Nvidia: A Bad Short
Having only been investing for a bit over 2 years I’m constantly looking for flaws in my investing process. I especially focus on the short side which I perceive to be at the same time more difficult and more entertaining from both an individual stock and general risk management perspective. Early this year, as junk rallied in January I isolated one name I saw to be ridiculously overvalued. Nvidia climbed 90% from its November lows and 60% in January alone as AI became the hot new fad. At around $210 I made it my largest short position, and it led to significant underperformance as the market rebounded downwards in February (which otherwise I was well positioned for). I have now trimmed it significantly down 20%. I want to explore what I got wrong here, why I trimmed it and what I think we can all learn from this.
The Thesis
The thesis for Nvidia was very simple, and seemed fairly obvious to me. This is a $600b company trading at over a 100 P/E multiple. Demand was slowing with the fall of crypto, loosening of the semiconductor supply chain and the exhausting of consumer demand for computers from stimulus. Between January 2022 and January 2023 revenue flatlined while operating income almost halved. Hype for the stock had exploded due to excitement around artificial intelligence. Nvidia is blatantly overvalued, and I still stand by that now. However, a company being overvalued doesn’t necessarily make a good thesis for a short.
Why was I Wrong?
The key thing to note here is that Nvidia is an amazing business. They have an absolutely market leading product with huge structural tailwinds and great margins. While I believe the AI hype is over the top, it’s objectively true that AI will likely continue to play a larger and larger role within society and that Nvidia will likely be a key player within that. The other thing to note is that this AI thesis isn’t a couple months out or a year out. This is a long term shift and potential investors are looking out many years ahead. The people that are buying and owning Nvidia don’t care about the decline in crypto or computer sales, their thesis barely cares about that. Even if the underlying business deteriorates, that doesn’t effect Nvidia’s long term prospects within AI. There’s a famous Cliff Asness line where he jokes about the irony of taking advantage of market irrationality and then getting upset when the market becomes more irrational. Why should the valuation come back down? Maybe in 3-5 years as we see the thesis unravel, but as a short seller with opportunity cost I don’t have time to wait that long. ‘I think this whole thing is stupid’ is not a thesis, and is how many short sellers have gotten blown out in things like crypto and meme stocks, especially when in an uptrend.
What Have I Learnt?
I think there is a very straightforward way to apply this to my own and others' portfolios. I’m still short Nvidia, however it’s now gone into my quant short portfolio (approx .3% of my book). I can deal with it not working out in that portfolio as I still believe it’s overvalued and probably a good risk adjusted bet. However if I’m going to make a sizable bet on a valuation short it better have a decent short term catalyst. Liquidity is the obvious one, and makes up the catalyst for most of my decent sized shorts. However, any thesis breaking catalyst would potentially work. Something being expensive isn’t a thesis for a short, no matter how expensive it is. Furthermore, even if you have a catalyst (which I did), it’s important to understand the motivations of the buyers and whether that catalyst is actually relevant to them. I also might investigate using a trend following strategy for shorts like this. Inherently a trend following strategy helps avoid getting blown out shorting against hype.