The Patient Chart Pattern Trader
Although HFT and algo trading dominate market activity nowadays to the tune of about 80% of transaction volume, there are still a number of old school chart pattern traders around. This is evident from social media messages where these traders post charts with patterns, such as head and shoulders, triangles, trendlines, double tops and bottoms, just to name a few. Although some of those chart traders aim to only teach their “art” to new traders, some are obviously patient enough to trade with it.
I will show below how being “patient” affects performance in a quantitative sense. But before doing this it may be useful to demonstrate in a visual way why this choice of word was made.
Below is an example of a failed head and shoulders pattern found in my book Fooled By Technical Analysis. This pattern was formed in Russell 2000 index in the first half of 2012 and was confirmed according to rules of chart analysis since prices broke well below the pattern neckline.
However, the price target of the above pattern was never reached; prices reversed and made new highs by September of the same year. The point here is not that some pattern failed. Actually, the longer-term mathematical expectation from trading this and similar patterns must be exactly zero before commission cost. In fact, the burden of proof of an opposite claim is with those who believe these random patterns can have economic value. The main point is that it took about four months for this pattern to form. In this time period a trading algo could have generated a sufficient sample of trades for testing its statistical significance. Yet, this is a sample of one for this pattern in the same period. A chart trader of this particular pattern would have to wait several years before obtaining a sufficient sample based on an average duration of say three months. This is one of several reasons of absence of studies of performance when these patterns were conceived as having economic value in the mid twentieth century since it was not feasible to perform simulations and backtest performance back then.
Now that we have defined what “patient” means in this context, we can analyze the impact of “patience” on performance. Some chartists claim to have 45% win rate and payoff ratio of 2 meaning that they win two units 45% of the time and lose one unit 55% of the time. Note that the odds against this performance record are given by the following equation
R = (1-w)/w (1)
where R is the minimum payoff ratio for breakeven performance and w is the win fraction. If the win fraction is 0.45, then R is 1.22. Odds against are 1.22 to 1. Therefore, a payoff ratio of 2 guarantees positive performance as seen by the Monte Carlo simulation results below:
The first problem of course is that the above simulation requires stationary win fraction. If the claim of 45% success rate is false, then the simulation results are also false. Actually, the burden of proof is with those who claim high win rate to prove that it is better than 33% when the payoff ratio is 2, as this is the break even success rate according to the following formula (see this article for more details and derivation)
w = 1/(1+R) (2)
where w is the win fraction and R is the payoff ratio.
The second and more important problem is that although profitability is almost guaranteed by 45% win rate and payoff of 2, performance depends on actual trade number and in the case of pattern formations that take weeks or months to complete, this number is small.
In order to see that, let us consider a chart pattern trader and an algo trader with the following parameters.
Chart pattern trader: win rate = 45%, reward = 2, risk = 1
This trader has low win rate but the claim is that he makes twice what he loses, on the average.
Algo trader: win rate = 67.5%, reward 1, risk =1
The algo trader has payoff ratio of 1 but at higher win rate.
The average trade for the chart pattern trader, avgTc, is calculated as follows:
AvgTc = 0.45 x 2–0.55 x 1 = 0.35
The average trade for the algo trader, avgTs is given by:
AvgTs = 0.675 x 1–0.325 x 1 = 0.35
Note that the average trade is equal to the mean of the population, also referred to as expectation or expectancy, only for sufficient samples. The issue of sufficient samples for chart traders was already discussed above.
Therefore, based on the above calculations, the average trade of the chart pattern trader is the same as that of the algo trader. Since final equity is the product of the average trade and number of trades, this means that a chart pattern trader must execute as many trades to achieve the same equity value as an algo trader.
Now, this could be a major problem; the patient chart pattern trader must find many markets where these chart formations occur and that may not be possible since they are not that common. The other choice is increasing risk. This obviously increases the probability of a large drawdown and ruin.
Chart pattern trading is a style that is more suitable for recreational trading rather than professional. This is one reason it was never considered seriously by the majority of hedge funds. In addition to requiring patience, slow chart pattern formations offer enough time for detection and competition is high at diminishing returns.
Note that in the above example, the algo trader has the same average trade as that of the chart pattern trader if win rate is 67.5% for payoff equal to 1. However, some algo traders maintain even higher win rate. Click here for specific examples. Needless to say that there are also unsuccessful algo traders. Those who do not understand the important of high win rate believe they can exploit the mathematical inverse relationship in equation (2) to their own benefit and can be wrong more often than they are right while maintaining a high payoff ratio. However, as shown above with simple math, in the process of doing so, they essentially match chart pattern trader performance.
The conclusion is that chart trading was a style for patient traders during times when everything was slow, from data collection, to chart drawing, to analysis and to executive trades. Nowadays the word is faster by several orders of magnitude. Good chart traders could obviously survive the new dynamics but the expectation should be low given dominance of algos. At the same time, learning that old style of trading is more interesting in the context of studying the reasons it is no longer applicable to the markets.
More information and specific examples can be found in this article.
If you have any questions or comments, happy to connect on Twitter:@mikeharrisNY
This article was originally published in Price Action Lab Blog
About the author: Michael Harris is a trader and best selling author. He is also the developer of the first commercial software for identifying parameter-less patterns in price action 17 years ago. In the last seven years he has worked on the development of DLPAL, a software program that can be used to identify short-term anomalies in market data for use with fixed and machine learning models. Click here for more.