Perhaps because of my research background, I am comfortable with reading academic research paper. One such paper is Market Cycles and the Performance of Relative-Strength Strategies after it was recently reviewed by Cxoadvisory. Among others, the paper investigates whether stock level relative strength (momentum) strategies and industry level momentum strategies are dependent on market cycles.
Reading such paper by academic, I usually just quickly scanned through the paper without going into detail (unless something caught my attention) and go straight to the results: figures and tables. Fortunately, academic usually put a self-explanatory graphs and tables. If the explanation is not sufficient, then I will go back to the main report to search for more explanation. Usually, by reading the tables and graphs, you can relate with the conclusion drawn. Sometimes, you can infer your own other conclusion not mentioned in the paper.
As a practical trader, our interest is in knowing whether (1) the strategy in principle was shown to have worked in the past. If it is, then we can add our bells and whistles to incorporate what we already know about market behavior. More importantly, we want to know (2) under what conditions the strategy fails.
Now, based on the two graphs shown above, what can we say about those two questions, assuming the author have produced accurate and unbiased results? The y-axes read percentage returns.
For the first question, the answer for individual stock momentum with 6 months horizon is clearly “Yes” as the lines are mostly above zero. That is, 6 months stock leaders are more likely to outperform 6 months stock laggards.
Just by eye-balling, we can see that industry momentum in 6 months horizon did not work very well. That is, 6 months industry leaders are not more likely to outperform 6 months industry laggards. Other academic papers showed that industry momentum worked best in 1 month horizon.
More importantly, the answer to the second question is also rather clear-cut. Momentum strategies (both at stock an industry level) worked most miserably when the market turned from bearish market to bullish market. This is more evident in individual stock level. You can see all the negative spikes coming after the red lines switched to zero.
To relate to our practical experience, after panic selling, the best performing stocks at initial rally were those stocks which were most oversold. This is about the only time that buying laggards is rewarded handsomely.
What you read here may not be what you read from the paper conclusions. Sometimes, you just have to look at the fact and draw conclusions meaningful to you. The academic usually took extra care in data collection and data cleansing. Their data history is usually also very long. We thank them for producing the output. But our mindset is usually rather different. So what we look for and what they look for is rather different.