I’ve taken a short break from the ETF Rotational Strategy series to try and figure out what is going on with this recent pullback. To be clear, I want to short this market. (In fact I did, this morning). It seems I’m not the only one, as even our dear Blogfather is sitting on mounds of cash, waiting for a lower entry for deployment.
After today’s finish, which “looks” bearish to me, I decided to run yet another study to see if there has been any edge to getting short here. I just wanted to check my head, as revenge trading and acting on emotional whims is no way to beat the market.You might recall the previous study did not paint a bearish picture. Still, today’s setup is the gloriously bearish Honey Hole pattern, which iBC old timers might remember served us well back in 2007 and 2008.
In case you were not able to avail yourself of iBC wisdom back in tha day, the Honey Hole pattern comes after an index has fallen beneath the 50 day moving average, and then rises beneath it, testing it, preferably following a multi-day run. It also helps to have RSI(2) rising above 90.
In the chart above, there are other indicators that are not so bearish, such as the recent MACD cross, and the recent peak in ATR(10). In regards to ATR(10), volatility tends to peak near bottoms, and it appears it might have peak already, near the bottom of the pullback.
Anyway, let’s test this honey hole.
The Rules:
1. 3 previous days all had closes higher than the opens.
2. Today, the close is less than the open.
3. The close is less than the 50 day moving average.
Sell short the SPY at next open and cover the short N days later. No commissions or slippage. There were 53 instances of this setup over all SPY history.
The Results:
Alas, it appears that there is not much honey in this hole, unless we are willing to wait around for over 2 months. Even the percentage of winning trades (axis on right side) is dismal, showing a greater chance of losing than winning.
As the proud borrower of [[SSO]] , based on these tests, I will be looking to cover this position quickly, on any weakness. While it is probable that some weakness can be expected over the next few days, the edge is not near strong enough for me to stay short with any real conviction.
It seems that the bears are early here at the honey hole. The conviction trade remains on the long side, deploying any cash into near-term weakness.
Wood,
You are amazing at coming up with these profiles. Does the fact you have two tests of the MA make a difference?
Thanks Mustard!
Re: two tests of the MA…This test was was a multi-factor test, with only one of the factors being a close beneath the MA50. Therefore, I don’t think that we should overweight the moving average part of the test. As we are including other factors, like consecutive closes above opens, and then a close beneath the open, it is hard to say exactly what the affect is of the MA50.
Hmm I wonder if any good edges appear for SMAs of different periods… however, and correct me if I’m wrong, I think optimizing the SMA period in this case falls under curve-fitting, which we do not want to do (unless the predictive value is confirmed by walk-forward testing???).
Troy, over-optimizing SMAs would be curve-fitting. What we would want to do is to choose a value that has performed steadily over the years. We don’t want the best value and obviously we don’t want the worst value. Instead, we want something in between where values on either side do well also. This gives us room for conditions to change without completely breaking the system.
Hey Wood. Just a thought. If one of the parameters above is simply “below the 50ma”, that would also include a lot of instances where SPY was well below the 50ma. (Im assuming your interested in catching a pullback to the 50ma.) I’d be interested to see the results if you were to change ” below the 50ma” to something like, ” < 5% below the 50ma”, or “below the 50ma,but above the 100ma”
Deuce, I had the same thought has you, and ran some of those tests. I don’t remember the exact results, but my thought for excluding those tests was that it reduced the sample size tremendously. Perhaps if I run a similar test soon I’ll dial it in to be a little more specific.
Anyone trying the Fidelity Select Rotational system? I see the SPX has closed over the 50 DMA which is the trigger to run the FSF system. If anyone is running with this let me know. I ran it and I have screen captured the results below. If anyone can verify my exploration that would also be appreciated. I am showing FSAIX having the top score followed by FSMEX.
http://www.screencast.com/users/geckoman/folders/Jing/media/5d6cb59f-c5a0-431e-ba87-32ad81819467
Johnny, I’ll run mine this evening and see what I get, and report back here in the comments.
Johnny B, I have FPHAX as top ranked via “score” and FSHCX in 2nd place. I ran the exploration really quickly though, and I might have changed some of the parameters over the last few weeks. I’ll take a more detailed look at it tonight.
What are you using for your ROC length? Mine is still set to 90.
I just emailed you the follow up
Wood…please check your yahoo email.