Tuesday, March 21, 2006

The Power Measure: What Happens Once a Trend Emerges?

Yesterday's entry on the Trading Psychology Weblog dusted off a proprietary indicator I developed a while ago, which I dubbed the Power Measure. It was my very first effort to measure a variable I call "trendiness": the market's tendency to persist in directional movement. After writing the recent articles that tracked the decreasing trendiness of the stock indices, I decided to modify the Power Measure and utilize it as an operational measure of trending that could be applied to a variety of markets and time frames. The nice thing about the measure is that it creates a normalized measurement, in which perfect upward trending earns a score of +100 and perfect downward trending earns a score of -100. Scores near zero suggest absence of trending: a tendency for price movement in period one to reverse in period two. Note that the Power Measure is a pure measure of price persistence; it does not confuse momentum/price strength with trending.

I went back to March, 2003 (N = 759) and looked at future price movement in SPY as a function of the Power Measure reading. When the measure was 90 or greater (consistent upside trending; N = 85), the next three days in SPY averaged a loss of -.02% (43 up, 42 down)--much weaker than the average three-day gain of .18% for the sample overall (443 up, 316 down).

When the Power Measure was -80 or lower (consistent downside trending; N = 55), the next three days in SPY averaged a gain of .53% (39 up, 16 down)--much stronger than average.

This is the clearest cut evidence I've yet seen of a countertrend bias to the market. Waiting for a distinct trend to emerge and then fading it has proven far more successful as a trading strategy than trying to ride trends. It will be interesting to see if this holds for shorter time frames as well. If my future investigations prove equally promising, I'll bring the Power Measure back to the Weblog.

6 comments:

John Wheatcroft said...

I have noticed in my own research that the best time to make a trade is when the slow stochastic oscillator using 5 periods moves below 50. I.E. long before it gets below 30 and starts back up again. The way the market has been moving is a little up - a little down - and mostly sideways. Consequently when the oscillator moves under 50 the move down is generally finished and it will slide a little sideways- down. Then it will throw off an NR7 signal or a candle hammer or something of that nature and a buy on that day will generally show profit for a couple of days into the future. Of course once it hits 50 ascending it is going sideways again and would be time to sell. Usually there is a clear topping signal.

Just thought I'd throw that into the mix. It is something I've been watching for awhile now and appears to be contrary to what we have been taught over the years.

When you developed the power measure did you share the method? If not - OK but if so - I missed it. I'm always looking for a new idea.

Brett Steenbarger, Ph.D. said...

Hi,

Thanks for your note, John. I do think technical indicators can help identify trends and places to fade them. My main problem with most technical measures is that they confuse two things: price change/momentum and trending. My preference is to treat those separately in analyses.

A few of my measures, such as Demand/Supply and the Power Measure, remain proprietary; they may, at some point, go into a book that I'm contemplating, along with other research. I appreciate the interest--

Brett

John Wheatcroft said...

Because of that I use three separate indicators (momentum, trend, and volatility) and insist that they all meet my parameters before I take the trade. The methodology for my three indicators will not be found in any book by the way - they too are proprietary. Many years of work, many more to go.

But I am always looking for the unified theory. Appreciate the proprietary nature of your work.

Brett Steenbarger, Ph.D. said...

Very nice observation, John. Your big three variables of momentum, trend, and volatility are right on the money. If I had to add a fourth, it would be sentiment.

Brett

Greg Lepiaf said...

Hi ! Brett and John

Many interesting things from both of you these days …

To your big four
Momentum/Trend/Volatily/Sentiment
and in order to trend to a "unified theory"
wouldn't you add
"weighted on various time scales".

Any comment on the hierarchy and weightings of time scales ?

Best Regards

Greg

Brett Steenbarger, Ph.D. said...

Hi Greg,

The idea of weighting various time scales is excellent. I'll see if I can pull together a little example of that in the very near future. I'm not sure a "unified theory" is needed: you can examine different historical patterns for the various measures (trend, volatility, etc) and treat each pattern as an "expert". The different patterns make up your "committee of experts", and you can use the committee's vote to decide whether you trade and how much of your capital to allot to the trade.

Brett