In this article, we will look at the best algorithmic trading strategies.
Before we do, we have to define algorithmic trading and how it differs from discretionary and systematic trading.
Contents
- What Is Algorithmic Trading?
- What Strategies Are Used In Algorithmic Trading?
- What Is The Best Strategy For Algorithmic Trading?
- The Work Of The Algorithmic Trader
- FAQ
- Conclusion
What Is Algorithmic Trading?
Think of systematic and discretionary trading as two ends of a spectrum, with most investors somewhere in between.
Discretionary trading is based on the personal decisions of the trader as to what he or she thinks the price will go.
Systematic trading is based on predefined rules based on price movement or other market-generated information.
There are no decisions for the trader to make. Hence, emotional trading is reduced as much as possible.
The trader simply follows a set of written rules.
They match the conditions seen in the market with the predefined patterns of the system.
When they see a match, they execute the trade. And follow the exit rules.
Algorithmic trading is systematic trading with the trader completely taken out of the picture. Instead, the computer will perform and execute the trades based on an algorithm.
An algorithm is a set of predefined rules and instructions for a computer program to execute.
What Strategies Are Used In Algorithmic Trading?
Algorithmic strategies fall into the following general categories.
Mean-revision strategies:
Also known as counter-trend strategies, these will attempt to detect when the price is at either extreme of the range.
If it is at an extreme high, it will generate a sell signal assuming that the price will revert lower to its mean.
Vice versa, it will buy if it detects the price is too low.
Momentum strategies:
These are the opposite of mean-revision strategies.
When the price makes a move, it assumes that the move will continue.
They take advantage of the short-term momentum of the move.
Trend-following strategies
These are similar to momentum strategies, except at a longer time frame.
They try to stay in the trend by monitoring various indicators such as moving averages, etc.
Hedging strategies
These strategies attempt to capture profits in the event of black swan events or major selloffs.
While they may be losing a small amount of money during normal times, they are designed to pay off big on those rare occasions.
Pair-trading strategies
These strategies look at two or more correlated assets.
If their prices diverge, there is a higher statistical probability that they will eventually converge again.
They are a form of mean-reversion.
Statistical arbitrage strategies
These strategies are similar in that they look for price divergences and price inefficiencies by taking the statistics and prices for hundreds of assets across exchanges, cash and futures, and options markets.
These tiny inefficiencies in the market do not exist for a long time.
Hence, these strategies are typically employed by institutions with high-frequency trading computers that make trades at sub-second timeframes.
These strategies are typically unable to be employed by individual retail algorithmic traders because they cannot achieve the kind of speed needed for these trades.
What Is The Best Strategy For Algorithmic Trading?
The best strategy for algorithmic trading is to run multiple non-correlated algorithms simultaneously.
This diversification is the key to reducing risk and drawdowns and provides a smoother equity curve.
For example, this could mean running a mean reversion algo and a momentum-following algo simultaneously.
Or one can run a mean reversion algo, a momentum algo, and a hedging algo simultaneously.
Or one can swap the momentum algo with the trend-following algo.
The reason that you want to run multiple algorithms at the same time is that the best-performing algo at any given time will depend on the market condition at the time.
When markets are in a strong trend, momentum and trend-following strategies work the best.
When markets move sideways, or prices stay within a range, then mean-revision strategies are the best.
Most of the time, the market is either in a trend or in a range and will alternate between the two.
The problem is you don’t know when the markets will switch.
Therefore, having both types of algo running, they each will pick their own spots to enter.
When the market switches regime, one algo will pick up more entries while the other algo are finding fewer entries. And vice versa.
There may be rare occasions (such as when markets are in selloff mode) when both algos are losing money because markets are neither in an orderly trend nor in consolidation.
This is when the hedge algo shines and kicks into action.
The key is that the algos are non-correlated.
Algo strategies are like foot soldiers, each having their own specialized task depending on what the market throws at them.
The Work Of The Algorithmic Trader
The job of the algorithmic trader is to define and refine the rules of each of the algos, to determine how much capital to allocate to each algo, and to decide how many and which algorithms to put together to run simultaneously.
They also need to make sure to put risk management rules into the algos in the event that they run into an unforeseen situation.
In order to refine the rules, the systematic trader runs a large number of back-tests with historical data.
The results are plotted as a P&L equity curve, which they want to see upward sloping with low volatility and low drawdown.
They need to balance how upward sloping the curve is with how steep each of the drawdown dips is.
Are they willing to sacrifice some returns in order to reduce the drawdowns?
These are the questions needed to be answered individually.
If they don’t like the look of the equity curve, they tweak the rules and run the test again.
And again. And again, until they see something that they like, or at least find acceptable.
They need to avoid the problem of “over-fitting.”
This is when rules are tweaked so much and overly modified that they end up working well on the sampled historical data but may not work well in future never-seen data.
That is why back-tests are done only on a portion of the historical data available.
The other untouched data is used to test the algorithms.
Forward Walk
The next step is to “forward walk” the algorithm.
That is to run the algorithmic trading system on live data in real-time, but not with real money.
Finally, then is the time to put the systems out in the live markets.
The system can be fully automated or semi-automated.
The fully automated system is run by computers that generate the signals, connect to the broker via APIs, and place the trade in real-time with no human intervention.
The semi-automated system only generates the signals.
The trader looks at the signal when they get the text alert or when they get a chance to get to a chart.
They confirm that the signal makes sense and then place the trade manually.
Although some will argue that this is not a true algorithmic trading system because now you have just introduced human discretion into the picture, admittedly so, we can say this is a semi-algorithmic trading system then.
FAQ
Is Algorithmic Trading Profitable?
Yes, it is. Algorithmic trading systems have been in use for many years. If they are not profitable, we would not see them continue to exist for so long.
In the United States, around 80% of trading is done by algorithmic trading. And their use has been increasing.
These systems have already taken over the job of market makers.
This is not to say that all the algorithmic trading systems are profitable. I’m sure some of them are not and have since disappeared.
This is not to say it is easy to build either. See the article on how to get started with algorithmic trading.
Is algorithmic trading legal?
Yes, it is. It is not a secret that many of the top hedge funds (including Ray Dalio’s Bridgewater Associates and Two Sigma Technologies) use algorithmic trading.
How to get started in algorithmic trading?
Learn a profitable trading strategy, programming, and data science. Find details and some resources in “How do I start algorithmic trading?”
Conclusion
We’ll end with a quote from one of the books in my library. The book is titled Automated Stock Trading Systems by Laurens Bensdorp. On page 14, it says:
“combining these systems and trading them simultaneously improves performance exponentially. What matters is not the performance of individual systems, but the nearly magical effect of combining them for exponentially better results.”
We hope you enjoyed this article on the best algorithmic trading strategies. If you have any questions, send an email or leave a comment below.
Trade safe!
Disclaimer: The information above is for educational purposes only and should not be treated as investment advice. The strategy presented would not be suitable for investors who are not familiar with exchange traded options. Any readers interested in this strategy should do their own research and seek advice from a licensed financial adviser.
Original source: https://optionstradingiq.com/best-algorithmic-trading-strategies/
