I have no doubt whatsoever that there are "trends" - periods of time when prices go from A to B without too much retracement or noise. I have also no doubt that trends are seen in every timeframe - that such patterns are fractal. It is true that you can not see a trend until it is over - as the market trades, you have no idea whether the price will go from A to B relatively smoothly in a given timeframe or to any other point, smoothly or otherwise. I donât care âwhyâ trends exist. I donât care to argue with those who consider markets to be a Markov chain. I donât care to argue with those who consider the market to be efficient or those who donât. Iâm damned if I know and Iâm pretty sure they donât either. The simple fact is that in the past there have been repeating patterns from which people could have profited if they had been in the markets with the right system at the right time. And some did profit. Simple observation tells us that there are trends (which can only of course be properly defined with hindsight) in all timeframes. No need for maths or statistics to tell us that. The only real question is therefore is how to design and adapt a system to profit from these recurring patters. Which is incredibly easy if the future continues to look anything like the past. As everyone and his uncle, mother, brother and granny now know, trend followers have increasingly hit bumps in the road over the past decade and in particular over the past two years. The patterns have changed â perhaps temporarily, perhaps not. Something has changed at any rate or these long standing CTAs would have ridden smoothly over the bumps without suffering historically high levels of drawdown. Many explanations for the different market conditions have been put forward: the rise of HFT, Government stop/go policies, globalisation increasing correlation between hitherto independent markets, too may trend followers doing the same thing in the same time frames. The list goes on. All that really matters is that a trader who makes his profits by following trends with a method which has provided profit in the past must make a âpredictionâ: will trends continue to form in the future of sufficient duration, sufficient magnitude and sufficient âefficiencyâ ( a trend with a straight enough line from A to B to profit from without getting chopped in an out with a series of losses) to make a given system profitable. Again there will be a lot of argument, doubt and uncertainty. But that is what it boils down to. A prediction about the future; a judgement call which only time will prove right or wrong. A series of tests like those I have been carrying out are helpful (trend efficiency in different timeframes, serial correlation, length of time a given system has historically been able to stick with trend) but no certainty exists. In my view at least. There may be times in the future where simple trend following systems will once again work well, there may not. As always in life, it pays not to put all the eggs in one basket.