Yesterday I wrote about a simple breadth system that has beat the market over the last ~18 years. This evening I want to describe another simple breadth system. I find it to be very interesting concept, but it is not going to set the world on fire, or make you rich beyond your wildest dreams.
The system uses an indicator which calculates the percentage of all major exchange listed stocks that are trading above their 50 day moving averages (PctAboveMA50) and 200 day moving averages (PctAboveMA200). All stocks, excluding OTCBB, trading above $1 are used to calculate the percentage. The calculation includes de-listed, non-surviving stocks, so the percentage 10 years ago should be very close to what it would have been in real-time, 10 years ago.
I observed that these metrics, particularly the PctAboveMA50, tend to be volatile. I used an EMA to smooth them, using a 5 period EMA on the PctAboveMA50 and a 35 period EMA for the PctAboveMA200. These EMAs were chosen through optimization, but varying the EMA periods didn’t make a huge difference.
Those two EMAs are what constitute the indicator that triggers the buys and sells.
The Rules
Buy SPY at the Close if:
- EMA(PctAboveMA200,35) crosses above EMA(PctAboveMA50,5)
Sell SPY at the Close if:
- EMA(PctAboveMA200,35) crosses beneath EMA(PctAboveMA50,5)
No commissions or slippage included. All SPY history used.
What interests me about this system is that it is counter-intuitive (at least to me). We are buying when there are fewer stocks above the 50 day average than above the 200 day average. Yes, after thinking deeply about what that truly means, it makes sense why it works.
The chart above shows the indicator and the signals.
The Results:
SPY –
- Compound Annual Growth Rate: 4.95%
- Winning Percentage: 71.93%
- Max System % Drawdown: -31.67%
- Exposure: 43.25%
- Number of Trades: 57
- Sharpe Ratio: .71
As I noted before, I like the concept more than the performance. My thinking is that it could be a jumping off point for something better.
The oscillator would work great for us “less-than-mortals’ WS.
Profound point. Sadly, I fight the habit of beliefs in outsmarting the tape.
I’m going to reflect on this idea.
I like it.
Once again, THANKS!
No problem Blind. Glad it was thought provoking. That was why I published it.
o..K
Buy/Sell MA convergences.
Not THAT wise WS.
Trying to be a Gentleman here.
lol…ok
So I gotta know; What software do you use to do your backtesting?
Amibroker with a data feed from Norgate Premium Data.
It’s interesting to compare this system’s performance to the previous one (on Nov 27), over the past two bear markets. This system performed poorly in 2000-2003, but well in 2007-2009; the other system did well in 2000-2003 but poorly in 2007-2009.
Sure wish you could find some happy middle ground 🙂 That would offer the potential to use leverage.
I noticed the same thing. In fact, I tried a buy statement using OR so that the system would use either buy criteria (the one from the 27th OR the one from this post).
I also tried to combine the sell statements with an OR.
Didn’t work very well 😉
I think it lowered the annual return to around 3%. Still, it was just a quick preliminary test. I would need to add more criteria show that the signals weren’t canceling each other out. There are likely better times to use each system, and we would want to weight the buy signals so that we are more likely to use the appropriate signal at the appropriate time. And then, it might be curve fit. 🙂
Thanks Wood. I love the work you post here.
Great post thanks