“How to improve trend following performance” – David Lundgren

Andrew, nice to see you. Thank you very much. Thanks for having me on.

Of course, it's my pleasure. I was looking forward to being the first guest in your new home, but Brett Steenburger upended me on that. So yeah, he beat you to the post. Apologies for that. But thanks for joining us today. Now, where are you, of course?

Okay, excellent. Sorry, we're just having a technical issue there, but I think it's caught up now. So you're coming to us from Maine?

Okay, excellent. All right, well, looks like everything's running clear now, nice and clean. I can see it's going through. So let's get to the show. Apologies for all that mix-up. Now, how about we start? First, I know you've got some great stuff to share with us today, something that people might even think is a little controversial, but we're going to get into that in a sec. But how about first, can you share a little bit of background on yourself so that people get some context of where you're coming from and what you're going to share with us today?

Yeah, of course. So I've been in the business for about 30 years. My first job was actually as a stock broker up in Canada. My wife-to-be then was finishing up her schooling, so I moved up there expecting to be there permanently. Events unfolded in such a way that I ended up coming back here. But in that job at Dean Witter as a retail broker, I ended up sitting next to another broker who managed almost his entire book using technical analysis. It was mostly risk management, looking for buy ideas, but also vetting the brokerage firm's buy recommendations through the chart. He wouldn't buy a buy-recommended stock unless it had the proper trend setup. Up until that moment, I had no experience or knowledge of technical analysis. I saw it come to life right in front of me, and I've been hooked ever since. Over the years, I've been on the technical team at Fidelity, which was my first indoctrination into institutional money management, where the mindset is fully invested in relative performance. Prior to that, it was mostly futures trading. I left Fidelity to start a hedge fund with a few folks from Fidelity, which was a great learning experience. I left the hedge fund to start my own research firm, Breakaway Research. That's really where I started to get into what I would call systematic investing, though I don't have a PhD. I developed the models that I currently use today. Through those models, I found myself at Wellington. They were my client for a couple of years, and they eventually convinced me to shut my firm down and join them. I was there for 13 years as a co-PM on the portfolio, advising other portfolio managers. My big change today is having left Wellington in December. I am now making this cavernous change from managing money for others to trading for a living, which is pretty exciting. Very different, all of which I expected, but it's just different in ways I didn't expect. I'm pretty excited about it.

