The efficient market theory suggests that the fluctuations in the price of financial instruments are random and do not follow any regular or predictable pattern.
The science behind system trading involves uncovering exceptions to the efficient market theory. These exceptions (or “market anomalies”) are sometimes of sufficient magnitude and frequency to yield a trading bias, or “edge”. The objective of every strategy developer is to identify these market anomalies, exploit the ones that yield clear, quantifiable and actionable edges, and dismiss those that are not statistically significant. The anomalies should ideally have some basis in financial theory and not just be the product of data mining. And, of course, trading edges must be of sufficient magnitude and potential profitability to warrant the time, effort and resources required to exploit them.
There exist two major types of market anomalies:
These are structural and pertain to manner in which information is made available to the financial community. Any inefficiency in the way real-time pricing data is disseminated creates opportunities for those with the knowhow and resources needed to exploit them. Market makers and high-speed traders are able to profit from these execution edges, either thanks to their status as privileged market participants or through the use of advanced information and network technology.
Some execution edge techniques are perfectly legal and can be argued to provide benefits to the overall financial community, namely because of the increased liquidity and reduced spreads they tend to generate. These include:
– Market-making: which involves acting concurrently as both buyer and seller in an effort to earn the bid-ask spread;
– Statistical arbitrage: which looks to exploit any price discrepancy – however small and short-lived – between a financial instrument and its derivatives, or in the price of an instrument traded in different markets or currencies, or in the price of a multiple currency pairs. This technique is used in the equities, bond, futures and foreign exchange markets;
– Trading the news: which uses high-speed and/or co-located computer systems to read and interpret electronic news feeds faster than their human counterparts.
– Index arbitrage: which involves predicting changes in the weights in indices. This is typically used in the equities/ETF market;
Other more controversial techniques, such as spoofing, quote-stuffing and layering were banned as part of the Dodd-Frank financial regulation law of 2010.
Quantitative analysis of historical financial time series allows system developers to uncover statistically significant patterns that can, in turn, be converted into actionable trading edges. The existence of these statistical edges serves to undermine the efficient market theory’s main premise that prices of financial instruments do not follow any predictable pattern.
There are several reasons that explain the presence of statistical anomalies in financial markets. The most obvious one is that market participants are human, and as such most of their actions are driven by emotions. These are at times irrational and lead to undesirable phenomena, such as herd behavior, panic selling and bubble buying. Most “momentum” strategies, for example, are based on understanding and exploiting herd behavior, and many mean-reversion systems are based on understanding and exploiting panic selling, or “extreme oversold” conditions.
Quantitative analysis also allows system developers to study the way markets have historically responded to specific events. These can be recurring (such as FOMC announcements, option expiration days, tax season, etc), or unforeseen (natural disasters, terrorist attacks, etc). Quantitative analysis also allows developers to study how market sentiment has historically reacted to changing environments: seasonality, long-term trend, macroeconomic conditions and government intervention.
Studying the presence and strength of statistical edges in financial time series is done using a number of technical variables. These include price-action, volume, volatility, breadth, relative strength, short-term and long-term trend and many others.
Statistical edges in financial data series can be fickle. Some edges will weaken, others will get stronger. Some persist for several decades, others last just a few years and then flatten out or even switch bias. However, since the bulk of statistical edges are based on the recurring behavior of market participants, many edges do remain intact over decades. The emotions governing the stock market (greed and fear), and the behaviors they create (bubble buying, panic selling) are as present today as they were 50 years ago.
Below we look at a number of techniques used by quantitative analysts to generate statistical edges:
Mean reversion theory suggests that “overbought” conditions will tend to revert back down to the mean, while “oversold” conditions will tend to resolve themselves upwards towards the mean. So what goes up must come down, and conversely, what goes down must go up.
Mean reversion is counter-intuitive to most novice traders, particularly those who have gravitated to trading from buy-and-hold investing. Moreover, some of the trades triggered by mean-reversion strategies can be scary – like stepping in front of a bus – so only seasoned traders have the confidence and experience to pull the trigger.
The forces that tend to push prices back up after a downswing are a) The need for short sellers to cover their positions, b) The perception that dip-buyers have of buying at a discounted price. And the bigger the fall, the stronger these forces. Conversely, the forces that drive prices back down after a pop are a) profit taking, and b) fading euphoria (the Musical Chair Effect). Generally speaking, mean reversion strategies work better on the long side than on the short side.
Gap trading uses mean-reversion theory as its main feature. Most intra-day gap strategies look to “fade” the opening gap expecting it to fill by day end. Other conditions, however, may warrant the strategy to “follow” the gap – i.e. to trade in the direction of the opening gap.
Seasonality is a powerful force that all traders should be familiar with. And while seasonality alone doesn’t always provide an exploitable edge, it is certainly easier to swim with the current than against it. For example, knowing whether the day’s bias has historically been bullish or bearish can help short-term investors trade in the “right” direction and/or help them adjust their position size to accommodate the presence of a head or tail wind. Seasonal effects come in many forms:
Daily Seasonality: each day of the week has its own personality. Some are bullish, some bearish. Some exhibit mean-reverting tendencies, other are prone to follow-through;
Intra-day Seasonality: specific days of the week or month are known to have particularly bullish or bearish mornings or afternoons.
Overnight Seasonality: certain overnight sessions have strong positive/negative historical expectancy.
Monthly Seasonality: some months are particularly bullish, others bearish;
Intra-monthly Seasonality: some days of the month exhibit a bullish bias, others a bearish bias;
Bi-annual: the May to October period is generally considered bearish, while the November to April period bullish. Hence the expression “sell in May and go away“.
Recurring Events: some specific events (annual holidays, macro-economic announcements, etc) have historically shown to have strong biases, some of which can be profitably exploited.
Momentum & Breakout
Momentum edges function on the premise that the same market forces that are in play today are likely to remain in play tomorrow. So a stock in a strong uptrend is likely to continue to rise, and a stock in a strong downtrend is likely to continue heading south.
Momentum strategies are somewhat of a self-fulfilling prophecy. Mutual funds and insurance companies will not be inclined to sell outperforming stocks, in fact they are more likely to invest in them, thereby pushing prices up. Similarly, if a stock has been generally overlooked and under-loved, institutional investors are likely to want to sell them, further accelerating their decline in price.
Pair trading, at least as applied to the equities market, involves hedging a long momentum trade with a corresponding short momentum trade – i.e. betting on the strong horse and betting against the weak one.
Pair trading has the advantage of exploiting the “consensus edge” associated with momentum systems while keeping the trader market neutral to protect him from an unforeseen market downswing.
Trend following is a concept widely used in both intermediate-term investment and short-term trading. The technique plays on the premise that once a market trend is firmly established then it is likely to continue into the future. Traders therefore simply need to determine the direction of the trend, and then “follow” it until the trend changes direction. Trend following strategies are typically both long and short – that is, they can recognize and profit from both rising and falling prices.
The most commonly used entry and exit signals are moving average cross-overs. When the fast moving average crosses above the slower moving average, a long position is taken, and when the fast moving average crosses below the slower moving average the long position is exited, and a short position taken.
Because it is so intuitive, trend following is widely used (and sometimes misused) by novice and advanced traders alike. The technique is easy to implement even without a sophisticated trading platform and entries and exits are easily identifiable on a basic chart. The technique is also easy to explain, which probably accounts for the large number of technical funds and CTAs that sell their products and services on the back of trend following strategies.