The Magic Behind Algorithmic Trading Bots

This is part of Kollider’s “Long Story Short” series, where we share short and sweet thoughts and tips with our community. It is…

The Magic Behind Algorithmic Trading Bots

This is part of Kollider’s “Long Story Short” series, where we share short and sweet thoughts and tips with our community. It is educational and is not financial advice.

Here’s a trick question: who would win at trading, a human or a robot?

To answer that, you’ll need to know what a trading bot actually does. We previously covered the basics of market making and spread trading. In those articles, we looked at strategies and the pricing philosophy that goes behind them. But who or what actually executes these ideas?

Today’s “Long Story Short” will help you understand trading bots and how strategies plug into them. Because after all, Kollider was built for all traders, human or bot alike.

There’s an air of mystery surrounding automated trading bots, just as there is about the other aspects of trading we’ve discussed. On the surface, they might appear to be impossibly complex creatures designed to be understood only by mathematicians and software developers. Yet our team has experience working at some of the “top” algorithmic trading firms out there, and we can tell you now that bots are much simpler than they seem.

Automated trading is easiest to understand if you imagine what a human trader would do. Let’s say our human is looking to trade a very simple (and possibly flawed or incomplete) bitcoin momentum strategy. They believe that the markets will probably continue in the direction they have been following since last hour. But how does one trade that basic strategy?

They would likely follow these steps for the simplified strategy:

  1. Check relevant information. The trader will examine past and present market and trade prices.
  2. Guess the future direction. From the current and previous prices, they know the last hour’s direction. In our scenario, the trader believes this direction will continue.
  3. Look at their existing positions. Let’s say that last-hour direction up. The trader is looking to buy, but now they assess their existing positions to see if buying makes sense. If they have already bought too much and are very long, they may sit this one out. Note that the same analysis applies if the last-hour direction is down.
  4. Trade. The trader sends an order to the market to increase, reduce, or reverse their positions.
  5. Repeat the process indefinitely. The trader will repeat steps 1 through 5 until they decide to stop.

At its core, a bot is not that different from a human. It will likely follow the same steps above. Some aspects that may sound different, including how they check prices (perhaps via a websocket versus visually), but most aspects are highly similar.

That’s the essence of it. Some major optimisations could still use the same framework. For example, if the same trader decided to apply a different “strategy” idea (say they now believe that the market will always reverse its direction if it has moved at least 20% in less than a week), they would mostly change what they do during the “guess the future direction” step. If they instead employed a spread strategy, they would also look at a second set of market prices and send orders out to a second exchange. A bot would make the same changes.

There can be pros and cons to trading one way versus another. Even though bots can be less emotional because they are programmed to act the same given the same information, that doesn’t guarantee they will be more accurate. They may not have to shut down and could trade all day and night, but does that mean they will be profitable during those extra hours? In fact, they can and do lose money. Are they faster? Yes, but it also means they can make more mistakes faster.

Can trading bots be completely hands-off? Not exactly. They must be constantly adjusted in order to keep up with market conditions. Then again, it’s no different from a human trader who has to adapt to new price dynamics or how macroeconomic, cryptocurrency, or corporate information is disseminated and consumed.

We like both humans and bots at Kollider. No matter which you choose, we believe you should be as educated as possible. And some people choose both, either separately or by doing machine-assisted trading.

Is there magic behind trading bots? Yes, definitely. When designed and maintained properly, they can be quite effective, but they are not a panacea. So a human trader could actually have a better idea and a much better execution than an automated trader.

So back to the original question: who would win at trading, a human or a bot? The answer: whichever one makes more profit.

Know anyone else interested in learning about bitcoin, Lightning, and trading? Follow us on Twitter and help us share the word.