The DecisionPoint Price Momentum Model (PMM) is a mechanical trading system developed by DecisionPoint's Carl Swenlin in the 1990's. The model favors stocks that are moving higher with strong momemtum. The model is still used today to create the DecisionPoint Tracker Reports found on StockCharts.com.
According to Carl, the model works well for any security or market index that has a history of relatively low volatility; mutual funds in particular. It is simple to use, and has a high reliability for being in the market during major up moves and for being out of the market during major declines. It was designed to minimize worry, hand wringing, and sleeplessness.
Note: The PMM has nothing at all to do with the Price Momentum Oscillator (PMO).
The PMM is strictly mechanical and is always set on a BUY signal or a SELL signal.
For a PMM signal change to occur, prices must: (1) reverse at least 10% from the extreme price for the current signal; and (2) pass through the 200-day exponential moving average (200EMA). Important: Both the 10% move and the 200-day EMA crossover must exist at the same time for the signal to change. Once these conditions are met the new signal is "locked in" until conditions for the opposite signal occur.
On our daily Tracker reports we calculate the “stop” for each signal. The stop is simply the point at which the signal will change. It is either the distance from the 200EMA or 10% move, whichever is farthest. It would be rare that a 10% price move and 200EMA crossover would occur simultaneously, so a given signal change will necessarily be dependent on one or the other. It the 10% move is the farthest, the stop will be fixed. If it is dependent on the 200EMA, the stop will change daily as the 200EMA value changes.
If the PMM is on a BUY signal and the price is at least 10% lower than the highest price for that BUY signal, AND the price is below its 200EMA, then the model changes to SELL.
If the PMM is on a SELL signal and the price is at least 10% higher than the lowest price for that SELL signal, AND the price is above its 200EMA, then the model changes to BUY.
Here is how PMM signals look on a chart. Note that beginning from the left there is a series of three BUY and three SELL whipsaw signals that were generated during a basing period. This is what happens when prices move in a narrow trading range. Later however prices moved up following the last BUY signal for a wildly profitable one-year move. At the end of the chart, the 2013 BUY signal remained intact after surviving a sharp -17.3% correction.
The 10% price move and 200EMA crossover criteria were chosen so that the model would only respond to fairly large moves in the market, and thus give primarily long-term signals. When we first developed the model in the 1990s, we tested both criteria separately with data beginning in 1980. The 10% model generated about 25 signals and the 200EMA model generated about 55 signals. In both cases, this was entirely too much whipsaw and, of course, not useable.
Then we had the idea to combine the two screens into one model. This reduced the number of signals to nine (9) and all but one was profitable. When tested over a period from 1920 through the present on a single market, the results were not as good because the market will ultimately present a set of conditions to defeat any mechanical model. However, when we back tested it against a wide range of market sectors, it was quite effective. This is due to multiformity–different price indexes have different shapes–and where the model results for some indexes may be poor, results of the model applied to other indexes may be satisfactory.
One of the strengths of the model is that it doesn't let you miss major moves up or down unless the move is extremely quick. As it so happens, the model did take the S&P 500 out of the market two days prior to the 1987 Crash, but I must allow that it might just as easily have missed the move by a day had actual market conditions been slightly different. Generally, the PMM usually does not miss large market moves that extend over long periods.
Another strength is that the model usually won't stay wrong for very long. While there is occasionally some overshoot, it will normally change directions after a maximum 10% loss.
A major weakness is that the model is still subject to whipsaw when market volatility is high and it is moving in a narrow trading range of about 10% or less. In other words, price must move at least 10% in the direction of the signal before a signal can break even.
Another weakness is that, if the price moves too far too fast, the 200EMA can be left far behind and a price move of a lot more than 10% will be required before the 200EMA screen can be tripped for a reversing signal. You can see what I mean by looking at the difference between the “Signal High % P/L” column and the “Signal Final % P/L” column. The first shows the maximum profit embraced by the signal, and the other shows the amount of profit the signal actually captured.
Performance is best in a trending market–that is when the market moves the same general direction over extended periods of time. For more information, please see our PMM Backtesting Results article.
Every trading day we publish a collection of DecisionPoint “Tracker” Reports for our members. We provide Tracker reports for the Dow stocks, the S&P 500, the Nasdaq 100 as well as several other popular stock groups. These Tracker reports are the primary way for following the PMM signals of popular stocks.
Here is a snapshot of the PMM table from the DecisionPoint Tracker Report for the 30 Dow stocks as of November 18, 2014:
Note: This is simply a snapshot of the PMM portion of the report. The actual report includes more tables and live links to the underlying charts.