Before I was a full time trader, I was a programmer for
12 13 14 years.
I am a much better trader than I was a programmer, but it payed well, and no-one cared how good the code was as long as the thing worked.
An extension of this was a period where I programmed other people Black Box Trading Systems, Trading Robots or Expert Advisers (EA’s. Whatever you want to call them.
In essence, automated scripts that ran on your charts, taking, exiting and managing trades for you while making copious amounts of money for the trader while they slept.
Yeah, not so much.
Here is the truth in 10 points of what really happened when developing other peoples systems into a rules based, no discretion system.
These opinions were formed based on over 50 different systems programmed over three years:
10 things I learnt programming black box systems
- Most black box systems had enormous thought put into where they entered. They almost all relied on other indicators, and involved the crossing of one thing over another thing with something else used for timing.
- Almost all hardly had a thought put into money management beyond the 2% rule or a variation of it (risk only 2% of your account each trade). The 2% rule is only the beginning of money management, it is initial risk assessment to determing an appropriate trade size, but what happens when the trade is on?
- Exits were almost always an afterthought, and quite often were their entry signal reversed. They also almost always were trying to catch an entire move, and the idea of exiting on weakening of strength was a horrific concept.
- If a black box system didn't make money when completed, it was usually blamed on an error in my programming (in fairness, at times a fair assumption). What was forgotten was computers don’t have bias, they don’t make exceptions, they don't subconsciousness skew their results to make themselves feel good. It is just black and white, it works or it doesn’t. Usually it didn’t.
- Not one of the black box systems I programmed took into account the time of day. None.
- Almost all the programming instructions came with “take this trade if in direction of the trend”, but rarely was it determined exactly what (or which) trend they were talking about.
- Most idea creators wanted a black box system because they “didn't have time” to trade themselves. These systems aren't a set and forget entity, they are a set and adjust, monitor, deactivate, reactivate, adjust, assess, monitor kind of thing. If you don’t have time to trade, you probably don’t have time to keep tabs on the mode the market is in.
- Rarely, was statistics brought into a black box idea, something that computers excel in, that seemed strange.
- I never enjoyed the process of creating one. It usually went along the lines of creation, accusations of bad programming, realisation of bad trading methodology, addition of multiple “filters”, rinse and repeat.
- Not once did a black box system make money over the long run (remember I have the source code, I would know).
No doubt there are black boxes out there that make some money, the algorithmic traders as they are called in the media.
But they are created by quants, mathematic PHD’s, market veterans and just plain geniuses and are funded by the big dogs, who can throw a spare million on something to make it work in most markets conditions.
They involve complex algorithms, placing computers as close to the exchange as possible, and are turned on and off at the right times.
If you have traded a methodology for years, which has clearly defined rules that you stick to, then translating this to a black box might work (we almost always under-estimate the influence of sub-conscious decisions and biases).
If you are starting out in the markets, don’t waste your money, or the programmers time. Learn the markets, be a trader, then see if you can automate components of it.
Oh and forget scalping with a robot as a private trader. You will never be faster than the big money machines.