How to Pick an Algorithmic Trading Strategy (Stocks)

Share This Post

Share on facebook
Share on linkedin
Share on twitter
Share on email
Share on whatsapp

Index

Free Investment Guide!
Get a list of the best performing algorithmic trading strategies. Sign up to the Aikido newsletter!

Algorithmic trading can yield huge long term returns if done correctly. It utilizes computers to build a portfolio and execute trades. In this context, an algorithm is a set of instructions that determines which stocks to trade and when to trade them.

There are many types of algorithmic trading and quantitative investing techniques that one can employ, ranging from the very simple to the very complex.

In this article, I will give an overview of the different types of algorithmic trading strategies out there and then delve deeper into how you can choose an algorithmic trading strategy.

I’m Shane; algo-investor, FIRE enthusiast, and the founder of two quant startups.

Let’s dive in!

Types of Algorithmic Trading / Quantitative Investing

Factor investing

Factor investing is an investment approach that chooses securities based on attributes that have historically been associated with higher returns. There are two main types of factors: macroeconomic and style. The investing in factors can help improve portfolio outcomes, reduce volatility and enhance diversification.

Factor investing is a form of algorithmic trading for long term investors – it is Aikido Finance’s vehicle of choice. The history of factor investing is very interesting, dating back to 17th century Holland – though really taking off in the 1960s. You can read more about that here.

There are 5 generally-accepted factors:

  1. Value
  2. Size
  3. Volatility
  4. Quality
  5. Momentum

Note: Dividend gets an honourable mention for position number 6, though is not typically part of the standard model.

A fantastic video was made by Investopedia giving an overview of Factor investing. In their words: “factors are to securities what nutrients are to food”.

Value

Value searches for stocks that have low prices relative to their fundamental value. 

Here is the list of common metrics / ratios that an algorithmic trader might use for the value factor:

  • Price-to-earnings
  • Price-to-sales
  • Price-to-book
  • EV-to-EBITDA
  • Price-to-Free-Cash-Flow

Backtested returns of price-to-earnings since 1965. Universe is split by decile (lowest P/E to highest P/E). Source: O’Shaughnessy, What Works on Wall Street

Backtested returns of price-to-earnings since 1965. Universe is split by decile (lowest 10% P/E to highest 10% P/E). Source: O’Shaughnessy, What Works on Wall Street

Size

Historically, portfolios consisting of nano, micro, and small-cap stocks exhibit greater returns than portfolios with just large stocks.

Market capitalization is the go to metric here.

Retail algo traders have a unique advantage here over large hedge funds. This is due to the face that the lion’s share of the gains come from nano-caps (<25M), which are normally off limits for funds as they are too small.

Backtested returns of Market Capitalization since 1965. Universe is split by decile (smallest 10% to largest 10% of companies). Source: O’Shaughnessy, What Works on Wall Street

Momentum

Stocks that have outperformed in the past tend to exhibit strong returns going forward. A simple momentum strategy uses relative returns from three months to a one-year time frame.

Backtested returns of 6-month momentum since 1965. Universe is split by decile (biggest gaining 10% to lowest gaining 10%). Source: O’Shaughnessy, What Works on Wall Street

Volatility

Lower volatility stocks earn greater risk-adjusted returns than highly volatile stocks. This may not mean better returns over all, but rather, a better Sharpe ratio.

Quality

Quality picks stocks based on companies that are considered health. Investors can identify quality stocks by using common financial metrics like a return to equity, debt to equity and earnings variability. 

Typical quality metrics / ratios are:

  • Debt-to-equity
  • Return-on-equity
  • Return-on-Assets
  • Return-on-Invested-Capital
  • Stable growth
  • Margins
Backtested returns of ROE since 1965. Universe is split by decile (Highest 10% ROE to lowest 10% ROE). Source: O’Shaughnessy, What Works on Wall Street

Dividend

A dividend is the distribution of some of a company’s earnings to its shareholders.

Backtested returns of Dividend Yield since 1965. Universe is split by decile (Highest 10% Dividend to lowest 10% Dividend). Source: O’Shaughnessy, What Works on Wall Street

Other Forms of Algorithmic Trading

Event-driven arbitrage:

Event-driven quant strategies involve capitalising on the short term mis-pricing brought about by corporate events. These include earning calls, mergers and acquisitions, capital restructuring and regulatory changes applicable to a given stock.

