Just a few points that I’d like to ask you to clarify:
How the Dynamic momentum algorithm selecting exactly 10 stocks out of the SP500? Is the number of stocks (ie 10) set by you or chosen by the algo?
In the logarithmic graph showing the magnitude of the most recent SP500 drop from March 2020? How would the Dynamic Momentum strategy do if started just before the crash of March 2020? I am asking this since we are in a bull market now and most markets are growing if one invests on the longer run, so most long only strategies should work. Especially if trading good quality stocks already in the select SP500 club.
Can you show us some historical examples of what kind of SP selection we had in 20016 and 2019. To have an idea if / how the hyped growth stocks are changing over time.
And lastly, is there a way to get periodic updates on the SP 500 selection? Would like to have a ‘crutch ‘ when doing the pickings , rather than using the gut.
On your question regarding the number of assets we select, it's based on the point of diminishing returns. For instance if we choose 15 assets instead of 10 and we count how many have a correlation greater than 0.7 we have 11. In the example with 10 stocks we have only 2 - I've got some images to showcase this, but it seems I can't add them in a comment.
Remember that we are aiming to get the least correlated assets from one index, from one single country. We are already biased in this selection. I think most of the time diversification is misunderstood from this point of view, as adding more stocks from the same economy to a portfolio doesn't always decrease risk, it might even enhance it, especially if it's done using the same strategy. As a matter of fact the results using 15 stocks are poorer than using 10 stocks simply because we are trading more and that adds a lot of cost.
On your second question regarding the drop in 2020, I actually went on and ran the algo from 2008 and compared it with the returns of the SP500. As you can see the overall loss was pretty similar that year. The March 2020 drop was small in comparison, around 5% and the illiquidity crisis didn't even compare to the one in 2008. One interesting point to make here is that the correlations increase in times of illiquidity, so if you remember the only asset that went up in March 2020 was the USD, while stocks, bonds and precious metals all became highly correlated and were in a free fall.
On this point remember that diversification is important as you don't want to be 100% in stocks, maybe have some cash on hand so that you are not forced to sell your assets at the worst possible time.
Year S&P 500 Index Dynamic allocation algo
2020 16.26% 86.00%
2019 28.88% 43.00%
2018 -6.24% 31.00%
2017 19.42% 32.00%
2016 9.54% 30.00%
2015 -0.73% 21.00%
2014 11.39% 31.00%
2013 29.60% 47.00%
2012 13.41% 28.00%
2011 0.00% 3.00%
2010 12.78% 39.00%
2009 23.45% 15.00%
2008 -38.49% -38.00%
Average return 9.17% 28.31%
As an example the 10 stocks that you would own in January 2016 based on this strategy would be: AMZN, GOOGL, NVDA, NFLX, ATVI, GPN, DXCM, EXPE, TYL, ABMD and January 2019: LLY, AMD, HCA, CMG, DXCM, ENPH, ETSY, PAYC, FTNT, AAP.
One interesting thing to note is that the algo gets all the rotation in the sectors. So for instance in March 2020 the selection was: AAPL, TSLA, AMD, TGT, LRCX, KLAC, DXCM, ENPH, TER, QRVO.
On your last question, the answer is yes, I will publish and article with the algo's selection each month.
Great learning from your article!
Just a few points that I’d like to ask you to clarify:
How the Dynamic momentum algorithm selecting exactly 10 stocks out of the SP500? Is the number of stocks (ie 10) set by you or chosen by the algo?
In the logarithmic graph showing the magnitude of the most recent SP500 drop from March 2020? How would the Dynamic Momentum strategy do if started just before the crash of March 2020? I am asking this since we are in a bull market now and most markets are growing if one invests on the longer run, so most long only strategies should work. Especially if trading good quality stocks already in the select SP500 club.
Can you show us some historical examples of what kind of SP selection we had in 20016 and 2019. To have an idea if / how the hyped growth stocks are changing over time.
And lastly, is there a way to get periodic updates on the SP 500 selection? Would like to have a ‘crutch ‘ when doing the pickings , rather than using the gut.
Hi Ylly, many thanks for the questions.
On your question regarding the number of assets we select, it's based on the point of diminishing returns. For instance if we choose 15 assets instead of 10 and we count how many have a correlation greater than 0.7 we have 11. In the example with 10 stocks we have only 2 - I've got some images to showcase this, but it seems I can't add them in a comment.
Remember that we are aiming to get the least correlated assets from one index, from one single country. We are already biased in this selection. I think most of the time diversification is misunderstood from this point of view, as adding more stocks from the same economy to a portfolio doesn't always decrease risk, it might even enhance it, especially if it's done using the same strategy. As a matter of fact the results using 15 stocks are poorer than using 10 stocks simply because we are trading more and that adds a lot of cost.
On your second question regarding the drop in 2020, I actually went on and ran the algo from 2008 and compared it with the returns of the SP500. As you can see the overall loss was pretty similar that year. The March 2020 drop was small in comparison, around 5% and the illiquidity crisis didn't even compare to the one in 2008. One interesting point to make here is that the correlations increase in times of illiquidity, so if you remember the only asset that went up in March 2020 was the USD, while stocks, bonds and precious metals all became highly correlated and were in a free fall.
On this point remember that diversification is important as you don't want to be 100% in stocks, maybe have some cash on hand so that you are not forced to sell your assets at the worst possible time.
Year S&P 500 Index Dynamic allocation algo
2020 16.26% 86.00%
2019 28.88% 43.00%
2018 -6.24% 31.00%
2017 19.42% 32.00%
2016 9.54% 30.00%
2015 -0.73% 21.00%
2014 11.39% 31.00%
2013 29.60% 47.00%
2012 13.41% 28.00%
2011 0.00% 3.00%
2010 12.78% 39.00%
2009 23.45% 15.00%
2008 -38.49% -38.00%
Average return 9.17% 28.31%
As an example the 10 stocks that you would own in January 2016 based on this strategy would be: AMZN, GOOGL, NVDA, NFLX, ATVI, GPN, DXCM, EXPE, TYL, ABMD and January 2019: LLY, AMD, HCA, CMG, DXCM, ENPH, ETSY, PAYC, FTNT, AAP.
One interesting thing to note is that the algo gets all the rotation in the sectors. So for instance in March 2020 the selection was: AAPL, TSLA, AMD, TGT, LRCX, KLAC, DXCM, ENPH, TER, QRVO.
On your last question, the answer is yes, I will publish and article with the algo's selection each month.