Für über 2 Millionen Euro nachgekauft: Meine 3 Top-Aktien (Michael C. Jakob von AlleAktien)

Finance-Focused Summary

Market Outlook and Sentiment

• The presenter cited a Fear & Greed Index reading of 31, indicating fear. It had recently fallen to 18, which he described as extreme fear (below 25). He said the index reflects sentiment around the S&P 500, not German, Chinese, or other markets.
• He argued that an index near its all-time high is not, by itself, a reason to avoid investing. At the time of the livestream, he put the S&P 500 roughly 0.7–0.79% below its high, while the XLK technology ETF was about 6.6% below its high, after being nearly 10% below the previous week. He also noted that some individual tech shares had fallen much more, citing CoreWeave as down over 30% in a week.
• His general recommendation was to invest spare cash when sentiment is fearful, while assessing each company’s business and valuation rather than relying on the market’s distance from its high. He pointed to the S&P 500’s roughly 25% return in 2024 as an example of how buying near a high can still work out.
• He said investor sentiment and seasonal interest in stocks often improve in October–December, as investors review and reposition portfolios for the coming year.

Shares He Said He Bought or Currently Favours

• He said he had bought more than €2 million worth of three stocks: Berkshire Hathaway, Amazon (AMZN), and Oracle (ORCL). He joked that even a much larger purchase would not move Berkshire’s share price.
• Amazon: He described it as more than 23% below his estimate of fair value, citing the potential of its cloud business.
• Oracle: He said it was more than 40% undervalued by his estimate and forecast a possible move to $200–$250. He cautioned that timing was uncertain: the stock could move quickly or take another year. He said he would be comfortable holding Oracle for 10 years and compared its long-term prospects with those of large technology companies such as Microsoft and Amazon. His claim that these kinds of stocks could return 30% annually over a decade was an optimistic personal projection, not a guarantee.
• Berkshire Hathaway was described as his third-best stock to buy at that point. He also suggested McDonald’s (MCD) as an alternative for an investor already heavily weighted in other stocks.
• He said ServiceNow (NOW) had gained over 70% and described software shares, including Salesforce (CRM) and Oracle, as undervalued. He also cited Alphabet (GOOGL) as a prior recommendation that he said gained 18% and then doubled within six months; the transcript’s timing and return reference are unclear.
• Other positive views:
  • Constellation Software (CSU): Still on his watchlist and, in his view, worth buying because sales and profits were continuing to grow.
  • Reddit (RDDT): Described as cheap and worth considering, with a valuation he compared favourably to Meta (META).
  • Rollins (ROL): He said he was not concerned by the share-price weakness, citing resilient revenue and profit growth and the defensive nature of pest control.
  • Waste-management-related company: He referred to a company with a local monopoly that had fallen over 40% and said it could be a good time to add. The company name is unclear in the subtitles, so it cannot be identified confidently.
  • Netflix (NFLX): In principle worth buying, but he said more attractive opportunities were available.
  • He also named 2G Energy, Pool Corp (POOL), Novo Nordisk, and Johnson & Johnson (JNJ) as examples of holdings or candidates across different sectors.

Cautions and Valuation Views

• Apple (AAPL): He said it was about 20% above his fair-value estimate, with a P/E ratio around 45, and saw no sufficiently strong new growth driver—particularly from AI—to justify the valuation. He estimated annual returns of 7–8%, below his cited 12% long-run S&P 500 return. He said he would sell Apple in a trading portfolio, but might hold long-term positions, especially where selling could trigger substantial taxable gains.
• Palo Alto Networks (PANW): He described it as heavily overvalued and said he would consider selling.
• Portfolio concentration: When asked about Amazon, Oracle, and Alphabet together making up 12% of a portfolio, he distinguished between 12% combined and 12% each. If each already represented 12%, he would not add more, stressing diversification.
• He said software businesses may benefit from AI by reducing their own costs while retaining much of their revenue, but he also discussed risks to AI-model providers.

