At this point, everyone has heard of the Leopold Aschenbrenner story with him being a child prodigy through his AI research and now, his hedge fund called Situational Awareness. Today, we dive deeper into the idea behind leveraged investing and the importance of timing when making these decisions.
Weekly Recap
This past week, we started with Palantir (NASDAQ: PLTR) beating earnings with an EPS of $0.41 relative to the expected $0.31. Revenue was $1.94B versus expectations of $1.81B sending the spot price soaring 29.5% to $162.66 from a prior close of $125.65. This was largely driven by high commercial spending but the jump in stock price can also be attributed to institutions covering short positions on the positive news.
On Friday, we also had a jobs report with nonfarm payrolls falling 23,000 in July which was a shock compared to the expected 83,000 gain. The unemployment rate fell slightly to 4.1% but the strong negative jobs report indicated more of a shift towards a rate cut rather than a rate hike as we look forward to the next Fed meeting in September.
What I'm Watching Next Week
For the week of Aug 10-15, key events I am looking at include CPI data coming out on Wednesday an hour before the bell. This will be a good indicator with the market estimating that headline inflation rate YoY will be at 3.5% with core inflation rate to come in at 2.5%. This will be a key indicator for where the Fed sets rates on the 16th of September. The market is currently pricing in a 55.6% chance that rates are held with a 44.4% chance that we get a rate hike.
On the earnings side, key names that I personally will be watching will be Berkshire Hathaway (NYSE: BRK.B) releasing this weekend with Rocket Lab Corp. (NYSE: RKLB) posting after hours on Monday. Berkshire Hathaway is typically a ticker with lower volatility traditionally following Warren Buffett's long-term investing thesis. The numbers have already been released as of writing, and I will be watching for the market reaction on Monday morning. Rocket Lab is known to be more volatile with valuations being linked to defense contracts. Finally, I will also be watching JD.com (NASDAQ: JD) for more of an idea of where China's AI spending is relative to the Magnificent 7's astronomical capital expenditures in recent months.
Research
Leopold Aschenbrenner spent most of 2024 and 2025 building his reputation as one of the smartest investors in the AI space to take advantage of high leverage plays. His trades consistently backed up the thesis that AI spending is still strong and has room to grow. While I could regurgitate this story like every news channel, Instagram account and LinkedIn post has done, I want to look at this from a different angle: the relationship between leverage, thesis and timing.
As of now, Aschenbrenner's thesis still holds and whether that holds true in the next five to ten years remains to be seen. The reason he got liquidated was not due to a bad bet but the level of leverage that he was using. We saw a sharp semiconductor and AI infrastructure sell off in the past month with big names like SK Hynix (NASDAQ: SKHY), AMD (NASDAQ: AMD), Micron (NASDAQ: MU), and Sandisk (NASDAQ: SNDK) among others falling from 6% to over 15% in single sessions. At roughly 4x leverage, as little as a 25% drawdown is enough to wipe out the position let alone 40% in some tickers. As the margin calls started coming, his fund, ironically named Situational Awareness, had lost 67% in one month. At time of writing, most of those tickers have rebounded substantially.
This proves that his thesis was not wrong. If anything, it strengthens the idea that there is more AI spending to come, though whether that spending is justified is a separate debate. Moreover, his issue was the amount of leverage that he used. This situation has taught me that the more leverage you use, the more awareness you need to have of your positions (pun intended). A fund with no leverage would indeed not have the headline making returns that he had but would also not have been liquidated through a small correction.
When trading options, you need to understand what your total risk will be, your time frame and where the trade fits in with broader macro trends. For example: if you want to trade short dated options, noise can really hurt your position whereas longer dated options have more time to play out within macro headlines. I am not saying that one strategy is better than the other as many people have made millions of dollars using both of these different strategies. The idea is to be wary that the more leverage and the less time that you have on the trade, the more right you need to be in your thesis with regards to maximum drawdown and direction.
Time in the market and experience can carry you through market cycles. This instance gave Ken Griffin the opportunity to buy $16B of the Aschenbrenner's liquidated fund at a discount of roughly 10% to the market. While these opportunities are few and far between, having the capital at a moment's notice can be useful in a time like this where Citadel's equities fund is up 14.2% in July, largely because of this purchase. Griffin did not need a better AI thesis than Aschenbrenner, he needed capital that was not forced to sell at the worst possible moment. This allowed Griffin to be able to turn a healthy profit to start off Q3 of 2026.
Personal
For this week, a lesson I keep having to learn is making sure data is clean and accurate. When trading algorithmically, there are many levels of data that you can use (e.g. Level 1 giving just the best bid and offer vs. Level 2 showing volume at different price levels). When trading options, you can get basic pricing or you can go further into option greeks. What I have realized this week is that just because data for pricing may be live, this doesn't mean that the greeks or the volume being fed through the API is live too. I encountered this problem when trying to incorporate a delta filter into my calculation for picking strikes and it kept screwing it up. To solve this problem, I first had to figure out the sample rate of the data I was receiving, which turned out to be every 20-30 minutes. This was way too slow for my use case so instead, I turned to using gamma. The data I am getting batches all greeks together. While gamma is not any fresher, it has a slower rate of change by definition. Delta moves close to linearly with the underlying price and gamma represents this slope. Instead of relying on old deltas, I can use the gamma value with the new live options prices to recalculate and estimate delta. It is not exact, but it works for my use case.
That is all for the first edition. I hope you learned something. The more I read, the more I realize I have to learn. This should hopefully give me a chance to stay in touch with markets while giving readers the opportunity to see inside the mind of a 20-year-old options trader for what that is worth.
Disclaimer: Everything here reflects my own opinions and is shared for informational purposes only. It is not financial advice, and nothing in this report is a recommendation to buy or sell any security.