This is a meditation on the essence of what makes for good quantitative trading. From a purely intellectual viewpoint this has attracted attention and has led to questions about what is at the heart of good quantitative models.

The Search For Structure

Whether a quant modeler is able to articulate it or not, eventually good algorithmic trading is about a search for structure in the noisy data of markets. It is about finding patterns, regularity or pockets of predictability. Here is a simple example of what is meant by structure. Let’s say that we observe that whenever the market goes up two days in a row, it usually goes up the third day. If this happens quite often, we have found the pattern or regularity we were looking for. The trading strategy immediately follows. If the market goes up two days in a row then buy at the close of the second day and sell it at the third day’s close. If only!

It is easy to get fooled by randomness and see patterns that in hindsight seem nonsensical at best. Technical analysis books are strewn with all sorts of patterns with colourful names, most of which will not stand even mild statistical scrutiny much less any systematic way of teaching a computer to identify the patterns.

To combine the problem markets are to a first approximation just noise. Yet they fail statistical tests of randomness in enticing ways. That is what randomness would prescribe except that the tail of the distribution is much fatter, i.e. periodically markets have much, much larger moves than a normal distribution.

When we study the dynamics of daily moves rather than the aggregation in a histogram, things get even more interesting. If markets were indeed a random walk, today’s move would be independent of yesterday’s move. Not so, say econometricians where there is a cottage industry of models (called the GARCH models) which suggest that big moves (in absolute terms) are usually followed by big moves i.e. there are times when markets remember yesterday’s moves. Greed and panic leave the markets quite shaken with bursts of volatility! This is a structure that has been modelled by academics and used by practitioners to model volatility for derivatives. From a perspective of trading it can be used not really as a strategy but as a signal of the onset of volatility and also lowering of trade size during this period.

The academics have discovered more that are about financial time-series. It may be hard to figure out what the future holds for a single time-series, but academics have found techniques (called co-integration) that allow us to create portfolios of longs and shorts of different assets in such a way that the value of the portfolio oscillates around a mean value much like a sine wave. Again, this is the regularity a modeler is looking for and the strategy is clear. Buy when it reaches the bottom and sell it when it reaches the top. The trouble alas is that markets change, the relationships break down and the sine wave starts moving out of its band or narrows to the point where the strategy is unprofitable. It is a non-stationary world. The message to a quant is clear – there is structure to be exploited, but beware of noise, fat tails and non-stationarity.

Market Insight

The search for structure normally starts with an insight about markets. This could come from traders or from academia or actually from a combination of mathematical, financial, and computational skills. A good instance from academia is the so-called factor model for stocks. The goal here is to be able to explain the daily move of a stock by a variety of factors. The factors could be broad market indices, macro-economic, or other technical factors. However, the crucial test of the model is that the part of the daily return that is unexplained by the factors (the residual), should again behave like a sine wave. We are back to a world of regularity.

The so-called carry model for currencies is another example of a model inspired by traders. When interest-rates in Australia are significantly higher than in Japan, it is tempting to borrow yen at cheap rates and invest in Australia. This requires considerably long holding periods and is prone to periodic but violent corrections. Where a model differs from a human being is that it is impervious to fear and greed. Even-though the origin of the idea started from traders, the modeler tests the strategy on past history, recognizes the deficiencies of a naive strategy, uses the full power of math.

The Hazards Of Modelling
Once the trade idea emerges, the task then is to start coding this insight into a strategy by teaching the machine when to get into a trade, how much to trade and when to get out of a trade. To simply teach a computer to buy a portfolio “roughly” when the portfolio is at the bottom of the sine wave requires a lot of code and parameters that control decision making. Over-parametrization is the bane of many a model.

In our anxiety to be able to capture all trades in the past, we sometimes inadvertently make the system overly complex and make it worse by choosing parameters that work best for past data but perform poorly in the future – overfitting as it is called by the practitioners. There are other subtle and not so subtle ways in which a modeler can trip up. Having too many models of a similar type, or looking ahead in the data are common mistakes. Chasing a phenomenon that has no discriminatory ability in picking the inflection points of a time-series is a mistake that even a seasoned modeler can make.

A key to modelling is to incorporate realistic transaction costs, otherwise it is easy to come up with strategies especially in high-frequency that seemingly make a lot of money in theory but are sure money-losers in reality. Brokers make a living out of transaction costs – it is important to remember that in modelling!

No matter how well a model may capture a phenomenon, the modeler should be prepared for what is called the “tail risk” – that rare but catastrophic event that accounts for the fat tails in the returns. This takes the form of deleveraging, reducing the trade size, and stopping out of certain trades or models. Risk management is usually an integral part of the modelling process and the triggers that lead to reduced trading are found by experimentation. All of this is an integral part of any good quantitative model.

If the models are added carefully, they can enhance the Sharpe ratio where one model compensates for the under-performance of another.

Pros and cons of quant trading

The biggest benefit of quantitative trading is that it enables you to analyze an immense number of markets across potentially limitless data points. A traditional trader will typically only look at a few factors when assessing a market, and usually stick to the areas that they know best. Quant traders can use mathematics to break free of these constraints.

