How to Exploit Arbitrage Chains Using AI-based HFT Algorithms

When trading arbitrage chains, traders exploit the mispricing of multiple currency pairs; but traders need the right technology to detect these opportunities and quickly execute the trades.


If you have ever taken a course in finance, you learned that in theory there shouldn’t be any arbitrage opportunities. “There is no free lunch,” they say, meaning that it should not be possible to generate a risk-free profit.

That makes sense in theory, where capital markets are perfect. In the real world, however, they are not, and arbitrage opportunities arise every day. The total size of arbitrage profits in digital markets just from December 2017 to February 2018 has been estimated above US$ 1 billion.


Trading arbitrage chains means exploiting the mispricing of different currency pairs

The price of a digital currency is mostly affected by supply and demand. As some exchanges react slower to prove movements than others, currency pairs and spreads are frequently mispriced. Mostly, these opportunities appear in a very short time window and at very low margins, so speed and deal size are essential.

Trading arbitrage chains involves trading several currency pairs and exploiting differences in their relative pricing. In a simplified form, it works like this:

Step 1: Start with some amount of currency 1 on an exchange

Step 2: Sell currency 1 and buy currency 2

Step 3: Sell currency 2 and buy currency 3

Step 4: Sell currency 3 buy currency 1

If applied correctly, after step 4, you should end up with an amount of currency 1 that is higher than the amount which you had in step 1. This strategy could also include more than just three currency pairs.


Let’s look at an example:

Consider the following currency pairs:

LTC/BTC: 0.009085

ETH/BTC: 0.0362

LTC/ETH: 0.235

There is mispricing between the BTC price of ETH and the ETH price of LTC. To take advantage of this arbitrage opportunity, you would execute the following trades:

Step 1: Start with 1 BTC

Step 2: Sell 1 BTC and buy 27 ETH

Step 3: Sell 27 ETH and buy 115 LTC

Step 4: Sell 115 LTC and buy 1.0447 BTC

Result: You have gained an arbitrage profit of 0.0447 BTC.


Arbitrage trading comes with execution and size risks, and transaction costs can eat up profits

The above trade seems simple and straight forward. In reality, executing such a trade requires fast reaction speed and therefore low latency. It’s unlikely that a retail trader could benefit from these trading strategies. However, AI-based high-frequency trading (HFT) algorithms can make it work within just a fraction of a second.

It’s not a completely risk-free strategy, because there is execution risk. The algorithm might be executing steps two and three, but before it can execute step four, the arbitrage opportunity might have disappeared because prices have moved.

That’s why advanced AI-based trading algorithms include multiple layers of defense that ensure execution of all steps at required volumes. Moreover, advanced algorithms also consider transaction costs. Depending on the exchange, every step during the trading process comes with a transaction fee. As the margins are tight, high transaction fees could quickly turn the whole trade negative.

The bottom-line: Chain arbitrage can be a highly profitable trading strategy, given that you know what you are doing, and given that you have the right technology.