Arbitrage – in finance and economics – is a transaction that takes advantage of the difference in prices of the same asset on two different markets, or on the same market but in two different forms.
When doing arbitrage (i.e. executing an arbitrage deal), an investor (arbitrageur) simultaneously buys and sells the same asset at two different prices, making a profit that results from the price difference. Due to the purchase and sale taking place at the same time, the result of the arbitration deal is known in advance. Arbitration transactions are thus risk-free.

This basic definition of arbitrage shows the general concept. But, as we all know, details are even more important…
To make profits on the cryptocurrency market, we have created an advanced, automatic system which searches for, analyses, executes and verifies deals in real time, earning the largest profit available at exchanges at a given moment.
We have also developed our original, advanced algorithms for searching for profit, which allow to generate revenues in places that are not taken into account by other systems.
Our system – in real time – downloads current exchange rates (orderbooks) of several hundred currency pairs from a dozen or so exchanges. This is done several times a second, thanks to which the data is up-to-date and allows deals to be struck very quickly.
The exchanges that are currently enabled on a production system basis: Bitbay, Coinfloor, Bitfinex, Binance, Kraken, Bittrex, Poloniex, Okex, Hitbtc.
Our system allows enabling of any additional exchanges at all, as well as any currencies (both FIAT and crypto).

 How does the system work?

The downloaded orderbooks (currency pairs) are analysed in real time by several independent algorithms. Every algorithm provides a different method for the execution of deals – i.e. a different form in which profit can be generated.
The system calculates possible profits several times a second and takes decisions as to which transactions will be conducted. All deals in the system – which works around the clock – are executed automatically (in their entirety /100%/).

MODE 1 (1st algorithm for searching for profit)

The simplest form of arbitrage: the system matches – 1:1 – offers for the same currency pair from many exchanges, checking the differences between the exchanges – “where is it possible to buy at the lowest possible price and sell at the highest possible price?”.

Examples of arbitrage opportunities.


That means that if we had funds (USD on BitBay, and XRP on Bitfinex), the system would immediately (in a fully automatic way) do a deal, making a profit of approx. 2%.
Similarly for the pair ETH – USD; here the profit would amount to 1.87%.
Our system in this mode can generate as many as several dozen deals per minute.

For the pair XMR – USD things looks even more attractive; in the above-presented example, the profit amounts to over 5%:

MODE 2 (2nd algorithm for searching for profit)

It is a more complicated algorithm, which searches for profit (and deal execution) in several stages. This allows our system to find profits that are overlooked by most of the arbitrage systems that exist on the market, which usually operate in Mode 1. Profits are often large – up to 10–20% per one deal.
Thus, it is no longer about “1:1” operations (i.e. the simple “buy and sell”), but about cycles of deals which consist of several stages.
It is our original algorithm; as it is a significant component of our system, it has not been described here in detail.

Below several deals are illustrated (details have been hidden).


The last column is the profit, which means that the first cycle of four deals was completed with a profit of 6.94%, the second one – 9.48%.
Both cycles have been completed in the same minute, one by one. The average cycle completion time is 10–15 seconds, which means that the system can generate several such transactions per minute.
After a deal has been executed, the system verifies statuses on the exchanges, to be sure that all transactions have been conducted as assumed, it sums up the profit earned, it signals any insufficient balances (e.g. as a result of a given crypto having been sold on one of the exchanges), and then it starts to execute another deal.
These are just two modes of execution of deals. The possibility of the system’s expansion to include more modes (algorithms) for searching for profit, also on markets wider than just the crypto exchanges, means that there is considerable potential for our system’s further development. We have made a reconnaissance in this regard and we have not found any solutions that would be even close to fulfilling the function currently fulfilled by our system.
One of the optional directions in which our system could develop is to make it available to external users (for a subscription fee and/or some percentage of the profit earned). (Along with the system, we ensure the entire know-how that is connected with it).

The Arbiteo system has been continuously developed, tested and improved for over two years now. We have executed hundreds of thousands of deals whose volume totalled several hundred million PLN, and with the initial capital of approx. 200 thousand PLN, we have earned profits in excess of 2 million PLN.

The system is fully configurable – it is very simple to choose which exchanges (as well as individual markets) are to be enabled (in the example below – the green ones), and which are to be disabled (the red ones).


Some of the most important variables to be defined in the system are:

  • minimum profit on a transaction (below which our system does not execute deals);
  • minimum value of a single transaction;
  • maximum value of a single transaction;
  • trading fee applied on exchanges (important when it comes to profit calculation).

The system has a built-in mechanism for e-mail notifications (slack). It can inform the indicated persons e.g. about every completed transaction and generated profit, low level of funds on a given exchange, or short-term problems with the exchanges (e.g. lack of communication).

Most processes in our system are fully automated and do not require human intervention. Such operations, however, as bank transfers or cryptocurrency transfers need to be defined manually. To make this component user-friendly, too, a mechanism has been prepared that allows to carry out transfers and fully control them. In the course of further development of the system it will be possible to implement full automation of transfers between exchanges.

The system has a built-in self-control mechanism which, once every 24 hours (the frequency can be changed) sends an e-mail with information about the number of completed transactions, the profit generated etc. This ensures full control over the funds in the system and the current profit.

The system is protected – both against external attacks (infrastructure with full security measures, access via a VPN only) and against erroneous execution of deals resulting from a lack of communication with exchanges (or errors) or internal errors.
If the system detects an irregularity on any of the exchanges, such exchange (or an individual market, that is a currency pair) gets disabled, and the system continues its operations on the subsequent exchanges.

Broad analysis mechanism (BAM)

At present, the system operates on 12 exchanges and several dozen of currencies/cryptocurrencies, however, it is prepared to support more markets. To examine which markets have the greatest potential, we have prepared an additional mechanism for the so-called “broad analysis”. It analyses profits for a much (many times) greater number of exchanges and currency pairs than the trading system, identifying those with the greatest potential.
An example: using the Huobi exchange (currently not enabled in the system) it would be possible to execute deals with a profit exceeding 6%.

Currently full analysis of the arbitrage market is not being performed, however, according to a 24-hour-long monitoring exercise, the potential volume of transactions possible to conduct with a profit exceeding 1% within 24 hours exceeded $ 4 million.
This is data only for the exchanges that currently remain implemented in our system. If we take account of data from Asian exchanges and additional cryptocurrencies, this would be over $10 million per 24 hours.
One should also note that the current (1st quarter of 2019) movements of cryptocurrency prices are not significant, and thus the volumes on exchanges are relatively low. Where price fluctuations are big, volumes tend to be much higher (by many times).