I just woke up but... I wonder how he scalps 5 ticks(cents) without any concept of "sustainability"...
1. calc the yield of (Cto/Oto) for all bars in your time series. 2. define a condition like (Hto>H(t-1)) AND (Lto>L(t-1)) and calc the yield for (C(t+1)/O(t+1)). i think this would be the simplest way to prove that trend persists. i think it fails for all time frames you look at. trend followers live from low hit ratios and high payoffs. i have seen no trend following system with a hit ratio of above 50%. thus the money is made in a an asymetric distribution of future returns not in the trendiness of future returns. imo an important difference. [which would make most discussions look like quarrels on wording rather than content]. peace
thank you, mind. given your post--does it not take large amounts of capital to be able to survive "trend following" and expect to earn a return of sufficent size to make the practice worthwhile? AND furthermore, in effect, neutralizing the benefits of such a strategy for the average trader. hank
you know the answers to your questions. IMO it is the starting point of quant trading. good shops do not stop here. maybe they never offer more sophisticated stuff to the public, but they run better things on their own. intraday multimarket futures is a very different story in terms of sharpe ratio IMO. for the average trader it is th easiest way to get the foot in between ... does it take large amounts of capital? our trend follwoing approach cannot be run efficiently with capital below USDm 3. sooner or later we will eperience a 30% draw down. decide for yourself if this is much ... but ... you know all that pretty well, don't you ... peace
Yes, there has be some semantical confusion going on. On one level, "trend" and "persistence of price change" would seem to be the same. However, attempting to follow specific directions of "trend" and using a system that exploits the persistence of price change in general are two different things.
Surely this *has* been the topic under discussion? Also, I disagree with mind's comment on the hit ratio of following a trend. This only applies if your sole signal for trend entry is the prior existence of a trend, broadly defined. In other words, you take a shotgun approach, entering on every possible signal that might indicate a trend starting, and hope that the profits from the good signals bail out the losses from the bad ones. An alternative possibility is qualifying the signals that indicate likely trends, according to the probability of them being right. For example, if you can identify market conditions that give a much higher than normal probability of a trend occuring, then you may be able to increase your hit ratio significantly, even to above 50%.
Given the proliferation of trend-following as a methodology across all spectrums and types of investors, I think that a valid case can be made for inefficiencies and edges being created by this use of this strategy in certain shorter-term time horizons.
By exiting a position at a more favourable price than he entered at. Examples - a market maker working the bid-offer spread in an illiquid security; an arbitrageur purchasing fungible securities at different prices.
What you are describing is unselective trend-following, with low hit rates and therefore high drawdowns which necessitate diversification and thus a reasonable capital base. But trend-following does not have to mean what Dunn & others do - that is simply one, increasingly worn out method with steadily and inevitably deteriorating risk/return characteristics. All "trend-following" means is identifying periods where the market tends to move in one direction with small corrections, and then trading those moves with the appropriate techniques (typically entering and sometimes pyramiding in the direction of the trend, using a trailing stop). It takes a lot more than the poor performance of one particular *subcategory* of trend-following in order to be able to write off the strategy as a whole.