After a tumultuous few months, June of 2025 saw a strong rally which took global markets to (or close to) new highs. The rally was broad-based, with international and U.S. markets all up strongly. Leading the pack were emerging market stocks, followed by communications and technology. International stocks continued to do well, with VXUS up 4% in June. European stocks–the global leaders so far in 2025–continued their rally, but were a bit of a laggard in June. Even Treasuries eked out a bit of a gain, while duration-heavy equities (REITs and utilities) were underperformers. QuantStreet had a strong month in June, in line with our benchmarks. You can see details about our performance here.
The biggest loser in the month was the dollar, continuing its dismal year-to-date performance. The emerging era of geopolitical competition–with the U.S., Europe, and China jockeying for global influence–has brought forth an end-of-dollar-dominance narrative, which the market has not been able to get past. Our view is that the end of dollar dominance is not yet upon us. The U.S. remains the global leader in rule of law, entrepreneurship, depth of capital markets, and technological innovation.
Unfortunately, some of these areas of leadership have been called into question this year on the back of U.S. foreign policy missteps. However, the Trump administration has also shown a willingness to change course based on feedback (from advisors or from markets) and current U.S. foreign policy appears more level-headed, with a reasonable likelihood of leading to good outcomes for the U.S. and our trading partners. And while it is true that China is catching up to the U.S. in the AI race, it seems that China is carving out a market position as a low-cost provider while the U.S. maintains its lead in state-of-the-art AI applications and hardware. Furthermore, the U.S. corporate sector is far ahead of Europe’s and China’s in AI adoption.
The bearish dollar narrative has done its damage, but a lot of bad news is already priced in, and the dollar is set up to rally in the back half of the year, should news flow turn just a bit less negative.
AI assessment of relevant news flow
One of the key challenges in managing a portfolio, whether focused on asset allocation (as we are) or on security selection, is the inability of any one individual (or even a group of individuals) to follow all the news flow relevant to a given set of positions. While our asset allocation process at QuantStreet is systematic–driven by machine learning return forecasting algorithms combined with portfolio optimization–we have long believed that the data that feeds the forecasting process is not rich enough to capture all relevant news flow.
Thanks to Google Cloud, however, we now implement a monitoring process which allows us to fill in the gaps. We send recent news headlines to an AI engine, along with our portfolio and prospective positions. The AI engine–currently the Gemini 2.5 Flash model–then generates an opinion about how the passed in set of news interacts with portfolio positions. The AI engine converts a very large volume of text into a relatively short (several pages) summary of how recent news impact our asset allocations. Our use case is not to redistribute these data, but to use them internally for purposes of monitoring current and prospective portfolio positions.
The actual amount of code needed to set up this process is not large, though the learning curve to connect all the pieces together was reasonably steep. But once the startup costs are paid, the process is seamless, requiring only occasional oversight. Finally, as the capabilities of AI models improve, it is trivially easy to replace Gemini 2.5 Flash with a more capable model. The process, therefore, doesn’t grow stale, but instead becomes stronger over time.
Insights
Prior to releasing our July 1st portfolio updates, we ran the above process using the last two weeks of news flow. Please keep in mind that we are not taking a stand on the veracity of the analysis shown in the bullet points below; indeed, some of the news flow is stale and we don’t agree with some of the interpretations, as we discuss below. And this is only one input into our decision making process (we are avid news readers ourselves). See the disclaimer at the bottom of the next section.
Here, in its own words (in blue), are some of the AI model’s interpretations of how recent news impact our portfolio positions:
-
Escalating Geopolitical Tensions: The primary and most impactful theme is the ongoing and intensifying conflict between Israel and Iran. News covers Israeli airstrikes on Iranian nuclear sites, Iranian missile attacks on Israeli cities and a U.S. base, and discussions about potential U.S. involvement, including a possible expansion of the conflict. This is leading to concerns about regional stability, oil supply disruptions (evidenced by rising oil and jet fuel prices), and a broader sense of global insecurity.
-
U.S. Political and Economic Policy Uncertainty: The Trump administration’s domestic policies are generating significant headlines. This includes ongoing struggles to pass a ‘megabill’ in Congress, facing setbacks due to Senate rules and internal Republican divisions, which impacts various sectors from healthcare to clean energy (EV tax credits).
- EWG tracks German equities, making it highly sensitive to European economic and geopolitical stability. The escalating Israel-Iran conflict, with its impact on global oil and jet fuel prices (“Jet fuel prices soar in Europe”), directly affects energy costs for European industries. Broader discussions about NATO spending (“Trump says US doesn’t have to meet NATO spending goal,” “NATO allies agree to 5 percent defense spending goal”) and potential trade tensions initiated by the U.S. (“Trump says US ending all negotiations with Canada over digital tax,” “Trump says he’s not planning to extend a pause on global tariffs”) create uncertainty for export-oriented economies like Germany. The generally negative SF Fed news sentiment also contributes to a cautious outlook for international markets.