I mean two things. One is, maybe go to the second page of the slides that I sent. Let me get this one here. Actually, go to the next one because it's marked up a little bit more. Most of your viewers are very familiar with the momentum factor, so I won't belabor what it is. This chart shows the 10 deciles of momentum back to 1927. Those interested can get this data from Ken French's website. I converted the monthly stream of returns into a portfolio for each decile and plotted it. A couple things jump off this chart. One is that most of these deciles are up and to the right, so there's an upward drift to the equity market. The pink line in the middle is the average return, which would represent the market. What's really interesting about the momentum factor is that when you decile the market based on its historical returns and then invest in each decile separately, the forward returns of the decile based on preceding history actually matches the decile it started in. In other words, if you invest in the top decile of performance, that top performing decile continues to be the top performing decile. This is not always the case with value, where the charts are much more like a bowl of spaghetti. Here, these are almost in rank order of the decile they started in, which is pretty fascinating. But when I say momentum doesn't work, first of all, I've been looking at charts for 30 years, and I'm always looking for anomalies. When I was putting these charts together, what struck me was that on the far left side of the chart, most of the decile returns went sideways. In that window of time, momentum didn't work. The top line, which is the orange line, is the top decile versus the bottom decile. It did work through that period, but it was quite volatile and had a lot of overlapping structure during that almost 20-year period. So it worked, but it kind of didn't work. And when you look all the way to the right, we know today we hear a lot of people talking about how momentum is not working today. You can see that the top decile versus all the other nine deciles has an upward drift, but it's laboring, very volatile, with sudden givebacks we call momentum crashes. Since the tech bubble, it's basically been a knife fight. There's been no real winner. You can't really say that decile one has been a big winner against any of the other deciles. This is why we say today that momentum hasn't been working. When I looked at this chart, it struck me immediately that those two periods where momentum wasn't working are the periods where there was no growth. In the 1930s, momentum struggled, which was a low growth, deflationary environment. Today, with the exception of US growth, specifically technology, most benchmarks are still in 20-year secular bear markets. Japan is in a 30-year deflation-driven bear market. Is it coincidence that momentum is struggling during this deflationary period? I'm not sure, but it's interesting. In the middle of the chart, momentum lined up perfectly well during the inflationary bear market. So it's interesting that during this period where we're talking about momentum not working, it dovetails with the last time it didn't work, which was that deflationary, depressionary environment in the 30s and 40s. When reflation hit after World War II, momentum started working again. If we think about how cycles work over time, we go from deflation to reflation to inflation to disinflation. If we just completed a deflationary cycle and everybody's talking about us growing to inflation, I would argue that we have to go through that reflation phase first. From a momentum standpoint, after 20 years of struggling, if we actually start to experience some inflation and growth, we might start to see the momentum factor generate some returns. That's one example of what I mean when I say momentum doesn't work. But what I really mean by momentum not working is that if you have a signal telling you to buy a stock or commodity, and that signal is wrong half the time or even worse, is that a good signal? I would say it's not. But we can look at the stream of returns for the factor over time and see that it works. Why is that? The reason momentum as a factor generates profit over time is not because of the signal. If you look at the next chart, page four, this is a chart of the SMH semiconductor ETF. In the bottom panel, you can see a momentum score. One of the problems with momentum is that it has very high turnover because the signal is off and wrong. You have to sell the stock once you buy it. That's the short-term mean reversion problem of momentum. You can see that as SMH was trying to make its way up during the 2017-18 period, the momentum score fell out of the top decile several times, so as a momentum investor, you have to sell that stock, generating turnover. Contrast that with trend following. Trend following gets you to the same stock. Trend following is when you have a series of higher highs and higher lows. With the first green arrow, trend following would buy this stock and continue to hold it until you have a series of lower lows. During that uptrend, trend following stayed with that stock, whereas momentum was forced to sell it several times. That's what I mean when I say momentum as a signal doesn't work. Momentum as a factor does generate profit. Why? If you go to the next page, it's because momentum as a factor has positive skew. The process of ranking the market, buying the top decile, and selling anything that falls out of the decile has built into it a stop mechanism that prevents left tail events. As soon as it falls out of the top decile, it's gone. One thing momentum can't help but do is capture things like Tesla when it goes into the top decile and stays there for several years. Momentum will catch it, but so will trend following. The point is that because momentum has such high turnover, I would qualify the signal as not working. When we credit trend following or momentum with working over time, we need to be clear that it's not the signal generating the P&L. It's the fact that the process has a positive skew. It won't allow left tail events to unfold because it's constantly selling anything that falls even a little bit, but it will capture those right tail events, which is critical to generating positive returns.

That's exactly what I would say. My conclusion is not to say momentum is useless. I actually use momentum quite a bit as an input to what I do. I'm a trend follower, but as I'm trying to determine which trends to engage with, if I'm looking at 4,000 stocks, in a very robust bull market like right now, 90% of the stocks in the S&P 500 are above their 200-day moving average. In a trend-following world, that seems very bullish. The challenge for a portfolio manager is how to winnow that down to 50 stocks. You can do that with momentum. If you take all the stocks in the S&P 500 and rank them according to their momentum, and only commit to trafficking in the top three deciles, that eliminates a lot of what you have to consider. You're trying to capture those right tail events. Tesla's not going to be in the fourth, fifth, sixth, or seventh decile. Even though it might be in an uptrend, many stocks have momentum scores down in the 30s and in the 7th, 8th, and 9th deciles because they're not keeping up with the top deciles. That's how I think you can use momentum. I have an example on page eight. This is looking at the technology sector broken down by industry. I have my own tech rank, which combines momentum, volatility, price structure, time frames overlapped, all coming into one momentum score. By doing this, I can immediately see that internet network solutions, with 10 stocks in that group, is ranked 19 out of 100 in the short term and 3 in the medium term. Although half that group is in an uptrend according to my model, I'm not going to spend a lot of time looking for ideas there because it's not the pond I want to fish in. I want to be fishing in leadership industries within a leadership sector. I find a leadership sector, like technology, and then the leadership industries within that sector, and then I focus my stock picking on those best performing stocks within those industries. By aggregating this for the 3,800 stocks I can look at, I can scale my research process and come up with a quick overlay of what's going on. Where is the best leadership? Where are the best trends? By systematizing my research process and putting it in tables like this, I no longer have to look at thousands of charts a day. I can do it quickly, knowing that these models capture what I would otherwise capture visually, but most importantly, they capture many more things that I would completely miss.

