Why create my own automated trading system? What are the pros and cons?

Fast - Trading is a time consuming activity. It requires a lot of hard work and dedication if done manually. From analyzing market trend to screening for the stocks that fit the trading criteria, a lot of effort is required. If I am to trade without an automated system, I would probably need to quit my job to do so. In fact, with so many steps to perform, I would probably won’t be able to complete the daily analysis in a day! With an automatic trading system, the system will perform the tasks for you automatically and at a shorter time.

Free up time - With the system trading for me automatically, It frees up time so that I can spend it on more important stuff in life like family, holiday or uh… perhaps create an even more automated trading system to free up more time. 

Reduce cost -  A complex trading strategy would be rather impossible to be manually managed by a single person. You will likely need to engage several others to help in the different tasks thus increasing the cost (salary, workspace etc). With an automated trading system, you do not need an extra helping hand. You can make the system do everything for you (yes, it is possible!).

Emotionless - Emotion can cause us to lose our logical decision making which is crucial to be successful in trading. It is especially critical if we are in the situation where our huge position is crashing or it is rising sharply. Emotion can cloud our judgement (even if we have a sound trading strategy) stirring up fear or greed which could result in us holding losing position too long or selling winning position too early. However, computer has no emotion. It has neither fear nor greed. It does not become euphoric when its position is skyrocketing or devastated when losing half the capital (though I would!) It just does whatever it is programmed to do.

Consistent - This is quite related to our emotional state. No matter how hard we try, we humans just can’t be consistent all the time. Our performance can be affected by many factors - our mood of the day (that guy just took off with my cab!), the situation at work (I just got a bad bonus this year), the amount of sleep we have or the coffee is just not right today. This could cause us to sway away from our trading strategy which would translate into bad trading decisions. However, an automated trading system does not waiver. It just perform the same task over and over again, day in day out. I can count on it to execute my strategy with great precision. 

To trade around the clock and around the globe - Yes, your computer can be awake 24/7. It can trade different markets around the globe at different time zone while I am sleeping or at work, giving me more exposure and diversification.

Can be back-tested - Whatever the strategy, it allows me to validate it before putting in my hard earned money. Back-testing is crucial to make sure the strategy works on historical market data and to determine the risk and reward of the strategy. Take note that successful back testing does not necessary mean it will be successful in real trading. There are many factors to consider during back testing to remove any form of bias as detailed below as disadvantages.


Automated trading system is not without its flaws. Here are some of the disadvantages.


Crashes sometimes - Like all software, a complex automated trading system will not be bug free. Sometimes, it can crash.The reason could be due to bad input data, hardware issues (eg out of diskspace) or coding errors. No matter how much effort is spent in testing the system before releasing it, it will still not be bug free. There is no way of preventing it. You just have to accept this fact and try to get as many bugs fixed as possible.

Still requires some sort of monitoring - As much as I want to, however, my automated trading system still need some sort of monitoring. I think this applies to all kinds of automated trading systems. Nobody should attempt to let his system trade automatically without having any sort of monitoring at all. Doing so would be suicidal. For example, one simple bug in the calculation of position size could end up with it buying a position with all your capital. As mentioned above, no matter how much time you spent in validating the system, it cannot be bug free. This becomes more apparent as your system gets more complex.

When I say monitoring, I don’t mean sitting at your desk and staring at the laptop monitor watching it trade all the time, as that defeats the fundamental purpose of automation. What I mean is there must be some ways you can communicate with your system no matter where you are. I am talking about things like an alert system or a manual override kind of stuff. You could be enjoying your holiday in Bali and the system should still be able to send transactions report to you. And if it detects any issue, it should allow you to manually resolve it remotely (eg sell off a wrong buy) or at least halt the system to prevent further damage. This is a kind of risk management too.

Can have back-testing bias - Although back-testing allows you to validate your system, it can turn against you if not executed properly. One example is over-fitting (or curve fitting) where you tried too hard to make your system profitable, unaware that you introduced too many parameters into your trading criteria. Over-fitting can mislead you into thinking that your system is working wonderfully while validating against historical data, only to end up scratching your head when it does not perform that well in live trading. The other kind of bias I encountered is forward looking bias where the system has an unintentional ability to ‘see’ the future of price movements. This could easily happen since you are validating against historical data and a subtle error in your price referencing code when running though the data would turn your system into an oracle, resulting in a system that works exceptionally well, apparently. So whenever you find that your system seems to work overly well during back testing, it is wise not be overly excited too early but rather, be a bit more skeptical in case you unintentionally introduced curve fitting or forward looking bias.