For example, company A announces that it is acquiring company B. Company B sees an increase in stock price. The trader sells company A and buys company B. strategy would be to take advantage of a merger are within the same industry and highly correlated.

Systematic global macro:

As the name suggests, this form of investing applies a quantitative model to global financial data in order to drive return. Assets held include equities, bonds, currency, and commodities. It has the benefit of providing high levels of diversification. Strategies in this form of quant fund will focus on assets with higher levels of either value, dividend or momentum.

Risk parity:

Balance a portfolio’s risk across asset classes, based on how each asset class tends to behave in different types of environments. The idea is that volatility and losses in one asset class will always be offset by the other asset classes.

Statistical arbitrage:

Arbitrage is the trading of assets with the purpose of profiting from price inconsistency between exchanges. It is a result of market inefficiencies.

Statistical arbitrage (Stat Arb) is a statistical approach to trading assets. It is also known as pairs trading, which depends on trading correlated assets. It involves observing the relationship between two stocks and buying or selling whenever the relationship deviates.

Smart beta:

Similar to factor investing, smart beta strategies are a low cost method of systematic investing, used to manage passive funds such ETFs. Rather than using market capitalisation to weight the positions in a portfolio, factors can be used. Its benefits include the potential for an improved risk adjusted return while being a transparent and simple fund type.

A.I. and Big Data based strategies are the newest type of quant strategy. They attempt to find new sources of alpha using techniques and data that until recently had not been used in the fund management industry. Indeed, many of the techniques discovered by AI may not be explained in terms understood by humans at all!

The Difficulty Levels of Algorithmic Trading

Stock Screener

Coming in at the most basic level of quant, you can use a simple stock screener to build an investment portfolio, like the method outlined in this video.

No-code Algorithmic Trading

This is one of the most exciting areas in fintech at the moment as the space is growing rapidly. Traditionally algorithmic trading has been confined to financial institutions and highly educated individuals. With the rise of no-code algo trading, retail investors are now getting access to the same strategies used by quant hedge funds.

One such platform is Aikido Finance, a no-code algo trading tool for stocks and cryptos. You simply connect your broker, pick a strategy, and in minutes will have a portfolio built, with the trades automatically sent to your broker / exchange.

Code Algorithmic Trading

If you are a software developer or data scientist with knowledge of in trading and finance – this could be the route for you. You will have the full rocketship controls at your fingertips. The results are literally as good as your imagination and skill levels allow. It is a lot of fun coding and testing your own algos, though it can be very challenging – people can spend years getting their algos live.

Two of the most popular platforms are:

How to Choose an Algorithmic Trading Strategy

Throughout this section, we will take examples from The Stable Dividend Strategy, in an attempt to give everyone a firm understanding of algorithmic trading principles.

Choose your geography / sectors

Choosing your country or region can be important in adding diversification and reducing risk. As a rule of thumb in investing, the more concentrated you are (eg. factor, industry, country, asset class), the more volatile the portfolio will be; it will also be correlated very highly with macroeconomic cycles.

Unfortunately, due to the algorithmic trading tools available to retail investors being quite new – most retail algo traders will only have access to the US stock market. Even at Aikido, the universe is currently limited to the USA, however due to our access to the Factset database we are expanding to global scope in 2022.

Investors can also choose if they want a sector weighting; either sector-neutral or a heavy weighting towards a single industry. Needless to say, sector neutral will be far less volatile.

Choose your Factor Explosure

This is hands down the most important part of the algorithmic trading process. Different trading strategies are weighted in favour of different factors. For example, The Stable Dividend Strategy focuses on the Dividend and Quality factors. Whereas Deep Value Blend focuses solely on the Value factor.

If your goal is to maximise long-term performance and increase your risk vs. reward, you will want to mix several algorithmic trading strategies together in order to get good factor diversification.

