Today marks the first time since early 2008 that the 200 day simple moving average (dsma) of [[SPY]] has turned up.
I have been waiting for this development to test a very simple trendfollowing strategy. The strategy will go long if the 200dsma is higher than the previous day and will sell the long position and go short if the 200dsma is lower than the previous day.
But before I get to this final test, I want to examine and discuss some variations.
The Method:
Variation 1: Buy the SPY if the close is above the 200dsma and the 200dsma is higher than the day before.
Variation 2: Buy the SPY if the close is X % or more above the 200dsma and the 200dsma is higher than the day before.
Variation 3: Buy the SPY if the 200dsma is higher than the day before. Sell short if the 200dsma is lower than the day before.
***Edit*** I forgot this important bit of information…The exit signal is generated if the 200dsma is lower than the day before.
All variations compound gains (this is to make for an accurate comparison to buy-and-hold investing) but do not include commissions and slippage.
Variation 1:
Summary:
Of course this is a small sample size, so all the usual caveats apply.
The average winner is 38.92% and the average loser is -1.35%. That is fantastic.
Annualized rate of 9.55% beats buy and hold by 4.81% over the same period with 1/3 less exposure.
Max system % drawdown of -19.03% is significantly less than the buy and hold drawdown of -56.47%.
All in all, not a bad way to invest with a buy and hold mentality.
Tomorrow I will post the results of variations 2 and 3.
Ahh man.. Bear steaks for labor day?
Hard to say Goldie. Pretty good setup though. I would like it better were it not so far extended over the 200 day. That’s why variation 2 will be interesting.
I did finally get a long setup for tomorrow. Probably means we’ll be down 3%.
A while back I did a similar study on the 20 month moving average, with reasonably good results. I think a basic trend following system like this has a place in a portfolio for the long haul… its the quant traders version of buy and hold.
Good stuff, shed.
i love this one. really crazy how it beats down a straight buy and hold.
also, i laughed for an hour looking at that douche bag with hot chicks link you put up. classic
Glad you like guys. I’m pretty sure variation 3 is going to be even better, but we’ll just have to wait and see.
Jim, http://www.hcwdb.com is an awesome site and highly recommended, especially when you have some beer and some time to waste.
2 more days gentlemen. 2 more days of green, then pure red that is going to wipe everyone out. Tomorrow and the next day will both be up days, one day will be a big up day and the other will be similar to today.
July 30, 2009
Please provide a rationale for your statements.
They may very well be correct, but give us a clue as to your thinking.
You don’t work for GS do you?
Wood,
Great post as always!!!
I’m looking into a long term trend following Index System that simply goes long the market when the RSI is above 50, and goes short when the market is below 50.
Following the DOW and the FTSE for long term positions.
I’m totally new to back testing, and the system I use doesn’t seem particularly intuitive. Would the system generate a profit at all?
EnglishGent, intuitive means little as if it is intuitive, it probably makes sense to many others as well. Therefore, the edge may be traded out of it.
I will test your idea and get back to you. Is the email address you provided correct? If so, I’ll email you the results, unless there are others here that would be interested in seeing the results.
However, my best guess is that it will be marginally profitable, but will not beat buy and hold.
Which RSI period setting did you want to use? 2? 5? 10? 14?
Wood, Thanks. I look forward to the updates.
Thanks Wood, I was using RSI 14, and e-mail address is correct
I’m not really a fan of buy and hold, and wanted to try to make money on the uptrends as well as the downtrends using a system that would not require more than a few minutes each day, and would not whipsaw too much.
This would just make up say 5% of my portfolio, but would generate a pretty decent return.
I was potentially going to incorporate MACD into it, but the backtesting system I’ve got is a nightmare!
The only issue I see is in an untrending market which will lead to drawdowns. However, i’m hoping that by keeping it to the DOW and FTSE indices these would be minimal.
EG, doesn’t look good. Results over the DIA are -9.27% annualized. The system is currently drawndown over 70%.
Results on the SPY are a little better, but still showing a negative annualized rate.
I did some testing one time that shows RSI14 to be pretty useless with no discernible edge.
Seems like the base case (“Variation 0”) would be long while SPY > 200dsma. Lots of folks use the 10-month or 200 day as a simple trend identification. Your test represents an advance on that, since the simple ‘above the 200’ strategy gets whipsawed terribly at transitions.
Eywwww! That is not the avatar/icon I would have chosen! (I guess they are trying to sell their ‘gravatars’.)
Yours is not as bad as EnglishGents!!! lol
By the way, the Gravatars are free. Just pick a picture of your choice, resize it, and upload it. If not, when you come to my blog, you will forever be the blue heart bunny ears with snakes dude or dudette.
well you have chosen a time period when it works. You niftily left out the 1994 triggered trade which was a loser. Try going back to 1930 or 1920 and see what you get; it’s not as good as it appears
PA, where do you people come from? Is there a troll rock someplace under which you hide?
Hmmm. I test over the entire data available for the SPY, which just happens to contain 2 of_the_nastiest_bear_markets_in_100_years, and you accuse me of purposefully manipulating the data?
Furthermore, I am getting ready to publish a variation that goes short, which, by your comment “niftily [sic] left out the 1994 triggered trade which was a loser” I assume you are a permabear, or don’t believe system trading works. Well I’ve got news for you, the losing trade you speak of will be a winner on the short side.
Do not forget that to make a 200 day average, you have to have 200 days worth of data…So the system cannot begin trading until 200 days. Did you consider that might be reason there wasn’t trade in 1994?
Finally, why don’t you tell me what happens on the SPX in 1930 or 1920?
My guess? You have no idea as you do not have the data or do not know how to test it.
Prove me wrong- post your work.
Silly Woodshedder; trolls live under bridges, not rocks…
You have thin skin! You don’t have to blow a gasket, I didn’t accuse you of anything. All I implied is that the recent data skews it more positive than it is over a lengthier period. If I can find the time I will post older data. In the meantime, chill out a bit. Peace
Uh, you did accuse me when you wrote that I “niftily [sic] left out the 1994 triggered trade which was a loser.”
However, due to you mentioning “blowing a gasket,” I might have to let you slide, as that is a favorite idiom of mine.
I have the older data, SPX going back to the 1920s. Maybe you are new here, if so, you wouldn’t know that a lot of people make assertions that are not backed up by data, and surprise surprise, most don’t even have the data to make the assertion in the first place.
Peace to you as well.
hi wood, nice work. it seems to be robust across other securities other than just the SPY. perhaps weekly or monthly data might be a new twist……
dv