I just want to stress that our process at Wellington and what I'm doing today are the two clearest examples. Trend following works all the time in the sense that there are different time frames of trend. If we ever find ourselves concluding that trend following no longer works, this world has much bigger problems to deal with. If trend following doesn't work, that means there are no trends. Trends are driven by economics, growth, prosperity, and progress. As long as trends exist, trend following will exist. Whether momentum is able to capture that as a factor is a separate question. I detailed some of the reasons why momentum would struggle. The higher the turnover amongst the deciles causes issues. But as for trend following, I showed you an example of the semiconductor index that had a very powerful trend. Trend following did fine capturing it, but momentum as a factor continued to have difficulty staying with it. I don't lean heavily on momentum, but I'm very aware of it. At the end of the day, if I'm capturing something in my trend following process, I'm probably also triggering well in the momentum landscape. But it's not uncommon to capture something that's trending and outperforming by traditional trend following models, but having a hard time gaining and retaining traction in the momentum world due to turnover and other factors.

I absolutely do. This gets to the main reason why, while I'm intrigued by backtesting and try to learn as much as I can from it, I don't put a lot of weight on it. First, I'm not convinced that the data sets we're backtesting are reliable, certainly not pre-1990 or 1980. I'm suspect of the data. I've found more errors than I care to find. Any backtest from that data, I have a lingering suspicion. Second, I just showed you 100 years worth of trend and identified four cycles. We really only have two deflationary cycles, one inflationary, one reflationary. Do I really want to put a lot of weight on backtests when regimes determine outcomes? Quantitative easing changes the landscape and materially alters what happens in a backtested model. You can't detect that change because the impact is not readily obvious on the surface. When COVID hit, everybody knew about it and could see what happened. It's easily ring-fenced. But with QE and other slow changes at the margin, it's not that easy to ring-fence and know what to do. The reason I'm a trend follower is that if something is going to trend meaningfully, there are four things that must happen. They are unavoidable conditions that must unfold before you can buy something. You must break a downtrend, break the last high of the downtrend, the short-term average must cross the long-term average, and the long-term average must turn higher. Those four things must happen. I don't care if it's a value stock or has high momentum. If this stock is going to trend for a sustainable period, those four things absolutely must happen. It doesn't matter what the cycle is. The market, in its infinite wisdom, figures out all these changes. The market's verdict will appear via this checklist of trend change. By the time we figure it out fundamentally or through momentum models, trend following will pick it up. If you can wrap your mind around the idea that trends happen for fundamental reasons, if you see a trend changing and inflecting in a time frame driven by fundamentals, that's a durable concept. It will work forever. On the day it doesn't work, we have much bigger problems because that means there are no trends. Trend following will work. Sometimes momentum can augment it. There's a chart on page 11 that addresses your question. It uses visual interpretation of regime drivers to help you get a sense for whether momentum should be working, whether we should be in a bull or bear market. In the top panel, the black line is the 10-year Treasury, the red line is a proxy for inflation expectations, and the other is credit. When they're rising, you're in a risk-on environment. When they're falling, you're in a risk-off environment. The panel below shows banks versus utilities and copper versus gold. You can also throw in Aussie-Yen. When you look at those regimes, they coincide with cyclical bull and bear markets, momentum doing well, and trend following engaging. You can use the feedback from the market and the various trends unfolding in related market activity to gain a sense for whether you as a trend follower should be doing well. That's one of the most important things we can do as investors: wrap our minds around this definition of edge. I have a slide on page five. In the bottom, you can see edge. This concept of discovering if you have an edge is critical. It doesn't matter who you are or your time frame. Too many people go through this grist mill without realizing it. Edge is how often you're right times how much you make when you're right, minus how often you're wrong times how much you lose. If you come out with a negative number, stop doing what you're doing. But the concept of edge can be misguided. During a full cycle, even if you have an edge over a bull and bear market, during a bear market, the peak of your frequency of small losses shifts to the left, and you incur more losses. The number of outliers to the right dissipates. When you're not doing well, you need to go through all your trades and put them on a diagram like this to determine why you're losing money. Is it because of something you're doing? Are you taking too many left tail events? More often than not, when I've not done well, like this year, I came out of the gates with one of my best years ever, but in mid-February, momentum fell on its face. In that snippet of time, I could have thought I was a genius, then after February 15th, I'm sitting here going, 'I can't believe I just left Wellington. Am I crazy?' Those are the mental gymnastics I would go through if I wasn't in tune with what my process is supposed to do in different regimes. If I look at the regime and see that risk-on is doing well but starting to roll over, and I see my trend model deteriorating, how surprised am I that I'm not doing well as a trend follower? Mentally, I'm perfectly comfortable with the fact that I haven't done well since February because I understand why.