If you fancy your hand at factor-timing, you could check out this whitepaper from the FTSE. It discusses the cyclical nature of factors and the best times historically to invest in different factors

OSAM Research

Above is a great guide to which factors to invest in at different parts of the market cycle. Consider this a cheat sheet from one of the greatest minds in quantitative investing: James O’Shaughnessy.

AQR also wrote a superb whitepaper on a Factor Momentum strategy they built which invests in factors depending on how well they are currently performing. Factor Momentum Everywhere” won “Best Quant Paper”  in the 2019 Savvy Investor Awards.

Choose a Rebalance Period and Portfolio Weighting

Rebalance Period

Some algorithmic traders / quantitative investors will choose to use a set rebalance period for their portfolio. This is the easiest way to keep a portfolio up to date with the algorithm.

For traditional investors, keeping a portfolio up-to-date (AKA rebalancing) is really hard work! I know from personal experience… It would take hours of my time every month to review old stocks, purchase new stocks, ensure correct weighting, review exposure and risk, etc. etc. It was really hard work and I really didn’t want a second job.

Luckily, for algo traders and quantitative investors, it’s easier; a lot easier! You utilize computers to do the whole process for you – saving hundreds of hours and ensuring you don’t sprout any (more) grey hairs.

Typically the algo trader will choose a rebalance period of weekly, monthly, quarterly, or yearly. More you rebalance (ie. keep your portfolio up to date) is not necessarily correlated with higher returns. Indeed, sometimes the more you look, the more you lose.

Many algorithmic traders use a intra-day, signal-based method to keep a portfolio up to date. eg. buy 10 shares of AAPL when 50 day SMA crosses the 100 day SMA. This means that you will be trading at any time (mid-month). This is a more risky method of trading as you are very likely to over trade and take away from the underlying algorithm. You could also be receiving high fees for this frequency of trading. However, if this signal-based method is paired with a very solid underlying factor-based strategy, it can be very powerful indeed.

Portfolio weighting

This is simply the weighting on each stock in your portfolio. Most investors will choose to use an equal-weighting – where the same amount is invested in each stock. The most common portfolio weighting methods are:

  • Equal-weighted
  • Volatility-weighted
  • Momentum-weighted
  • Value-weighted
  • Kelly-weighting (Kelly Criterion)

Ironically, the S&P500 and most of the best known indexes use a market cap weighting (ie. the biggest weighting is on the largest stocks); this is a terrible idea for retail quants! As we saw earlier, this goes against the size factor entirely. Though it might yield lower volatility, it also tends to yield a lower sharpe ratio (risk vs. return).

Choose your Risk vs. Reward

There are several very important metrics to look at when picking a strategy. Those are:

Taken from Top Quality High Momentum Strategy

The two Most Important Metrics

If I were to boil it down to the two most important things to look at, I would say Sharpe ratio and the Base rates.

The Sharpe ratio compares a strategy’s return to its risk (volatility). A higher Sharpe ratio is desirable for investors as it implies a higher risk-adjusted return.

Base rates are a little more complicated, but equally as important. They are frequently overlooked by novice algo traders. The base rates is the amount of time a strategy has outperformed the benchmark (eg. S&P500) in in-sample and out-of-sample tests. Below are sample base rates showing rolling 1 year, 3 year, 5 year, and 10 year excess returns. If you are a long term algorithmic trader, you want to know that your strategy consistently beats the market in the long term.

Do your Research

Before you choose an algorithmic trading strategy, it is important to do your homework. What research papers and academic journals is the strategy based on? For example, at the bottom of any Aikido Strategy, you will get a list of reference papers that the strategy is based on. Read through the research and use strategies with reputable sources.

Conclusion

Choosing a trading strategy can be a lengthy process, and there are things I didn’t touch on in this article. In this article, I have tried to break it down into its most important components. Everything discussed above must be considered.

The most important thing you can do for long term success as an algo trader is to use several different strategies (algos) at once. After all, “Diversification is the only free lunch” – Harry Markowitz

Thanks for reading,

Shane

More To Explore

Share This Post

Share on facebook
Share on linkedin
Share on twitter
Share on email
Share on whatsapp
Free Investment Guide!​
Get a list of the best performing algorithmic trading strategies. Sign up to the Aikido newsletter!