Project Tradebot I
Tuesday 22 March 2016
Tradebots are cool, so I’ve decided to make one. I’ll be documenting my methods and progress in this mini-series.
I’ve written a few articles on tradebots over the past year or so. If you didn’t already know or couldn’t guess from the name, a tradebot is a piece of software (bot) that places trades for you based on a specified set of criteria. There are many different flavours of tradebot, but the aim is pretty straightforward: profit from movements in the markets - in this case, crypto - with the software automatically executing trades for you. A computer can place trades and make calculations faster than any human, can monitor markets 24/7, and doesn’t get tired, drunk or scared and make mistakes. The attractions of this are obvious.
The range of bots is as wide as the range of manual trading strategies. There are extremely sophisticated bots that seek to determine not just the overall trend of the market but to predict short-term movements based on all kinds of criteria, from support and resistance levels to more technical indicators. These can be very profitable but, just like manual trading, there’s a high degree of risk involved. Markets aren’t as predictable as people like to think, even if the movement makes sense with hindsight.
This badass is going to be trading for me
At the other end of the spectrum, there are the relatively safe strategies, like market-making and arbitrage (see the LIQUID and NEXTBOND articles above). Arb bots seeks to exploit the difference in price between different exchanges; because bitcoin and crypto are fairly thinly traded and inefficiencies exist across markets, you can generally expect some discrepancies. When that happens, you can profit by a simultaneous buy on one exchange and a sell on the other. This is something that is best carried out by a bot, which has faster reactions than a human and concentration that doesn’t wander. Market-making bots profit from the spread, the difference between bids and asks. It’s a small difference - perhaps only satoshis in crypto - but if you close thousands of trades, it can be very lucrative.
It’s never that simple, of course, thanks to the principle of GIGO: Garbage In, Garbage Out. A bot is only as intelligent as its programmer, and a bug in your software can prove catastrophic. LIQUID recently lost tens of thousands of dollars when a bot opened a series of leveraged shorts against Ethereum in its testing phase. Ethereum rose sharply, wiping out a large fraction of their reserves. A bot will do exactly what you tell it to, no more, no less. It’s the ultimate Sorcerer's Apprentice.
I really like the idea of creating my own tradebot, so I’m going to give it a whirl. I’m somewhat hampered by my lack of programming skills, and my lack of knowledge of trading strategies - or, rather, lack of confidence in implementing what I do know. I’m not a trader and don’t really want to go down that route (it’s a steep and expensive learning curve, and anecdotally some 80% of traders lose money in the long run).
So I’ve got some choices to make and skills to learn. I’ll be discussing my strategies and progress here over the next few weeks. Wish me luck!
comments powered by Disqus