You trade less, take smaller bets, go on vacation. Jesse Livermore said in his book 'Reminiscences of a Stock Operator' that two or three times a year you really need to press your bets. Outside of those windows, the money is made in the waiting. From a trend following perspective, if you're a long-only trend follower, we're probably in that window where you just need to sit on your hands and wait for things to get better. If you're a long-short trend follower, there are opportunities cropping up on the short side. A lot of the formally strong charts from the reopening trade have rolled over. Those stocks still have great momentum ranks because they've done well over the past 12 months, but the trend is broken. That's the power of trend following over momentum.

I don't lose sight of any of those traditional indicators. But when I started Breakaway Research in 2004, I did a lot of modeling of the market to overcome what I thought were deficiencies in the way we were looking at it. Things like percentage of stocks above the 200-day average are undeniably useful. You'd rather have 90% above than 90% below. But the reality is that if you go to page 9, as of June, you can see in this table what I do every week in three time frames. On the left-hand side, you can see scores from positive 5 (best) to negative 5 (worst). In that week, 34% of the market had the best technical rank. When you say 90% of the market is above its 200-day average, that sounds bullish. But according to this model, only 34% are actually trending. The rest are in this knife fight we all experience. When you drill down on individual stocks, you can see that only 35% of the market was actually trending. The rest were in consolidation or an actual downtrend. Then you take that same model and look at what happened over the course of the week. The best bucket went from 34% to 28%. There's more deterioration. If I were to show you this model this week, it's 47% in an uptrend, and the biggest bucket today is negative 5. I've been watching this. The percentage of stocks above the 200-day average is at a record period of time, but I'm seeing an awful lot of deterioration in the market. This table allows me to see that without looking at 3,800 stocks. In the bottom left-hand corner, 51 stocks went from a downtrend to an uptrend or from consolidation back to an uptrend. I want to look at those. In the bottom right-hand corner, 125 stocks are now in the worst trend definition this week that weren't there last week. I'm definitely going to want to look at those charts. The high-level observation from the top left corner shows an improvement of 51 but a deterioration of 125. This market is deteriorating beneath the surface despite the record number of stocks above the 200-day average. This table reveals the knife fight we've been in all year long that you can't see by looking at simple things like percentage of stocks above the 200-day average. The advance-decline line is also important, but it has flaws. It can't capture breadth participation today like it used to in the 70s and 80s. You need to be creative and come up with ways to take the same data but look at it properly.

