As a long-time user of ratios, I've followed your development of the BPNYA:CPCE (the Seldon Ratio) with keen interest. The following are unsolicited observations of course, but they're offered with the sincerest respect of a fellow technician. Use these ideas as you wish, up to and including ignoring them altogether.
The Seldon Ratio has great promise as a usable indicator. The problems you're currently experiencing can best be described as raw data distortion. 7+ years of using ratios as a trading tool has taught me that their effectiveness comes from how you smooth the raw data. Fortunately, the tools are now so good that this process is easier and more enjoyable than ever.
The first chart below is the 3-year view of the Seldon Ratio set up the way I've seen you post it. It oscillates beautifully in a tight range, the hallmark of a great ratio indicator. Since you're using BPNYA, I've superimposed the SPX as a reference, and assume you'd trade SPY.
My own work has shown -- as does this raw ratio -- that it's next-to-impossible to devise a stable, one-decision trigger using just the raw data. As you can tell by the various multi-touches on both upper and lower boundaries, raw data creates painful headfakes at both boundaries. At the slow speed of real-life trading, those false entries will ruin the efficacy of the system -- and your portfolio returns!
The best solution I've found in using ratios -- and in 7 years, I've tried every one I could think of -- is to smooth the data with moving averages. Boundary triggers aren't the answer, EMA crossovers are.
The chart below shows the same 3-year period, but with one key difference: the raw data is hidden. In its place are 21- and 50-day EMA's of the data. EMA crossovers trigger the Buy/Sell signals, and these are expressed as Green and Red arrows respectively. The arrows are placed on the SPX graph for ease of interpretation.
The beauty of this approach can be seen immediately: over the 3-year sampling period shown, 21/50 crossovers have produced profitable long trades 100% of the time. 5 long trades, all profitable. The 6th is currently on, and it looks good as well.
Note that this system doesn't work consistently on short trades. Regardless of the ratio system, timing short trades is far more difficult. I'm sure that one can be devised, I just haven't spent as much time over the years crafting one.
To better assess the current market situation, below is a 6-month zoom in of both the raw data version and the EMA version. Currently, the boundary system appears very stretched and about to trigger a Sell. Meanwhile, the EMA system just triggered a Buy last week, and suggests that the current rally has a ways to go.
To compare the two, the raw data Nov Sell was very early, exiting the trade 3% earlier than the EMA exit. Meanwhile, the EMA stayed in the Buy trade another 3 months (the orange arrow on the EMA chart is the first touch; the Sell [red arrow] is triggered by the second touch). Then, as you know, the raw data version offered a very dubious entry point in March, while the EMA version is now in a Buy trade. MA versions of anything will always have lag, but in the case of ratios, it helps their accuracy. I call it the Certainty Put.
To summarize, you're on to a great thing with the Seldon Ratio. However, consider smoothing the data in some way. EMA crossovers do the job nicely, but you may find another tool that works even better. Also, I'm sure you can come up with something to improve the success of the short trade.
Best of luck, and thanks again for letting me share,
PS.. You can leave comments here if you like. I'll receive an e-mail when you post one.
Monday, January 01, 2007