In February of this year I wrote the following in How High Can Nvidia Go?:
I believe Nvidia will continue its run and could even become the most valuable company in the world for a short time, but then a pullback (but not a crash) is inevitable for the factors outlined in this article. It’s difficult to predict when this will take place, so I hesitate to put money into the stock now. If I were forced to predict, I would say the pullback will take place towards the end of 2024 or into the first half of 2025.
If this opinion was sincerely held, then why would I have not put money into Nvidia at that time? Today Nvidia is up ~80% since this article was written, so it would have been a marvellous trade. The reasons are my self-imposed rules for stock investing which have served me well in the past:
Buy low and specifically during downturns
Hold for at least one year (to pay lower taxes and reduce risk)
Only sell when the investment thesis no longer holds (do not base on price alone)
Now and into the future these rules will govern the bulk of my portfolio, but over the last few months I’ve realised that they should not be treated as gospel when an opportunity presents itself which violates them.
Updating the Model:
There were several stocks I hesitated to buy this year which did not fit the model of buying when low and holding for 1+ year. Broadcom (AVGO 0.00%↑) was one near the top of my watch list, which announced earnings June 12 and rocketed up 23% this past week. It is a strong AI momentum stock that has market beating returns for several years, and thus in our model is perceived as being “too expensive.”
Like Nvidia and many other market leading semiconductor stocks that utilize TSMC’s manufacturing, Broadcom announced a beat and raise (the past was better than expected, and the future will be better than anticipated) at earnings. When all of these chip companies are clamouring for the latest TSMC process node and more supply, it should be obvious that TSMC has an immense amount of negotiating power to demand higher prices from its customers (heck, even Intel is using TSMC for their latest products). TSMC announces earnings in mid-July, and I believe these will be strong based on the forward guidance and earnings of its largest customers. My belief is that this trade will gain 10-20% in the next couple months, and the position will then be sold prior to the November election.
Previously I mentioned my aversion to stocks which have China revenue exposure, and in particular, Taiwan war exposure. Perhaps there is no stock more exposed to the latter than TSMC, so it may seem odd to take a position in it now. Yet as I wrote in Intel Earnings Pushes the Stock Down 12%, it is extremely unlikely that China will invade/blockade Taiwan prior to the 2024 US presidential election. It’s not just that Biden has let slip on multiple occasions that he would defend Taiwan militarily, but also more recently how China has leaned into its reliance on advanced manufacturing exports (because of overcapacity) to propel the economy this year and next. If China invaded Taiwan, the economy would be so badly hurt that it would put the Communist Party at grave risk, which is the ultimate barometer of whether something is likely to happen in China. The US election could change everything, particularly with Trump’s insistence on extremely high tariffs for all imports, especially Chinese ones. This is essentially why the position in TSMC is intended to be short-term, and still fits within the geopolitical risk model.
Risk On, Risk Off:
For much of this year my internal watch-list of stocks has skyrocketed up, while the stocks I own (other than Meta) have been stagnant or worse. Many stocks in my watch-list (in particular I am thinking about Nvidia, TSMC, Broadcom, and HIMS) did not fit into the three criteria of the original investment model. As a result I missed out on opportunities this year that would have outperformed the market.
The TSM 0.00%↑ trade is higher risk than my usual. TSMC could be the first to show cracks in the AI craze as their forward guidance (which is inherently further into the future by the nature of the foundry business) is more of an indicator of distant things to come than say Nvidia or Broadcom guidance. Nonetheless, based on my research from these AI giants and others, there is a high chance TSMC will beat and raise in their July earnings. Because of the competition for TSMC’s supply and latest node, it would be a dereliction of management duty to not have charged considerably more for the luxury of using their facilities when the customer of their customers are throwing oodles of money around for GPUs and AI infrastructure silicon made at TSMC. This has of course already begun to happen this year, but I see very few reasons why it would not continue at least through July.
Perverse Incentives:
For someone who wishes to make a career out of writing about the stock market, there are strong incentives to make more short-term trades and also to increase the risk of those trades. There’s just simply more to write about under these circ*mstances, and it’s exciting to consume content when one knows the author has skin in the game (e.g. observe the saga of RoaringKitty) on a high risk venture.
I am of the belief it is nearly impossible to know the truth behind why someone takes a certain action when the incentives differ from the stated reasons one acts — most especially for oneself. Therefore the judge of whether this shift in the model is warranted must be the quantifiable performance over time (and by that I mean market returns, not article views/subscribers). My success metrics will not change, even if the methods evolve. Perhaps the old aphorism that we cannot let the perfect be the enemy of the good should also apply to a theory on stock investing. Time to find out.