This was from when I put it together for the podcast in June. Now, the biggest bucket by far is negative 5, and a large part of that is the reopening trade. Those are the stocks that roared off the COVID low. Airlines, hotels, car rental, all that stuff. You're seeing a lot of violations of the trend change checklist. When I look at that, I know what those charts are doing without having to look at all of them. Today, despite the fact that we're still 90% above the 200-day average, something materially bad couldn't happen unless we crashed in a COVID scenario again. Any shorts I put on, I want to keep my leash short, but I want to put them on. Things have tilted more negative than what would be perceived by looking at the S&P 500 or the advance-decline line.

My answer to that question is no. That's not because what Elon Musk is saying is irrelevant to everybody. It's just irrelevant to me because I'm operating in a time frame that isn't influenced by what he tweets on any given day. I'm trying to capture trends that you can see on the weekly chart. I don't look at hourly or 15-minute charts. If I were still doing my day job as a technical analyst at Technical Data, sitting in front of a Telerate screen typing trading recommendations to traders on the Chicago Board of Trade floor, what Elon tweeted would be all I cared about. So it's not irrelevant; it's just irrelevant to me.

I'll give you a brief overview. There's a list of books at the back of my slides. The first four books are essentially what I consider myself a disciple of: Jesse Livermore, Nicholas Darvas, William O'Neil, Mark Minervini, and my former colleague Frank Desharnais. Those five people are among the best at what I do. It's trend following at its core. Starting from the top, by bucketing the 4,000 or so stocks into proper buckets by virtue of trend, I can almost always eliminate a third to two-thirds of the market just by bad chart structure and poor relative performance. I can then go through the rest and get rid of all the poor performing industries. Studies show that 50% to 80% of what happens to an individual stock is determined by what happens to the market, the sector, and the industry. If those things are drivers, I want to know what the industry is doing and what the market is doing. Through that effort, I can narrow down my list of focus to companies that are triggering. On page nine, those 51 stocks are ones I would want to consider as a new entry. I know what they're going to look like: reversing a downtrend, breaking a downtrend line, breaking through the last swing high, the 50-day crossing the 200-day, or the 200-day average turning higher. That's what triggers me to consider a buy. Then you take that exact process and flip it on its head for exits. If the industry starts to deteriorate, the market starts to deteriorate, and the trend model triggers negatively, those are the things that trigger me out of a stock. The example of SMH on page four is a perfect example of what would get me into a stock. The pink arrow shows that from a trend following perspective, there is no trend. It's an overbought range where you don't want to buy. A momentum investor would buy because the momentum score triggered into the top decile. A trend follower would not. That's the distinction between trend following and momentum, and why I'm a trend follower.

The systematic rules are basically position sizing, profit taking, raising my stop, and chasing a stop with volatility trailers. That's all systematic. For me, that's where I feel I have to be systematic. All the decisions I make pre-trade are discretionary based on a systematic research process. But I know myself well enough to know that once the stock starts going, I have zero problem selling if it violates. I don't care if it's in the ARK Fund or Bitcoin. I will sell it if it violates. If you look at page six, this is a frequency diagram of my trades. The heavy orange line is me stopping myself out. I have zero reservation about getting out of a trade. My issue is more about getting out when a stock is doing really well due to behavioral bias. I have systematic rules for when to take a partial profit, when to raise stops, and how big the position should be. That part is systematic not because I think it's the best way to invest, but because I'm protecting me against myself. I thought I was a genius in mid-February, but I was trying to tell as many people as possible that I knew I was overearning. I knew my process inside and out. I was actually concerned we were on the verge of starting another 1999 bubble because a lot of the ingredients were cropping up. I had funded some of my own models to try systematic investing. The model was really good at finding durable leadership. I decided to systematically invest in that portfolio in January, rebalancing at the end of every month. I would do half my portfolio that way and the other half the way I've always invested. Both portfolios were doing well, but the systematic one was destroying my discretionary portfolio. When things started to turn over, I started cutting positions in my discretionary portfolio, but I couldn't sell what was in the systematic portfolio until the end of the month. I watched a significant percentage of that profit evaporate. That was my foray into systematic investing, which I'll never do again.

