The early part of this week showed the difficulties in following a trend following strategy. By the very nature of the system, you would never exit a position at the high or low of the move.
When a retracement occurs, you do not know whether this is a short-term pullback, or the start of a trend reversal. These periods can be difficult to deal with, as you can see significant profits vanish.
I have read an interesting study of the merits of 'pyramiding' both in counter-trend and trend-following systems, which I came across in an old copy of Traders' magazine. To me, the pyramiding described in the article actually refers to scaling in and scaling out of positions. To understand the article, the reader was given the definition of entry and exit efficiency.
The entry efficiency is measured on a rate between 0% and 100%, indicating how close the entry was to the low (or for short trades to the high) of the trade in comparison with the total price range of the trade. The exit efficiency measures how close the exit was to the high (or for short trades the low) of the trade in comparison with the total price range of the trade.
In a counter-trend system, the exit efficiency is generally higher than the entry efficiency. This basically means that the entry of the trade was not optimal, but that also the trade improved before the exit. This therefore demonstrates that averaging down the position (i.e. scaling in) can be advantageous.
On the other hand, a trend-following system generally has the opposite characteristics, in that the entry eficiency is higher than the exit efficiency. Because of this, a reduction of position size (i.e. scaling out) as the trade progresses can be advantageous.
All other things being equal, the above is the background on why the percentage of profitable trades is lower when using a trend-following system when compared to a counter-trend strategy. The study went on to take both basic systems, and then apply the scaling in/out of the positions, with significant effects on both overall profit achieved, and a lower maximum drawdown suffered.
Traders familiar with the Turtles trading methodology will know that scaling out of a trend-following position is the polar opposite of the aggresive pyramiding of positions which gave them such big winning trades, but also contributed to the overall volatility of their results.
In my own (trend-following) experience, I can see how the scaling out of positions would work in that, by taking partial profits when an initial target is met, a higher overall percentage success rate would be achieved, as you would be 'locking in' an initial profit and also reducing the risk on the remainder of the trade. In theory, this could minimise the profits earned on a substantial move. However the test showed that the overall profit achieved was higher, with a reduced drawdown.
The point at which to take partial profits can either be determined by using a specific percentage distance from the initial entry, or by reference to a multiple of Average True Range.
To me, this is better than completely exiting a position based on the assumption that a long trade is now 'overbought', or that a trader thinks that a trend is coming to the end. Should you wish, you can always re-instate the balance of the position if the dominant trend re-asserts itself, when a new high/low is attained.
If you have several profitable positions open, then the risk of a significant retracement should be looked at in respect to the total equity in the portfolio, NOT to each individual position. This is where scaling out of your positions really scores, in that you won't be losing the bulk of your hard-earned gains.