- Gold is traditionally considered a safe-haven asset during times of geopolitical instability and economic uncertainty. The continuous escalation of the Israel-Iran conflict, potential U.S. involvement, and the general negative news sentiment indicated by the SF Fed index (-0.087) all increase demand for safe assets. News of stock futures falling due to conflict concerns directly correlates to increased appeal for gold.
- VGSH invests in short-term U.S. government bonds. Positive: During periods of elevated geopolitical risk and general market uncertainty, government bonds are seen as safe havens, increasing demand and potentially supporting prices. The negative SF Fed news sentiment aligns with this flight to safety. Neutral/Slightly Negative: The Federal Reserve’s economic projections hinting at “stagflation concerns” and the political discussions around a new Fed Chair or interest rate cuts will directly influence bond yields. Short-term bonds are less volatile than long-term bonds, but their returns are still tied to the Fed’s stance. The struggle to pass the “megabill” might also imply more fiscal uncertainty, which can be mixed for bonds depending on how it impacts inflation and borrowing.
- XLI covers industries like aerospace & defense, machinery, and transportation. Positive: The “US Steel Sale to Nippon Steel Poised To Close After Trump Deal” could be seen as positive for the steel industry, although foreign acquisition of a key domestic asset can also be politically contentious. SpaceX’s successful Falcon missions indicate continued activity in the aerospace sector. Negative: The Air India plane crash, involving a Boeing 787, raises safety questions and could lead to increased scrutiny or potential negative sentiment towards Boeing (a significant industrial company). Trump’s stance on tariffs and trade (“ending pause on global tariffs”) creates uncertainty for global supply chains and manufacturing, impacting some industrials.
- XLY covers consumer-facing industries like retail, automotive, and entertainment. Positive: Box office success of movies like “How To Train Your Dragon” and “F1,” the FIFA Club World Cup, and positive sports news (Caitlin Clark’s return, NBA Finals developments) boost the entertainment and leisure segments. Negative: Trump’s bill potentially ending EV subsidies (“Trump’s Bill Would End EV Subsidies: Could This Kill Tesla?”) and Elon Musk’s related “utterly insane” comments directly threaten the Electric Vehicle segment, a significant component. Tesla’s robotaxi ambitions facing a “reality check” and regulatory scrutiny add to this. The general negative sentiment and economic uncertainty could also lead consumers to curb discretionary spending.
There is a lot more in the analysis, but the above are some of the highlights.
Assessment
First, the above analysis is impressive. It hits on many subtle points about both news flow and the impacted assets. And it is automatically generated! Second, the prompt (or question) which is sent to the AI model matters a lot. Our early, simplistic prompts often led to model analysis which was far more negative than seemed reasonable to us. We needed to provide the model with lots of context, e.g., news headlines gravitate towards negativity and many positive developments in the world are not regularly covered in news headlines, as well as to provide guard rails around the model’s degree of negativity. Even so, after much experimentation, it still feels like the model’s analysis is too negative and too simplistic in many places.
There is a lot of first order logic, e.g., bad news lead to lower prices, without incorporating the second order logic that after bad news have hit, prices might already be quite low, thus offering investors attractive forward-looking returns. Another example is the model’s interpretation of NATO news headlines as negative, without considering the positive effects additional defense spending might have on European economies. Despite these flaws, the model’s ability to reason about world events and the prompts we send the model will both improve over time. The analysis is likely to get better.
Our goal at QuantStreet is to construct cost-effective, risk-appropriate portfolios which respond to changing market conditions. Our (new) ability to use AI to monitor and risk manage our portfolios–though the process is certainly imperfect–gives us a valuable tool to better serve our clients.
Importantly, please note that forecasts of financial markets may fall far from actual outcomes (and often do). The use of AI does not change this simple fact. The context we provide the AI models might be improper; the news data might be inaccurate or stale; the model’s interpretations might be incorrect. There are many things that could go wrong, so the above analysis should be interpreted with caution. Please see our full disclosure about the limitations of AI at https://quantstreetcapital.com/terms-of-use/.
Working with QuantStreet
QuantStreet is a registered investment advisor. It offers wealth planning, separately managed accounts, model portfolios and portfolio analytics, as well as financial consulting services. The firm’s approach is systematic, data-driven, and shaped by years of investing experience. To work with or learn more about QuantStreet, join our mailing list or contact us at [email protected].
A message from Advisor Perspectives and VettaFi: To learn more about this and other topics, check out some of our webcasts.
© QuantStreet Capital
More Innovative ETFs Topics >