AI Business Models and Competitive Risk

• He distinguished between AI companies selling subscriptions and charging for API usage. He said API sales accounted for over 85% of Anthropic’s sales; this figure is his claim in the livestream.
• He said his own company was spending roughly €5,000–€10,000 per month on OpenAI and Anthropic services, and that it had shifted some work to free, locally run models to reduce costs.
• His main risk thesis for OpenAI and Anthropic was that open-weight models—many of them Chinese, including Qwen and DeepSeek—could be run locally for little more than hardware and electricity costs. He said a model could run with about 30 GB of RAM, with 100 GB preferable, and claimed current free models were roughly six months behind paid models.
• He argued that this could weaken the providers’ competitive advantage because customers may find it easy to switch between APIs or use local models. He viewed the risk as less urgent over the next one to two years, but potentially more significant later. He suggested local AI could become more practical over roughly four years, as hardware improves.
• By contrast, he saw less risk for Amazon, Microsoft (MSFT), and Oracle because they own data-centre infrastructure that can serve AI, conventional software, and other computing demand.
• He expected Anthropic’s IPO near the end of the year and OpenAI’s early the following year, though the transcript does not establish the year clearly.

Leveraged ETFs

He said he did not favour leveraged index ETFs, including 2× S&P 500, MSCI World, or DAX products. His reasons included:

• They typically reset leverage daily, which can cause leverage drag in volatile markets; returns are not simply twice the index’s long-term return.
• He cited an example where a leveraged product had returned about 220% versus roughly 175% for the S&P 500 over the comparison shown, rather than the roughly 350% he said would represent twice that index return. The exact dates and product are unclear in the subtitles.
• He also cited a 0.6% product fee and claimed embedded financing costs could be equivalent to 15–20% interest. These were his stated figures, not independently verified in the video.

He contrasted these products with borrowing separately and investing in the S&P 500, describing this as a potentially cheaper way to create two- or three-times exposure. This involves borrowing and can amplify losses; the video does not provide a full risk assessment.

Portfolio Construction and Analysis Framework

The presenter described this approach for testing portfolio allocations with Monte Carlo analysis and a Pareto frontier:

1. Supply the model with explicit expected returns for each holding, preferably using the presenter’s watchlist exported as a CSV or Excel file. Do not let the model substitute its own return estimates.
2. Include risk measures such as volatility, maximum drawdown, and the Sortino ratio.
3. Use historical data—he suggested looking back 10–20 years—and calculate risk metrics through the Eulerpool API or its risk-and-return analytics.
4. Define how objectives such as maximum drawdown, minimum drawdown, and Sortino ratio are weighted. He cautioned that optimizing heavily for drawdown can produce a different allocation than optimizing for the Sortino ratio and may reduce expected returns.
5. A viewer’s analysis had produced a 40–50% S&P 500 allocation, but he said he could not assess whether that result was sound without seeing the inputs and weightings.

He gave a platform example showing Microsoft with a one-year maximum drawdown of 13% and a Sortino ratio of 1.13.

Tax Comments and Disclosures

• The livestream promoted an AlleAktien “0% tax” and emigration training. The presenter asserted that U.S. estate-tax concerns generally would not affect someone who remained resident in Germany or another non-U.S. country, but could be relevant where assets or accounts were held in the United States. This was not presented with a detailed legal analysis.
• No explicit “not financial advice” disclaimer appears in the subtitles. The presenter’s stock valuations, return forecasts, and tax comments should therefore be understood as his views, not guaranteed outcomes or individualized advice.

Assets and Instruments Mentioned

Stocks and companies: Berkshire Hathaway, Amazon, Oracle, Microsoft, Alphabet, Meta, ServiceNow, Salesforce, CoreWeave, Constellation Software, Netflix, Rollins, Reddit, Apple, Palo Alto Networks, McDonald’s, 2G Energy, Pool Corp, Novo Nordisk, Johnson & Johnson, Nvidia, SAP, Waste Management, and an unidentified waste-management-related company.

Funds, indices, and other assets: S&P 500, XLK, MSCI World, DAX, leveraged ETFs, data centres, Bitcoin, and crypto mining.

AI models and providers: OpenAI, Anthropic, Google Gemini, Qwen, and DeepSeek.

Sources and presenter: Michael C. Jakob of AlleAktien presented the livestream. He referenced Bloomberg, the Financial Times, the Wall Street Journal, Comdirect, and Eulerpool during the discussion.