By removing emotion from the selection and execution process, it also helps alleviate some of the human biases that can often affect trading. Instead of letting emotion dictate whether to keep a position open, quants can stick to data-backed decision making.

However, quantitative trading does come with some significant risks. For one thing, the models and systems are only as good as the person that creates them. Financial markets are often unpredictable and constantly dynamic, and a system that returns a profit one day may turn sour the next.

For this reason, quant requires a high degree of mathematical experience, coding proficiency and experience with the markets. So it certainly isn’t for everybody.

Want to try out using an automated system, but not sure if you’re ready for quant? Find out more about algorithmic trading.

Quantitative Trading Defined

Quantitative strategies sometimes implement a high frequency approach, arbitrage positioning and trading algorithms that’s supported statistical averages. Many market analysts argue that Quant trading artificially adds to cost volatility, due to the fast in-and-out positioning that has little or nothing to try and do with economic fundamentals. Some won’t know that indicators and oscillators (such as the MACD or RSI) also can be characterized as quantitative analysis tools, and this essentially means Quant strategies and technical analysis are very close cousins.

Trading opportunities are largely defined by probability — mathematical computations that decide the chances prices will rise or fall based on current market events. Price, volume, and moving averages are common inputs used in trading models. In early computer trading Quant models were reserved for hedge funds and enormous financial institutions. Meanwhile today, easily accessible applications like MetaTrader or Trading Stations are perfectly capable of executing Quant strategies and these methods can even be implemented from remote servers (meaning your personal computer doesn’t even need to be on to open and close trades).

Common Strategies

Quant strategies might seem new because modern computers can put incredible processing power in nearly every home. But the strategies behind even the most modern strategies have been in place for nearly a century. So just because you might see an advertisement for a “90% effective EA” doesn’t mean you are seeing a new and innovative strategy. If anything, it is probably the opposite. This can be true even for strategies that are back-tested. What worked in the past might give you no insight into what will happen next, which is a problem because this is what is truly needed to make profitable trades based on price behavior.

Arguments Against Quant Strategies

People selling Quant strategies as Expert Advisors will, by almost definition, possess techniques that have strong backtesting results. But if the markets were as simple as that, everyone would be a successful trader and also the guy inquiring for change the street corner would have a penthouse apartment. Life — and the financial markets — don’t work that way. So, opponents of Quant strategies will argue that if you want to have confidence in your next trade you will need to monitor the actions of human beings, because human beings are what determine the net-worth of market assets.

This is one among the reasons why many traditionalists label Quant EAs as “black boxes,” and it’s beyond debate that there are even as many losing EAs as there are winning EAs. And when EAs fail, they tend to fail on a massive scale. One of the most reasons this may occur happens when market dynamics change, and historical events don’t have any way of including future events. For example, when the Swiss National Bank enacted a price floor in the EUR/CHF at 1.20, market values in the currency pair rose by nearly 1,000 pips in a matter of minutes. There would have been no way for an EA to predict this type of event, so it would have been very likely you would have accrued losses if you were using an EA to trade the CHF/EUR (or any highly correlated forex pair) at the same moment.

Quantitative strategies can be constantly re-defined and streamlined to account for potential market occurrences, but the stark reality is that it would be impossible to account for all possible occurrences in an EA. When markets are experiencing above average volatility, these strategies also tend to suffer. In these cases, buy and sell signals are sent so often that rising transaction costs can significantly erode your potential for gains.

Arguments Supporting Quant Strategies

On the other side of the debate, the main strength of Quant strategies is that they operate using the highest level of discipline and objectivity. If your Quant model accurately forecasts what is going to happen within the market, your positions will use the available quantitative data to successfully exploit market inefficiencies and generate profits. Quant models can be composed of as little as one or two inputs (such as price activity relative to a moving average, or overbought/oversold readings on an indicator tool), or much more complex (which can mean thousands of inputs working in conjunction with one another).

Another benefit is that EAs are ready to pick-up on trend activity just as it develops. A running EA is consistently monitoring price activity and market scenarios to spot opportunities. Human beings are simply not capable of this level of attention or awareness. Quant models are often set to research large groups of assets simultaneously, whereas an individual could monitor only some at any given moment. These models will then rate scenarios altogether cases, often using numeric or alphabetical grade levels (such as A-F, or 1-5). Assets with the very best ratings trigger long positions, assets with the lowest grade levels trigger short positions. This simplifies the trading process and allows traders to position themselves only in the most extreme cases (which offer the highest probability for gains).

But Quant models are capable of doing much more than simply opening and closing positions. Trades also can be structured to account for correct risk levels (outlining stop losses, profit targets, and position sizing). Once trades are opened, it is also possible to limit exposure in correlated assets. For example, long positions in both the USD/JPY and USD/CHF would mean the account is taking over double exposure within the USD. Proper diversification is needed for any successful strategy, so it is important for those running Quant strategies through EAs to keep this in mind in order to avoid over-leveraging.

