For example MOMENTUM. This thing printed money from 1999 to 2008. But something happened. The 2008-2018 ten year period the factor doesn’t work anymore. Yes markets are becoming more efficient or maybe factors go through periods of over/underperformance. Value/growth is an example. Anybody care to comment?
Momentum and other factors still work. They just don’t work as easily as they did pre-financial crisis. Perhaps because of the mass acceptance of systematic investing, many of the straightforward momentum factors like buying positions that performed best over the last 2, 3, or 6 months - or the ones with the highest Relative Strength, don’t work any longer.
Today you have to get more sophisticated in your construction of formulas. It’s still possible to exploit many market conditions, but you just can’t throw a low P/E ratio up and hope to get good results.
An analogy: when I was a teenager I used to get good performance out of driving through McDonald’s and ordering a hamburger, fries and a Coke. I got lots of energy out of it and no problems. Now I’m an old fart, and if I ate that crap, I would crash and burn. Today I have to eat a low-fat diet with fresh fruits, vegetables, and salads or dinner at a top-end Japanese sushi bar. THAT gives me excellent performance now.
The same thing occurs with investment factors. Just like my body, investment using computers has matured. Therefore, you have to stay away from the hamburger, fries, and Coke of factors.
Fortunately, P123 accommodates sophisticated algorithms in system construction that can achieve high performance using more advanced approaches. I’m happy with my results, over the last four weeks I’ve received an 18% return from my ETF portfolio and a 42% return for my top stock portfolio, all while a turbulent correction played out. However, these strategies are my most sophisticated, multi-component, adaptive creations since I joined P123 some 14 years ago. They call on five external data series, 3 universes, 4 ranking systems, and six custom lists. All compiled into a carefully crafted machine that performs well regardless of what the market is doing.
You can get to the same place - if you are so motivated.
Chris
I have a momentum system and it is up 32% YTD. Fairly simple in concept yet some tweaks that 20 years of history support. Actually has done much better “live” in 2017 and 2018 than it did in simulations on data from 2000-2016. Sim performance from 2010-2016 was much better than years prior. I have “decent” ETF systems and am working on a value system to complement the existing systems. What I have found over the decades is that I have to have mechanical systems. And lots of careful backtesting to back them up. That gives me a feeling of confidence and sidesteps most of the fear and greed that I have.
As you can see in the replies, Momentum seems not to work in the way you try to capture it. Academically the factors work in a way, presenting an edge. In the real live, capturing that edge is not easy and it can be carried out through different methodologies.
As I shared in the other forum about momentum, this article will make you think about how to implement momentum in the systems.
I don’t think it should be a surprise that momentum doesn’t work as well in bull markets…200-day SMA (buy SPY when it closes above 200-day SMA, buy IEF below) underperformed B&H during the bull run of March 2003 to Jan 2008 also.
Where momentum helps is avoiding large drawdowns, which overall will improve your returns long-term.
Or are you talking about momentum on individual stocks?
Right I understand there might be versions that have some edge.
But the general form of buying the top 10% and shorting the bottom 10% (past 1 year adjusted return minus past 1 month) - this type of strategy has stopped working as well in the past 10 years. The strategy has poor return in the general form on a broad basket of stocks … so makes you wonder if it robust anymore.
In fact, there are 2 momentum “crashes” in the past recently. One in 2000 and one in 2008. Each crash took down about 3 to 4 years worth of “gains” from following the strategy.
A similar thing happened in “trend following” commodity hedge funds. CTAs have not had a great return since 2008 after the GFC compared to the golden age of the 80s and 90s and early 2000s.
I understand for hundreds of years trend following/momentum has made serious money.
I’m just hypothesizing maybe this factor has been arbed away by the Renaissance and Citadels of the world.
Here’s where I stand. Trend-following (or momentum investing) “works” on the level of the stock market as a whole and on the level of sectors and industries. It might work to some degree on the level of individual stocks, but less so during a strong bull market and less so for small caps or microcaps. But what one must focus on is the reason for it, which you must clearly articulate. A lot of people use price momentum as a substitute for quality. A stock that most investors think has strong growth potential will increase in price and volume, so momentum signifies growth potential, and there is indeed some correlation between long-term volume-weighted price increases and future EPS growth. On the other hand, a short-term price change could signify either a reaction to an earnings report (in which case pay attention) or a rumor (in which case it will probably correct). Using momentum as a signal on its own can be pretty problematic for individual stocks, and I prefer to use measures of quality, growth, and sentiment, which are often the causes of price changes, rather than the price changes themselves.
My very brief thought: I think momentum effects are still present, but they often don’t survive integration with the other factors in a model. This agrees with Yuval’s view above (if I understand it correctly) that momentum may be a secondary effect/phenomenon, and that the primary drivers of momentum are perhaps better captured and described with other variables if available.
This said, I think it’s still there in places if you look. For example, look at Toronto Stock Exchange universe. I haven’t worked much with the Toronto Exchange, so don’t know if the factor survives integration with multifactor models there, but the histogram looks nice. In US mkts though, traditional relative strength momentum is certainly more vanishing.
Just my random thoughts. By no means have I studied this intensively, so ymmv.
I still use momentum because I am not sure if it is dead or not. Better to muddle than miss out.