Discretionary traders have been thinking that they don’t need to know any programming languages and that they don’t need to rely on a computer to do things for them because they are not algo traders.
The line between algo traders and classic discretionary traders is becoming increasingly blurred and less easy to define.
There are, for example, many sophisticated traders who use machine learning to select stocks that are undervalued. However, in the end, they make the final decision on whether to buy or not.
How would you define someone who uses complex artificial intelligence scanners but does the final selection manually? Are they algorithmic traders or discretionary traders?
Years ago, I built a model for a big institution that could synthesize a large quantity of information—all coming from different sources—on oil.
My client had too many numbers on the screen, so to help him, I used principal component analysis to reduce the data into more readable and usable indicators.
This final users of this model were discretionary traders who used to trade out of the counter derivatives on oil. They used my model—and hopefully they were happy with it—but they made the final decision on a completely discretionary basis.
I can keep going with examples but the message should be pretty clear. Even if you are a discretionary trader, an algorithm can help you. The whole trading process won’t be “externalised” to a computer and you’ll be still the center of the whole trading process.
In the retail world, there are many very useful things that discretionary traders, with a bit of effort, can implement themselves. All it takes is a bit of programming and basic statistics knowledge.
Every discretionary trader uses indicators when they are starting out. They usually try out the classic ones that have been available to everyone, such as Bollinger Bands, RSI, Stochastic, and so on.
After few months, they will ask a programmer to code an indicator for them based on their idea.
With this approach comes a big problem: effective communication between a discretionary trader and a programmer is very difficult.
Before programming something, it is necessary to define in a unique way what you want to code. 99% of traders do not know what they want or they are not able to define it in a quantitative manner.
For example, someone can ask for an indicator that identifies the most volatile stocks or the strongest stocks, not thinking about how these two qualitative attributes, strengths and volatilities should be defined in order to be coded.
In this case, the programmer, instead of trying to get the necessary information from his client, will often just do whatever they want. For example, he might define volatility in the way he prefers or maybe in the way that is easiest to code.
Or sometimes the trader wants to code something that cannot be translated into a set of rules. Thus, it cannot be coded or it simply doesn’t make any sense.
You’ll be surprised about how many people want a screener that tells them which stocks are going up at a steepness of “45 degrees”. Of course, the steepness depends on the scaling used on the chart, so just zooming in or out on the same stock can make it visually go up at a different “angle”. It is like making someone could look fatter or thinner just by stretching their photograph.
How can you overcome this problem?
There is only one solution.
If you want your own screener or indicators, you must learn how to code them on your own. By doing that, you’ll see for yourself if something cannot be defined and therefore cannot be coded. By learning how to define things, you will understand much better what you actually wanted to know from an indicator.
For example, most traders define volatility as a standard deviation of returns. Then, when they use a programming language to define it, they see that for their purposes, some easier indicators based on average true range or length of the candles are much more useful.
Testing Strategies :
A discretionary strategy cannot be tested completely because it relies on human intuition and how every person reacts to external geopolitical and economic events is different.
But it is at least useful to test if a given approach to a certain market has some chance of making money or not.
With a few lines of code, it is possible to see if a commodity is more volatile in a given month or if a certain currency reacts to the 200 moving average.
This is very useful information, so discretionary traders should test some of the rules upon which their strategy is based.
As an example, if you used to buy natural gas after a strong downward movement, you can test if, in general, this is profitable or not.
When reading trading books, some strategies are clearly defined with rules and can be easily coded and tested. It is also useful to test whether the author is a potential source of inspiration or is just a scammer like most trading mentors.
If you liked this article, don’t forget my algorithmic trading course.
And ask for my real statement (email@example.com) (I am one of the few trading mentors who trades real money.)
Warning: There is a risk of loss in trading. It is the nature of commodity and securities trading that where there is the opportunity for profit, there is also the risk of loss. Commodity trading involves a certain degree of risk, and may not be suitable for all investors. Derivative transactions, including futures and forex, are complex and carry the risk of substantial losses. Past performance is not necessarily indicative of future results. Please read additional risk matters on our web site www.londontradinginstitute.com
It is important you understand all the risks involved with trading, and you should only trade with risk capital. This communication is intended for the sole use of the intended recipient and is for informational purposes only. It is not intended as investment advice, or an offer or solicitation for the purchase or sale of any financial instrument.