Hi GJ, KookeboyInteresting discussion on the use of MAs as...

Hi GJ, Kookeboy

Interesting discussion on the use of MAs as entry/exit signals. I have been studying them quite a bit in the last few months as I have been developing my FX trading methodology around them. A good opportunity to share some of my thoughts on the subject as it's topical today.

As I see it, the main issues with the use of MAs as primary buy/sell triggers are:

1. They obviously lag price movement in proportion to the length of the MA period used - the shorter the MA, the more whipsaws/false signals you get, whereas the longer the MA, the more of the range you miss out on before the crossover gives you the buy/sell signal (such as GJ's BHP example)

2. Whichever MA period(s) you use, MAs will make you money in trending markets (up or down) because they are primarily trend indicators, but will cost you money in ranging markets, because there is too much noise as price or lower period MAs repeatedly cross up & down over the higher period MAs

The net effect of these two drawbacks from what I have seen is that as long as you are trading in an asset that has a net long term directional bias (whether up like most indexes or down like some stocks) then using a MA as the primary buy/sell trigger will result in a net profit over an appropriate period of time. However the smaller the net directional bias, then the more time spent in ranging movements versus trending moves one way or the other, and if the ratio of ranging periods to trending periods is high enough, the net losses incurred during the ranging periods may outweigh the profits incurred in trending periods.

I can give you an example I studied recently, although this is an FX example, I think it demonstrates what happens very well. The example is the EURJPY FX pair. I studied the daily price movement in this pair over the period August 2008 to July 2010. In this period there is a net directional bias (down), so you would expect profits from trending periods to outweigh net losses in any ranging periods, and they do.

It's a good example because there are two nice trending periods (August 2008 to April 2009, and January 2010 to July 2010), separated by a ranging period (April 2009 to January 2010).

It's also a good example because it shows the comparative effect of different period MAs and how much of each swing or trend can be captured by MA crossover buy/sell signals for that swing or trend.

The daily chart for this period is shown below. Included on the chart are the following:

1. EMAs - 8 (blue), 41 (pink), 100 (yellow), 195 (black)
2. Buy & sell signals (green or red arrows) based on each possible MA crossover - 8/41, 8/100, 8/195, 41/100, 41/195, 100/195



Thus there are 7 possible MA crossovers going up & down, and I put a buy/sell trigger on each crossing of each MA pair (a buy/sell trigger meaning the position is reversed from short to long (buy, and long to short (sell)). If the currency pair establishes a long enough trend, then all lower MAs will have cross over the 195EMA and there will be a net 7 contracts (say) in that trending direction. When the trend turns, the 8/41 crossover occurs first and eventually the 100/195 crossover occurs if the trend goes completely the other way, and so on until eventually there are a net 7 contracts in the other direction. When the market enters a ranging period, the net number of contracts will be less than 7 as some shorts will offset some longs.

What I did for the purpose of this exercise was to start entering positions at the start of the period, as crossovers occurred, and I closed out all positions at the end of the period.

The results were very interesting. I have two spreadsheets which I put together which calculate the result of all the transactions indicated by the buy/sell arrows on the chart.
The 1st spreadsheet shows when the buy/sell transactions occurred, the p/l for each, the total p/l over the period for each MA crossover pair, and the grand total for all crossovers for all pairs over the period.



The results show that the 8/41 crossover pair (the lowest order pair) generates the most profit over the entire period, and the higher order crossover pairs generate increasingly less profits. A quick scan of the spreadsheet shows that the number of transactions decreased from the lowest order pair (8/41) to the highest order pair (100/195), but interestingly the number of transactions for all the EMA8 pairs (i.e. 8/41, 8/100 & 8/195) were about the same, but the total profit decreased strongly from the lowest order to the highest order.

What these results indicated to me was that:

1. provided there is a net directional bias in the selected market & period, a net profit can be generated just using MA crossovers as the primary (only) buy/sell trigger, provided the ratio of any trending periods to any ranging periods is high enough

2.that lower order MA crossovers will generate more profit than higher order MA crossovers, despite any higher number of whipsaws in price that generate more transactions for the lower order MA crossovers compared to the higher order MA crossovers

In the 2nd spreadsheet, I modified the buy/sell strategy so that lower order MA crossovers were not reversed unless higher order crossovers had occurred, as a way of reducing the number of whipsaw transactions within the ranging period. This strategy was effective in achieving that objective. The period from 3/4/200 to 17/1/2010, within the red outline in the spreadsheet shows the ranging period and the number of transactions have been reduced significantly.



The results of the 2nd spreadsheet show about a 25% higher grand total profit than in the 1st spreadsheet, which shows the impact of the whipsaw trades within the ranging period on the profit for the total period. In other words, the ranging whipsaw losses ate about 25% out of the profits achieved from the trending periods.

GJ's BHP example is closest in approximation to the 8/195 crossover pair, simply because the BHP example looks at the price crossover of the long term MA in that case, or in other words a 1/200 crossover. I'd suggest if a similar exercise is done on BHP's price movement in that period as I've done here, a significantly higher profit would be generated by implementing a mechanical buy/sell system based on a lower order MA crossover pair such as the 8/41.

Anyway, hope this was useful. Cheers, Sharks.