The Power of Trend-Following Alongside Discretionary Trading

In my last several updates, I have written extensively about the bubble-like condition in AI, with a particular focus on Nvidia. Nvidia’s current pricing has already discounted an enormous amount of good news, so I naturally feel that the market risk is heavily skewed towards the downside. It is exactly the sort of scenario that is perfect for option strategies, and as you can probably guess, I have put on several limited risk strategies that will make significant profits if we get a sharp downward price correction, with only minimal losses if the stock continues climbing.

In fact, I have put on similar positions on a number of stocks, so Nvidia is far from alone.

I have also started to build positions in Advanced Micro Devices (AMD), Arista Networks (ANET), Coinbase (COIN), META, Netflix (NFLX), and several others. Each of these stocks have had wild runs over the past thirteen months, and it is sensible to expect large corrections before too long.

I could well imagine a variety of triggers for such a correction, although we don’t actually need a trigger to have a major sell-off from such elevated positions. For example, AMD has more than tripled, ANET nearly tripled, META has rallied four-fold, and COIN has rallied six-fold. My trend-following system has performed brilliantly, capturing the lion’s share of these moves, albeit with risk-adjusted sizing due to the hyper-volatility of some of these stocks.

Two-month options in Coinbase, for example, are trading at roughly an 85% volatility. ANET options with the same maturity are trading at a volatility of 65%, and META options are relatively mild in comparison with a 33% volatility level. Risk-adjusted positioning means that the more volatile stocks get smaller notional-sized positions. My notional amount of META stock will be roughly twice as much as my exposure in ANET. Their respective impacts on the portfolio will be similar as META will probably move about half as much ANET. Compared to currencies, which are trading at volatility levels between 5% and 8% in the same period, stocks are flying around like wild beasts on amphetamines. Gold is also relatively stable. Crude oil, a bit less so.

Aside from the hyper-volatility in the stocks, they have also been trending beautifully. This trending factor, more formally known as auto-correlation, has been the source of a very steady revenue stream for a few years now. For example, we entered a long position in AMD on the fifth of January 2024, and we got an order to sell on the opening Tuesday morning, February 20. The trade earned about $35, but because of the high volatility of the stock the position wasn’t huge. We got similar flattening orders in a number of other stocks today, so our positions are much, much lighter than usual.

In Coinbase, we got long on the 9th of February at $131.50, but the long position there, as well as the long position in Nvidia, got flattened today, locking in some very nice profits. No, we did not exit at the top, but we don’t expect to ever get in at the bottom or out at the top. Rather, the system is designed to capture the “meat” of the trade once the move is underway with some decent momentum. In Nvidia, we had been long since the beginning of the year, so we caught the move from $488 up to $692. I will be thrilled to have a few trades like that every year, even though I missed the top around $740. Conversely, we had been short almost the whole year in Tesla, until we got a recent buy signal. This locked in a $50 dollar profit even though we clearly missed selling at the top.

I am mentioning these things just to give you an idea about the power of trend-following as an excellent diversification from discretionary trading. There are going to be plenty of times when my discretionary trading is just flat-out wrong, or maybe early, and my systematic trading program should be there to capture the moves that I unfortunately miss. Personally, I find it aggravating that a computerized model can outperform me during parts of the year, but I love the challenge of beating it. I also love the fact that I have a “trader” working for me who diversifies my trading and jumps on trades that I would otherwise not capture.

I started developing and implementing these models over thirty years ago, and they still perform. I have made many adjustments to them over the years, but the reality is that they make money quite consistently. On average, they are profitable 9 months out of 12, and the average up month is bigger than the average down month. It is important to have a well-diversified portfolio that is risk-adjusted so that one instrument won’t dominate the rest of the trend-following portfolio. As noted, I know beforehand that I will almost never get in at the bottom or out at the top, but the model will nearly always capture a major move.

Stocks like Tesla are amazing performers. They have consistently captured 20% moves, or more, for years. Other stocks like Oracle and Netflix might seem to be “trendy,” but I have very small allocations to them because they tend to have large price gaps after earnings reports. That is a different type of market risk that is very hard to manage. I can either go flat the stock before earnings or carry very light exposures. I tend to go for the extra light exposures. Commodities also play an important role in the overall returns, although we need to have much larger positions in things like gold in order to compensate for the fact that gold moves about one fifth as much as stocks like AMD and ANET. The net impact on the portfolio gets normalized that way, so the relative portfolio impact of gold and AMD will be about the same. Put differently, capturing a 6% move in gold will bring the same benefits to the portfolio as capturing a 30% move in one of the more volatile stocks.

In stocks, I run exposure across many types of sectors. This means that I will have exposures to banking, construction, IT, alternative energy, auto, manufacturing, retail, and so forth. There will be periods when one sector is outperforming another sector, but over time, all the sectors will tend to make profits.

Currencies are normally a great diversification from the other portfolio allocations, but they have really been quite disappointing for a long time. For example, currency pairs like eur/usd and aud/usd have scarcely moved 3% the entire year so far. That is not only uninteresting, but it is also nearly impossible to generate much alpha with such a narrow trading band. I know from over thirty-five years of experience, however, that I still need to allocate some risk to foreign exchange because at some point, the currencies will have huge moves. They are generally not correlated with other asset classes so they provide a wonderful portfolio benefit.

Fixed income is another major asset class that gets a healthy risk allocation. Their correlation with stocks, for example, is sometimes positive, sometimes negative. That is ideal for a segment of the portfolio. Commodities provide similar diversification benefits.

By the way, one of my former top traders will be joining me in a few weeks as I am rebuilding my macro team to capture what we believe will be staggering opportunities over the next six or seven years. This trader played a key role in helping me generate over 300% returns during the Great Recession, and generate compounded returns at roughly 60% a year for many years. We anticipate even bigger opportunities coming shortly.

This is a short update as I wanted to explain a bit more about my trend following strategy. I hope this helps you understand why I was very clear about telling you to not look at my system’s exposures as specific trade recommendations. They are a small slice of a large, diversified basket. This strategy is very different from my discretionary trading which will focus on a dozen or so big ideas over the course of many months. Both strategies can generate excellent returns, but they are very, very different.

I will write again in a few days. Wishing you all the very best of luck.

Andy Krieger