Bivek Neupane: Is the MA Crossover Strategy Even Profitable?

- Bivek Neupane

Today we shall be doing a fun exercise. We shall be questioning one of the most well-known trading strategies: the infamous moving average crossover. In this article, I am not going to go through the details of what a MA crossover strategy is. If you do not know the meaning, just stop reading this and google the term and then come back later because from now on what follows assumes that you at least have knowledge about how the strategy works.

Research question

1. Does the MA crossover strategy generate positive returns? If yes, how much?
2. What about risks associated with the strategy?
3. Is it worth trading on the signals generated by MA crossover considering all the transaction costs and tax implications that come along while actively trading?

Research methodology

Essentially, we are putting this strategy to test on the historical data. We are just back-testing to see how the strategy would have performed if we had implemented it from the very start. The study shall be conducted on a sample of data that spans from 1997-07-20 to 2020-12-23. It contains 5369 daily observations. I think this is a pretty sufficient amount of data to conduct a brief study such as this.

Note: There is one caveat that we must first keep in our head before we proceed and that is we are supposing that there is an index mutual fund product that we can easily buy in the market. This mutual fund replicates exactly what happens in the NEPSE. Also, the tracking error is very low.

For your better understanding of what we are exactly looking at, here is a graph for you:

So, what does the result show? Let's look at this table below:

Interpretation

Let’s start with the benchmark i.e. NEPSE. If we had just bought and held the shares of our hypothetical index mutual fund, we would have earned 8.86 % compounded annual returns with an annual volatility of 24.79% which is pretty substantial actually. Similarly, Sharpe Ratio is just a risk-adjusted return. It means how much additional return is your strategy generating per unit of risk. The larger the number, the better the strategy.

An interesting thing that we can notice here is that SMA 20/50 actually outperforms SMA 50/200, though the difference is fairly marginal. Furthermore, we can also see improvements in volatility (it has decreased thinly) and Sharpe Ratio (it has increased).

Q) Is it worth it to stick to this strategy?

In this study, I have completely ignored the transaction costs and tax expenses. If I had included these two factors in the study, I am pretty sure the annualized return from the SMA crossover strategy would be well below Buy and Hold. Next, the standard deviation will not be the same as the Buy and Hold if I had used individual security like NABIL bank or any other company. This is solely because the individual stocks are way more volatile than the whole aggregate index. This applies in our case because we do not have products like index mutual funds.

Furthermore, the strategies like MA crossover fall into the category of momentum strategy which means these strategies work when markets are trending. If the markets are choppy and going sideways, then you are bound to lose money. This kind of strategy fails to generate any returns. Hence, the reason why I used the last 20 years of data.

Now, I am not here to say you should not use this strategy or you should go all-in for it. Since I do not have any idea about your risk tolerance and your return expectation, I cannot give you any kind of investment recommendation. What I can definitely do rather is to show you the evidence. This is what the data shows us and now you got to decide what is best for you. 

Here are some more graphs. The first graph below shows the interaction between SMA 20/50 and NEPSE index. Similarly, the second graph shows the details of SMA 50/200 vs NEPSE.

Article by Bivek Neupane

Bivek Neupane is a MSc. Finance and Economics student. He is also a CFA candidate. He is specializing in quantitative finance and research. His other interests include Factor modeling, Portfolio optimization, Behavioral finance, Alternative asset management, etc. Connect with him via LinkedIn or his blog. Check out his GitHub page if you want to access the R code (https://github.com/biv-neupane7/NEPSE-stat-analysis