|
I find market myths fascinating. There is the outright fantastic like the idea that as dress hemlines on the catwalk rise, so do stock prices. Then there is the more believable suggestion that all Stockbrokers wear boat shoes and drive Porsches.
One so called market truth that I have seen come up relatively frequently is that moving averages act as support and resistance levels. The theory goes that large institutions watch certain moving average periods and that they will make buy and sell transactions based on them.
Below is a recent example of this theory in application on the Dow Jones Industrial Index (DJX). The 50-day moving average is a popular choice with this approach and below we see an example of it acting as ‘resistance’ highlighted in yellow.
Chart 1
click to enlarge
Take a closer look however in chart 2. Ignoring the smaller whipsaws, there are 8 major moves where price action trades to the moving average. Of these 8, only the 2 highlighted can be considered successful signals. That is a 25% hit rate. Expanding the time frame will yield similar results.
Chart 2
click to enlarge
Understanding how something works properly is the cure to believing any myth. As moving averages are based on their underlying data series, it is the moving average that reacts to (and follows/hugs) the data, not the data that reacts to the average.
If you look through any chart on any moving average period you will see examples of bounces off that moving average. Take any random number. Smaller numbers will have more examples of bounces because they have less lag, but will also have more failed signals. Take the number 111 and the symbol AAPL. Both choices are random and were selected without first looking at the stock chart.
Chart 3
click to enlarge
As you can see it worked really well except for when it didn’t. More importantly, how would you know to use 111 without the gift of hindsight? What’s that? 111 is too close to 126 which is half a trading year? OK how about a 427-day average on GOOG?
Chart 4
click to enlarge
I am aware that using number games we could probably find some significance for 427 as well, but realistically how could you possibly use such an approach if you can justify the use of almost every single period? Take the 2002 low on IBM after the dot com crash. This finally ends around the 2889-day average. Chart 5 shows simple moving averages in period increments of 500 and then 2889 to illustrate that figuring out which moving average IBM was going to stop on would be impossible.
Chart 5
click to enlarge
As previously stated, any moving average period will offer seemingly perfect examples of support/resistance with hindsight. The fact that this works on any period proves that these reactions are in fact random and not based on Institutions trading off them.
The conclusion here is that moving averages are still a useful tool, but should be used as a gauge for a where a market is currently at. They are akin to measuring the temperature today and do not directly suggest what will happen tomorrow, next week or next month in much the same way that a thermometer alone cannot forecast the temperature tomorrow.
Happy trading
Jordan Craw
|