Disadvantages of Automated Trading
Automated trading has serious disadvantages. In an earlier post we talked about the advantages of automated trading. But every light has shadow. Here at trade-robots we are not over optimistic using automated trading, we see also disadvantages of this approach. Moreover we try to actively handle the negative aspects and least minimize consequences of those difficulties. In this post we will have an in depth look at the disadvantages of automated trading.
This is not a disadvantage of automated trading in itself (as it applies similarly to normal propietary trading), but often when you talk about people getting into automated trading you realize they lack the required knowledge.
Not necessarily knowledge of the markets, but of the technology. You are at a serious disadvantage if you try automated trading without any idea how to program, how to operate a computer. You are at a disadvantage in automated trading if you try to compete on speed without knowing that it takes to win. Yes, there are tools out there doing drag and drop strategy development. Nice. Great. Inefficient. The graph (they all are graphical) for a more complex trade logic (not the signals, just the order handling with all special cases) likely would be good as a wallpaper on our office. In a font size you could not read. End user tools like NinjaTrader have some seriously limitations (look up the NinjaTrader tag on this blog). Even more so – quite brutally – when one cannot program around them. Data analysis of any larger kind requires in depth knowledge of Excel or even better SQL. Without this knowledge, one is at a serious disadvantage. And this knowledge is often harder to get than market knowledge, taking long hours of reading and theory that non-technological peeople will find boring. Automated trading, backtesting require some knowledge of statistics – that often is lacking and requires additional reading.
This puts a trader that tries to get into automated trading at a disadvantage unless he either is willing to learn, or willing to pay (or comes from an area wih a strong IT backtroung). A combination – learning and paying someone for the more complex things – is likely the best approach to handle this. Coming from IT – for example by being a programmer – the market side is often easier to acquire than the technical side, although getting a profitable strategy is yet another story. Robert Pardo’s “The Development and Optimization of Trading Strategies” is a good starting point.
Electronic trading needs investment. Data feed, software. Even more if you consider trading using trading robots, as automated as possible.
To have a properly optimized and tested strategy flawless data and software that make use of this data are needed. Saving and using poor historical data for optimization will affect real time trading. Getting software that - and the necessary computers that can fully utilize this data is another story. Simply said - this kind of software is hard to find and not cheap and requires even more investments. We tried - when we found all the ridiculous bugsi in NinjaTrader, and we did find nothing we liked. Infrastructure costs a lot of money: database servers to collect results, servers to analyze them. You need a good internet connection and in some areas a battery backup system for power failures. And then there is the trading - you will want a good computer in a data center close to the exchange.
As an old fashioned propietary trader this is easier. A desk, a computer, some screens. A data feed. Finished. Nice. We have more than 20 computers just testing strategies, multiple people working on the same setups. A quite capable file server and a 3000GB database to handle backtest results. We get a special data feed of the complete exchange and have years of data to test – all that costs money. The electricity bill alone is likely higher than the data feed costs of smaller traders. Add to that external servers for actual trading (close to the exchange), backups (tape drive) and it gets costly.
There is no way to handle this. It is offset by the ability to run many more signals and strategies in more markets at the same time than any human ever can hope to come close to. And then, there are still the risk of a proprietary trader. Our automated robots, with money that they earn, make it easy to forget about investments. Without risk there is no reward.
This is another disadvantage of automated trading. Machine don’t need food or sleep, but they need electricity and stable environment. Technical risk takes into consideration that now there is a full techniical layer involved that can be a nice disadvantage for autmoated trading. Problems with electricity, internet connection and other random incidents are not that uncommon.
A human trader is able to “cope” with bad events. Phone off? Use mobile, go to the neighbour. Market closed because of a bomb thread? Open counter-positions at another markets. Yes, the hedge may not be perfect, but in this case “not perfect” is quite ok. I was heavy in TBills futures when Chicago closed for days due to "water in the basement". And being a human we were able to open counter-positions in London. “Shit happens” is a common saying and a serious disadvantage for automated trading. Because a computer cannot cope. A computer follows a strict program. A close order is rejected – what does the strategy do? Maybe just the provider is down. Happens. Even good providers like Rithmic have occasional glitches. A human trader grabs the phone. A computer….?
This can be handled by saying goodbye to the idea that automated trading means “without a human watching”. This puts automated trading at a serious disadvantage. But an administrator with limited trading training (enough to close positions, know where to open counter positions) and an emergency trading desk (available from all good brokers) that can be called in case of… an emergency can handle such situations. Add to that a dead man switch (another computer watching the first) and multiple layers of risk management and the technical risk is low – because at the end there is a human to call, alarm and handle whatever comes up. Obviously that comes at a cost and adds to the financial disadvantage automated trading has.
Automated trading has disadvantages - a fact. Not many, but they are significant and require planning and execution to handle them. All those risk are in addition to any risk that a trader is taking anyway, on a regular basis. This means automated trading is from this side more risky than proprietary trading. They are a disadvantage of automated trading. On the other side, which proprietary trader has exact statistical data about how well his methods work? One has to be careful of curve fitting, but a backtest is a lot more reliable when done by a computer, automated, than by just “winging it” by paper-trading a signal for a short time before risking real money, which is hat proprietary traders have to do.
At the end, automated trading is a different game. It has different requirements, and a trader not willing to learn the necessary skills is at a serious disadvantage. As is someone who wants to fly a plane without flying. A proper, technical, approach to automated trading can handle all of those risks. If you want to fly a plane, learn it. To do proper automated trading, one has to acquire the skills, and it is at parts very technical.
If someone approaches automated trading with the right mindset, all the disadvantages of automated trading can be dealt with.