- Exploiting Order Flow for the Discretionary Quant - Part 1
- Exploiting Order Flow for the Discretionary Quant - Part 2
- Simple Mechanical Trend Following in the Forex Market
- Is a Reward to Risk Ratio Inherently Better Than Another?
- Robots Aren’t What They’re Cracked Up To Be
- Creating a Trading System Using Neural Networks
- Function Based Trailing Stop Mechanisms
- The Seven Deadly Sins of Automated Trading
- Exploiting the Volume Profile
- Building Robust FX Trading Systems
- Know Your Currencies
- Automating FX Trading Strategies
- Grammatical evolution
- Identifying an Edge
- Interview with Salvatore Sivieri
For advertising, contact
To illustrate the effect of time of day on spreads, Figure 3 shows the best bid and offer in EURNOK over the course of the trading day. As we might expect, the spread widens out during the relatively illiquid Asian session (highlighted area).
As we might also expect, spreads widen during periods of uncertainty. This is illustrated in Figure 4 . Before the release of data in ‘normal’ market conditions the spread in EURUSD was less than 1pt wide and on the release of the data, the price widened to 1.3110/21; well over ten times its ‘normal’ spread. Although EURUSD is the most actively traded currency pair, and this example is during the most liquid time of day, when you would expect the tightest prices, we can see that there are times when this rule doesn’t hold true.
Similar widening of spreads occurs as the market breaks through key levels, such as previous highs, lows and trend lines. Therefore any strategy based on momentum, breakouts, or other key levels, tend to perform much worse than expected in actual trading, than in simulations. Even with the best intentions, and assuming a much wider than ‘normal’ market spread, the assumptions are often far too conservative, compared to real market conditions, at the moment a trade needs to be executed. This behaviour often leads to strategies that seem to work well in simulation, even in sample and out of sample, with seemingly conservative execution assumptions, to lose money in a live trading environment.
As we can see from Figure 1, the most actively traded currency pairs are all against the US Dollar. We also know that liquidity is directly correlated to spreads, so the less liquid currency pairs generally have wider spreads; which in turn means that the cost of execution is greater. However the less liquid currency pairs can be traded via the US Dollar, thereby reducing execution costs.
For example, Figure 5 shows the price for AUD/JPY, as well as its components, AUD/USD and USD/JPY, courtesy of the Currenex ECN.
If we wanted to sell AUDJPY the best direct bid is 93.117. However, we could achieve exactly the same result from selling AUDUSD at 0.93271 and selling USDJPY at 99.839. This would achieve a rate of 93.120834 [0.93271*99.839], an improvement of 0.003834 [93.120834-93.117]. The offer is also improved, saving over half a point on the spread. We can also calculate similar efficiencies by executing NZDJPY and GBPJPY, via their USD components.
You could take the efficiency a step further by putting an offer in the market for AUDUSD and USDJPY. Of course there’s no guarantee of getting paid with a passive order, but if one were able to execute the order ‘passively’, rather than being the aggressor, then the resulting fill would be 93.1336, an improvement of 0.0166 pts.
These savings may seem small, but to put them in perspective, trading just once a day, saving 0.5pt in 1mio AUDJPY, would represent 126pts over the course of the year or ¥1.26mio.We can therefore see how efficient execution can literally make the difference between a strategy making money and losing money.