Ask Colin

The monthly 14 period RSI shows a strong triple divergence in August 2001. Is RSI the appropriate tool to get out before the market tops out?

One of the problems with momentum oscillators is that they should only be used to enter uptrends, not to exit them. This is because they tend to give many sell signals that are premature. In an uptrend a momentum oscillator goes overbought quite quickly and can stay in that position for long periods in a sustained uptrend. However, if you think about how a momentum oscillator is constructed, it is obvious that in a consistent trend it will show divergences. For it not to do so, the trend must actually keep accelerating.

In late August 2001, which is when you are talking about, the uptrend had been showing signs of loss of momentum for a long time. I personally do not need a momentum oscillator to see that on the price chart. The question is one of timing. One of the divergences will be the last one. But how does that help if it was preceded by premature signals? It is like the boy who cried wolf. Likewise, a loss of momentum observed on the naked price chart is also of no use for timing. We have to wait for the trend to fail, for a signal to be given.

This is why the thrust of all my writing for over a year has been that this market is high risk. Yes it is still technically in an uptrend. Yes we should still be involved. I have said that. However, I have also said that our exposure to the trend should be reduced and that we should focus our positions on defensive situations. All this and also the advice to be ruthless on quitting stocks that fall out of uptrends have I given repeatedly.

This is the thrust of my approach. Trying to predict markets is futile. However, the sensible trader and investor cuts his or her cloth to suit the situation in the ways I have described.