Trading the S&P 500 and Nasdaq 100 with the 5-10-20 Strategy
As promised, I’m following up on this week’s 5-10-20 Trend-Following Strategy to look at how it has performed on both the S&P 500 and Nasdaq 100.
S&P 500
In a nutshell, applying the strategy directly to the S&P 500 hasn’t work very well. What worked a bit better (but still not as good as traditional longer-term 50/200-day moving average crossovers) is using the Nasdaq Composite to generate the signal, but placing the actual trades on the S&P 500.
To illustrate, see graph below of the S&P 500 (blue), this modified 5-10-20 Strategy (red), and a 50/200-day crossover strategy (green) from 1972:
[logarithmically-scaled]
And for the number-lovers:
Both strategies reduced downside volatility and improved risk-adjusted returns, but the 50/200-day strategy clearly performed better historically.
I am considering adding the 5-10-20 strategy to the State of the Market report only because it has improved the predictiveness of the 50/200-day crossover strategy a bit (meaning, the market has performed better when both are signaling a long than when just the 50/200-day is). Will ponder this a bit more.
Nasdaq 100
The strategy has worked about as well on the Nasdaq 100 as on the broader Nasdaq Composite. Results have been slightly better using the modified approach discussed above, generating the signal using the Composite and then applying it to the 100, but the difference hasn’t been significant.
This is my last post (for now) on the 5-10-20 trading strategy. I hope readers have gotten a useful nugget out of all of this. I have it on the to-do list to put more of the indicators discussed on the dk Report to the test because many of them are in the same vein as what I do here at the MarketSci Blog. As always, more to follow.
[Edit: click for a summary of all related posts in this 5-10-20 series]
Happy Trading,
ms
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Trading Strategies | 3 Comments
So what account for the huge behavior differences in these strategies comparing the S&P500 vs the nas composite?
Over the longterm the 10/20 is better than the classic 50/200-day EMA crossover on the nas composite, but not better on the s&p. And your graphs showed the 10/20 doubled returns on the nas vs buy & hold, while the 10/20 barely surpassed buy & hold of the s&p.
Is the Nas that much more volatile? Or does the big drop in 2001 account for some of the behavior?
RE to gregh: part of it I attribute to just dumb-luck/curve-fitting and part of it I attribute to a fundamental difference between the indices. Is it because the Naz is “faster” or more “volatile”? I can’t really answer that. But based on the numbers, I know which MA crossover version I would employ for each index. Good comment. michael
Hi Michael,
I have found that several systems I have tested in the past perform well on the NASDAQ and the Russell 2000 but fail to work on the Dow or S&P 500. To combat this I have taken to testing systems on 16 different global indices. I would test on more but the historical data is difficult to find. One index that I think is very useful to include is the Nikkei 225 because in my mind it is the ultimate acid test for a system due to the fact that it is down 70ish% over the last 20 years. If a system can pull a profit from all 16 markets then the probability of continued success in the future is good.
So far I have been unable to break down into a formula the numerical explanation for why the NASDAQ lends itself so much better to trading models than the S&P 500. Logic tells me it is due to volatility with low noise but measures of volatility have been useless in proving this theory. That is, when the S&P has been more volatile than the NASDAQ on average it has still has not produced good trades and when the NASDAQ has had low volatility it has still produced good trades.
Have you come across a measure that can identify when an Index lends itself to a trading model more kindly?
Keep up the great research and thanks for sharing. Your approach to the market is very different from mine but it is so nice to come across another numbers dork who extensively tests their ideas. You are one of the few mechanical trading model engineers that I think provides real value to their clients.
Cheers
Derry