On Friday, SPY closed at a 50 day high. Today, the SPY pulled back a little more than -1.9%. Is this bullish or bearish going forward?
The Rules:
Buy $SPX/SPY at the close if
- Yesterday Makes a 50 Day Closing High
- Today pulls back more than -1%
- Sell X Days later
- $SPX data goes back over 50 years. All SPY data used.
- No commissions or slippage included.
The Results:
Analysis of Results:
Sample sizes for the baseline setup (blue and red lines) were decent, with 45 $SPX trades held the full 100 days and 20 SPY trades held the full 100 days.
To make the modeling closer to our current environment, I added the variable of a close beneath the 200 day moving average (MA200). Adding this variable significantly reduced the sample size to 9 $SPX trades and 6 SPY trades.
The bottom line is that even with the market trading beneath the MA200, a 50 day high and shallow pullback is bullish.
I was curious though how allowing for a deeper pullback would change the results. To that end, I required the pullback to be more than -1.9% (much like today’s pullback).
Pullback More than -1.9% Results:
Allowing for a larger pullback significantly reduced sample size to 13 $SPX and 6 SPY trades held for the full 100 days. Adding the variable for a close less than the MA200 reduced samples even further, leaving only 4 $SPX trades and 1 SPY trade held for the full 100 days. The single SPY trade was from 11.7.2002 to 4.3.2003.
While a deeper pullback still yields solidly bullish results going forward, the initial pullback appears to be deeper and last longer than the shallower pullback (to see this effect, compare the first 10 days of each graph).
I added the pullback variable to more closely model recent market action, but it is likely that the most important variable here is the 50 day high. It doesn’t appear to matter all that much whether this high comes in a bull or bear market. After it is made, the S&P 500 has tended to continue higher.
Good work wood. Have anybody in history been able to come up with a reasonable model for making profitable trades. Are you doing a lot of experimental work with these things. Apparently Isaac Newton lost money in stocks too. Can we scientifically trade this market. Or is it wishful thinking like my trading, hoping that someday I will make money in stocks on consistent basis.
Fish, there are ways to trade these sorts of models. Rob Hanna over at Quantifiable Edges has a system that basically takes these kinds of studies, or groups of them, and uses their aggregate results to make decisions. It is very neat the way he does it.
Beyond that, there are many different things that work, over many different time frames. The thing is, they don’t always work, and sometimes they really work well. The problem is continuing to trade them when they aren’t working well so that you absolutely catch the periods when they are working very very well. I’ve been watching a very simple system that I designed over 3 years ago, and it is doing very well, making new highs recently. I should be trading it, but I stopped when it went through a drawdown. Had I not stopped, I would be sitting pretty right now.
Find a simple edge, and have enough discipline to keep trading it as it is supposed to be traded. That is the key.
QE is a good blog – worth checking out.
And you nailed it here, Wood – it’s about the discipline of sticking to a plan combined with proper money management. We humans tend to fiddle with systems and when they don’t “work” immediately, ie have a losing trade, we often either discard or try to change, when in the long run, there’s gold in them thar hills.
That’s my personal Waterloo fo’ sho’
Wood, the problem is in guessing whether you are going through a temporary drawdown in a system, or whether you instead have a system that has stopped working permanently. There have been numerous systems that have stopped working permanently, so that’s always a risk. In fact, most of them are that way it seems. They are really good if the market stays in a certain pattern, which it does for a while. Once the market gets out of that temporary pattern, the system is scheisse from then on.
Woodshedder, in the text you are testing a pullback greater than 1 and 1.9 %, but in the pictures the pullbacks are smaller than 1 and 1.9 %, according to the title.
I guess the text is right, but could you confirm whether the test results are calculated with a bigger or smaller than 1.9% pullback?
Thanks.
Learner, it is < -1% or < -1.9%.
In other words, it pulls back deeper than -1% or deeper than -1.9%. Sorry for any confusing with the terminology.
Thanks, Woodshedder.