-of-mouth advertising, and its popularity on TikTok makes the video streaming app a crucial part of its marketing strategy.
However, most of ELF’s products are sourced and manufactured in China. As the company’s management dryly notes in its annual report, tariffs will have an impact on both costs and pricing:
As a result of the additional US tariffs announced since early 2025, the Company will raise prices globally for all products sold, which could result in the loss of consumers and materially and adversely affect our business, financial condition and results of operations.
If the sudden thawing of U.S.-China relations holds, the impact could be less than feared.
That’s likely why the TradeSmith system is modeling a 15% upside for ELF. Shares have already doubled since April, and history tells us that rising stocks typically keep going up in the short term, especially retail-facing stocks.
The Crypto Coaster
Finally, the TradeSmith system suggests Robinhood Markets Inc. (HOOD) this week, a leveraged play on cryptocurrency and retail stock and options trading. The system is forecasting a 13% upside and has an 84% historical target accuracy on the stock.
The reason for the system’s bullishness is likely because Robinhood earns an enormous amount of its revenues from crypto trades. In the first quarter of 2025, that segment accounted for 43% of total transaction-based revenues – slightly more than the 41% from options, and well ahead of the 10% from stocks.
Crypto markets are less efficient than stock markets, allowing Robinhood to earn more from market makers through a system called “payment for order flow.”
Where Robinhood’s revenues come from
The recent surge in crypto prices will likely boost Robinhood’s bottom line. Over the past several months, we’ve seen:
- Rising Bitcoin (BTC) prices. The world’s largest cryptocurrency rose above $100,000 in early May – a strong psychological barrier.
- Crypto-friendly legislation. On Tuesday, the U.S. Senate passed the GENIUS stablecoin bill, opening the door to banks, fintechs, and retailers to use crypto.
- Improving investor mood. The share of “bullish” investors, as reported by the American Association of Individual Investors, has risen to 37%, up from 21% in April.
Together, these are typically strong signs for Robinhood, which caters to retail traders. The average Robinhood user is just 35 years old and tends to favor momentum trades that have gone up significantly.
That’s made HOOD’s stock a leveraged play on Bitcoin prices. Shares have risen 250% in the past year, compared to a 60% gain in BTC.
Our TradeSmith system is now forecasting another double-digit upside for Robinhood to $84.69 as Bitcoin heats back up. And much like TradeSmith’s other two picks this week, an investment in Robinhood must come with a system that knows when to sell.
Modeling the Future
The rise of AI has created a revolution in weather forecasting.
Traditional models typically used principles like fluid dynamics and thermodynamics to predict what would happen next. AI systems use data to find what actually happened… and then build future estimates from those insights.
That’s allowed AI developers like Google’s DeepMind to create better forecasts. Its weather model is now more accurate than traditional ones in 90% of tested cases. The real world simply doesn’t follow theoretical formulas to the letter.
The same is happening in the stock market. After all, academics have spent decades working on formulas to capture what should happen in financial markets.
- Higher risk should lead to higher rewards…
- Companies should be indifferent between debt and equity financing…
- Prices should follow a random walk…
But we know that none of those are necessarily true in real-world investing. People are emotional… markets aren’t perfectly efficient… and traders often leave “clues” that reveal their next moves.
That’s why TradeSmithGPT is so powerful. It’s able to take far more data than any human can manage and creates real-world models of what should come next with a 75% accuracy rate for its target prices.
In a free presentation on Wednesday, June 25, TradeSmith CEO Keith Kaplan will reveal exactly what his new breakthrough AI software is, how it works, and why it will be critical to your success in the markets during 2025.
But fair warning: His live broadcast will not be available to the general public.
To receive your private access link, please click this link to get signed up.
When you do, you’ll also get immediate access to a “lite” version of TradeSmithGPT for a limited time to test it out for yourself.
After that, be sure to tune into Keith’s presentation on Wednesday, where he explains how to use the system to trade fast-moving stocks like the ones I mentioned in this email today. AI-based systems are quickly becoming the gold standard of modeling, and it’s becoming essential to know how to use these new tools.
Until next week,
Tom Yeung, CFA
Market Analyst, InvestorPlace