I've come to appreciate that if you have a process that works, in the short term you will never be as smart as you think you are, but you also won't be as dumb. Right now, I don't feel particularly confident. I'm looking at my performance, up a little bit, but the market's up 16%. I know I'm lagging significantly. I'm thinking, 'I can't believe I got paid for 30 years to do this. I suck at this.' But when I look at page seven, this is the momentum factor ETF. There are periods where momentum just doesn't do well because the market is rotating and leadership is transitioning. It happens when the market itself is going sideways. Jesse Livermore said you only have to press your bets two to three times a year. Today, you can see how extreme momentum got into the peak in February and what it's done since. I look at my frequency diagram. The peak of my small losses is painfully large right now. I do have right tail events, so I am making money, but it's not enough to make me feel smart, but it is bad enough to make me feel stupid. Then I look at this chart and say, 'What do I expect? Momentum is not doing well. Am I going to blame myself?' I go through the process, go through all my trades, and make sure I'm executing properly. I do make mistakes because it is discretionary. I hone in on those mistakes and try to get better. This is how I backtest. I don't backtest 20 years of history. I go back and walk through every trade step by step, incrementally trying to get better, writing down my diary, and trying to be as disciplined as possible. When I look at it today, momentum is struggling. How can I be too hard on myself?

I have my Twitter handle: dlundgren3333. We do have a podcast for the CMT Association. Myself and Tyler Wood are interviewing CMT charterholders who are thriving with their charter. We've had the great fortune of interviewing legendary technicians like Bob Farrell, Ralph Acampora, Walt Deemer, and others. We just interviewed Frank Desharnais and Jeff DeGraff. Andreas Clenow was on your podcast at one point. I can't recommend his books enough. You're going to put my slides in your show notes. Just highlight the last page with a list of books I recommend. I'll finish with the most unorthodox book recommendation you've ever had: 'The Advanced Archer: How to Stay Calm at the Center.' It has nothing to do with trading, but it has everything to do with trading. I found it at a library where they were giving away books for a dollar. It's one of those books I read several times a year.

My pleasure. We're getting some thanks in the chat. One final question from Jeff: 'Do you plan on writing a book or offering a class?'

I would love to write a book, but the truth is that if you look at the book list, there are four books to start: Livermore, Nicholas Darvas, William O'Neil, and Mark Minervini. There's nothing I could write that they haven't already written. This is over a century of successful trend followers. Mark Minervini was on the IBD podcast the other day and made a comment I thought was brilliant. He had a whole slew of books behind him and said he read them all, but there are a few he read hundreds of times. He said he would rather read those books over and over again than read another book. I would just say those four books. There's nothing I could write that's not in those books. If you actually read them, they all say the same thing. The value is a hundred years of trend following by successful guys who have actually done it. Mark Minervini is in the trading championship this year, up 200% doing exactly what he wrote in his books. He credits Jesse Livermore, Nick Darvas, and William O'Neil. It's all the same thing. As for starting a class, I had the privilege of teaching a class at Brandeis University to graduate students for about five years. I've thought about doing a class and putting slides together. I'm actually considering starting my research product again because a lot of folks have reached out to me since I left Wellington. But in the interim, I can't improve on what the other four guys have already done. Mark Minervini is on Twitter all the time. It's classic stuff, time-tested and proven. Frank Desharnais has a public record. He's one of the best of our day and a cookie cutter of William O'Neil. I'm just trying to be some version of those five guys. Just read those books.

If I could chime in with one other thing, I know the time is coming up. The other thing you should re-read, but you can't re-read unless you do it first, is your journal. I've learned a tremendous amount from those books, but I've learned the most about the most important thing: about me. There are many successful value investors, quant investors, macro investors, and trend followers. All those processes work. So why are we struggling as individuals? It must be us. The best thing you can do is write a journal. Write down why you did what you did when you did it, as soon as possible after you did it. Try to be as verbose as you can. It's difficult to do because in the moment you think there's no value. But when you go back and read that a month or two months later, especially if the regime changes and you were struggling or doing particularly well, your learning curve will be so steep. Read those four books written over the past 100 years, and read the stuff you wrote two months ago.

Thanks, Andrew. We'll see you now. Bye.