The Trading System That Maximizes Our Edge..
At the current state, cryptocurrency markets provide more than enough opportunities to make a profit (we’ve shown 11.3% ROI last week, without a single losing day; follow our weekly reports on our Facebook page). But to achieve maximum results, one must have a system to consistently track, quantify and exploit market inefficiencies. So let’s have a look at our system:
Studies show (source) that the cost of 10 milliseconds of communication delay is about twice that of an algorithm configured to run on only 1 millisecond (1/1000th of a second or 1/300th of a blink of an eye!) of latency. In turn, 100 milliseconds of latency result in threefold latency cost as compared to that of an algo using 1-millisecond execution latency. That’s why we have “gateways” as close as possible to every exchange that we work with. These are the servers with modified Linux kernels optimized for real-time and low latency performance. They act as feed handlers, processing all data from the exchange and as execution platforms, executing orders coming from the main trading algorithm. In most cases, we’re able to achieve sub-millisecond latencies.
If you trade on one of the many online algo-trading platforms, market data may be delayed by as much as 15 seconds(15000 milliseconds!). If you trade via exchanges’ API directly from your home or office, your latencies are probably within 100–300ms range.
Ticker plant is the system that receives all the data from all exchanges, stores it in the historical and real-time database and is very efficient at querying that data so that the strategy system can spot market anomalies as well as historical patterns with minimal delays.
Our strategy system consists of:
-Alpha Model that generates trading signals based on theoretical valuations or statistical indicators. We use statistical arbitrage (or “stat-arb”) strategies which are, essentially, value-motivated strategies that capitalize on assets’ mispricings that may appear for short periods of time due to market inefficiencies, and market-making strategies that have no material market insight and aim to profit from providing liquidity. Basically, we only use market-neutral strategies that deliver absolute returns not affected by market direction. If you want to see examples of our strategies and how they make money, please read our last article “How do HFT firms and quant traders consistently generate absolute returns that are not affected by crazy swings of cryptocurrencies?”
-Risk Model tries to minimize the total risk exposure. It assesses all trading signals and decides what share of total available capital may be used to open that trade. Even though an opportunity exists, it doesn’t mean we should take advantage of it or invest too much capital in it!
-Transaction Cost Model takes into account commissions and fees, tries to predict slippage (the change in price between the time an order is sent to an exchange and the time it is actually executed) and market impact (the difference between quoted price and actual average executed price, since large orders consume liquidity from multiple levels of an order book and, as a result, shift market price);
-and the Portfolio Construction Model which acts like an arbitrator, hearing the arguments of the optimist (Alpha Model), the pessimist (Risk Model), and the cost‐conscious accountant (Transaction Cost Model). Too much emphasis on the opportunity can lead to ruin by ignoring risk. Too much emphasis on the risk can lead to underperformance by ignoring the opportunity. Too much emphasis on transaction costs can lead to paralysis because this will tend to cause the trader to hold positions indefinitely instead of taking on the cost of refreshing the portfolio.
As a group of “quants” with academic background in Numerical Methods, Computational Mathematics, Game Theory and hands-on experience in High Frequency Trading and Machine Learning, our interest was in exploring opportunities in cryptocurrency markets, with the goal of exploiting various market inefficiencies to generate steady absolute returns (not correlated with market movements) with low volatility, or simply put, steady profit without major drawdowns. For more information please visit http://www.TensorBox.com or read about our Initial Token Offering and the Early Bird Promotion here: https://medium.com/tensorbox/tensorbox-initial-token-offering-details-oct-2nd-2017-ce391140d86e