Types of cryptocurrency trading algorithms and why you may need one?

For more than a year, the cryptocurrency market is in bear territory. Many are wondering when is the next bull run? Some believe the catalyst for the next price rally is driven by demand from institutions. I am bullish about bitcoin and a few alt coins but also very cautious and fear for members of the public who will FOMO(fear of missing out) in and get their behinds handed back to them just like last time. This is because in my opinion, institutions buying into crypto is great but also implies a more efficient market with many more sophisticated participants.

Make no mistake institutions are not coming into crypto to make you rich! They are looking for you to take the other side of their trades. Because in my view that's just the game. Lion's have always eaten gazelles. With that said and as always the best way to profit and manage risk is to educate yourself.

It is estimated that between 60-70% of US equity trading volume is attributed to silicon or machine based traders. Trading with machines offers many advantages. The most obvious is speed. Machines are faster than humans and won't feel tired or let emotions affects their trading decisions. They are programmed to execute strategies for their human owners.

 Another advantage of machine based trading is some strategies can't practically be executed by humans. With increasing computing power, algorithms analyse vast amount of data while trading at higher frequencies. This is impossible for humans to do efficiently. Whilst discretionary or trading by humans still has a place the trend towards ever increasing volume of activity by machines is more likely. Below are examples of crypto trading algorithms I help build for friends and clients.

1. Arbitrage bots: Arbitrage bots profit from price inefficiencies between exchanges, which once identified, buys at a lower price at one exchange and sells at a higher price at the other exchange for a quick profit. Please note in crypto you do not need to transfer the traded asset from wallet to wallet between exchanges that takes too long. Rather alter balances held at both exchanges and settle after the trading period. This trading strategy should not be confused with statistical arbitrage which uses mathematical concepts of co-integration and stationarity to discover pairs of assets that move together in some linear fashion and are mean reverting.

2. Trend following algorithms: These types of algorithms execute trades by detecting a shift in momentum or change in direction of price. They are relatively simple to implement and are very popular among traders who use technical indicators. An example is the dual moving average crossover, which uses 50 and 200 day moving averages crossing to enter and exit trades.

3. Execution Algorithms: When an institution decides to buy big, they won't just place a big market order which spikes up the price thus making the asset more expensive( you be so lucky.). They likely want to reduce the market impact of their trades to get the best prices. Execution algorithms such as Time Weighted Average Price(TWAP) or Volume Weighted Average Price(VWAP) break big orders into smaller chunks to buy at the average price during a time period or volume profile. 

Note some high frequency trading algorithms use mildly sophisticated probability estimates to detect the presence of VWAP or TWAP algos and could "front run" them whilst they still have large chunks of the parent orders left to execute. 

4. Machine learning based algos: Machine learning(ML) is a branch of Artificial intelligence which detect patterns in data and try to make predictions. In trading we analyse data and develop ML models to predict price or other metrics like expected volatility. Increasing and cheaper computing power plus easy access to data makes researching and developing models with high precision possible. One disadvantage is over-fitting, where the model predicts training data accurately but fails to achieve similar results in live trading. 

5. Quantamental algorithms: Quantamental algorithms combine human traders and machine predictions. They are often machine learning models that learn the bets of a human trader and improves accuracy by filtering out the false positives. The human trader makes a call and checks to see if the machine agrees before placing bets. 

If you have an idea for automated trading strategy and need help developing it complete this strategy builder form.