Quantitative trading strategies

Quantitative traders can utilize a huge amount of strategies, from the simple to the incredibly complex. Here are six common examples you might encounter:

• Mean reversion
• Trend following
• Statistical arbitrage
• Algorithmic pattern recognition
• Behavioural bias recognition
• EFT rule trading

Mean reversion

Mean reversion can be defined as the financial theory that predicate that prices and returns have a long-term trend. Many quant strategies fall under the general umbrella of mean reversion.  Any deviations should, eventually, revert to that trend.

Quants will write code that finds markets with a long-standing mean and highlight when it diverges from it. If it diverges up, the system will calculate the probability of a profitable short trade. If it diverges down, it will do the same for a long position.

Mean reversion doesn’t have to apply to the price of a single market. Two correlated assets, for example, may have a spread with a long-term trend.

Trend following

Another huge category of quant strategy is trend following, often called momentum trading. Trend following is one of the most straightforward strategies, seeking only to identify a significant market movement as it starts and ride it until it ends.

There are lots of different methods to spot an emerging trend using quantitative analysis. You could, for instance, monitor sentiment among traders at major firms to build a model that predicts when institutional investors are likely to heavily buy or sell a stock. Alternatively, you could find a pattern between volatility breakouts and new trends.

Statistical arbitrage

Statistical arbitrage builds on the theory of mean reversion. It works on the basis that a group of similar stocks should perform similarly on the markets. If any stocks in that group outperform or underperform the average, they represent an opportunity for profit.

A statistical arbitrage strategy will find a group of stocks with similar characteristics. Shares in US car companies, for example, all trade on the same exchange, in the same sector and are subject to the same market conditions. The model will calculate an average ‘fair price’ each stock.

You would then short any companies in the group that outperform this fair price, and buy any that underperform it. When the stocks revert to the mean price, both positions are closed for a profit.

Pure statistical arbitrage comes with a fair degree of risk: chiefly that it ignores the factors that can apply to an individual asset but not affect the rest of the group. These can result in long-term deviations that don’t revert to the mean for an extended time. To negate this risk, many quant traders use HFT algorithms to exploit extremely short-term market inefficiencies instead of wide divergences.

Algorithmic pattern recognition

This strategy involves building a model that can identify when a large institutional firm is going to make a large trade, so you can trade against them. It’s also sometimes known as high-tech front running.

Nowadays, almost all institutional trading is done via algorithms. Firms want to make large orders without affecting the market price of the assets they are buying or selling, so they route their orders to multiple exchanges – as well as different brokers, dark pools and crossing networks – in a staggered pattern to disguise their intentions.

If you build a model that can ‘break the code’, you can get ahead of the trade. So algorithmic pattern recognition attempts to recognize and isolate the custom execution patterns of institutional investors.

For instance, if your model flags that a large firm is attempting to buy a significant amount of Coca-Cola stock, you could buy the stock ahead of them then sell it back at a higher price.

Like statistical arbitrage, algorithmic pattern recognition is often used by firms with access to powerful HFT systems. These are required to open and close positions ahead of an institutional investor.

Behavioral bias recognition

Behavioral bias recognition is a relatively new type of strategy that exploits the psychological quirks of retail investors.

These are well known and documented. For example, the loss-aversion bias leads retail investors to cut winning positions and add to losing ones. Why? Because the urge to avoid realizing a loss – and therefore accept the regret that comes with it – is stronger than to let a profit run.

This strategy seeks to identify markets that are affected by these general behavioral biases – often by a specific class of investors. You can then trade against the irrational behavior as a source of return.

Like many quant strategies, behavioral bias recognition seeks to exploit market inefficiency in return for profit. But unlike mean reversion, which works off the theory that inefficiencies will eventually rectify themselves, behavioral finance involves predicting when they might arise and trading accordingly.

ETF rule trading

This strategy seeks to profit from the relationship between an index and the exchange traded funds (ETFs) that track it.

When a new stock is added to an index, the ETFs representing that index often have to buy that stock as well. If ABC Limited were to join the FTSE 100, for example, then numerous ETFs that track the FTSE 100 would have to buy ABC Limited shares.

By understanding the rules of index additions and subtractions and utilising ultra-fast execution systems, quant funds can capitalise on this rule and trade ahead of the forced buying. For instance, by buying ABC Limited stock ahead of the ETF managers and selling it back to them for a higher price.

Conclusion

Forex traders using technical analysis strategies as the basis for their positions will often run into practitioners of Quantitative techniques. There is a possibility the it can easily open up a “can of worms” because of some significant similarities between classical technical analysis and Quantitative strategies.

The main difference is that Quantitative strategies use automation to remove most of the human element, but also this characterization is not entirely true because all quantitative strategies are actually human developed technical analysis techniques that are triggered by computer signals.

It’s good to have a look at some of the strengths and weaknesses of the Quantitative approach, so that technical analysis traders can decide whether or not these strategies are an appropriate addition to daily trading.