Countless currency traders wake up every morning with just one question: how to trade FX and make lots of money (ideally without blowing up spectacularly thanks to the 50x leverage).
As it turns out, the question may not be "how" but "when." According to research by Deutsche Bank FX, many investors do not give much thought to the timing of currency trades within the day. But they should, because intraday timing matters greatly on FX patterns. As it turns out, currencies tend to depreciate during daylight trading sessions in their own time zones. Specifically, EUR/USD (and other European currencies) falls in the London session and rises in the NY session while AUD/JPY and NZD/JPY decline in the Tokyo session and rally through the rest of the day.
As to why wake up at 3am Eastern, or when the London session begins? Because that's when a clearly documented pattern of strength in the USD versus European crosses emerges, followed by a traditional subsequent reversal during the NY session, one which is worth between 2-3bp in each direction, if timed correctly. Do this a few hundred times per year with some leverage, and there's your 2 and 20 model right there.
So why should FX traders - especially those who wish to capitalize on EUR volatility - set their alarm for 3am? According to DB's Daniel Brehon, while university students are often night owls, as people get older and assume more responsibilities, they tend to like sleeping at night when they can. So it is not surprising that return volatility is higher for currencies during their daylight trading hours – European crosses during the London session (defined in this paper as 3-11AM EST), USD crosses in the New York session (11AM to 5PM EST) and JPY crosses in the Tokyo session (5PM to 3AM EST) as seen in Figures 1 and 2.
Deutsche then uses 30-minute currency fixings to calculate average returns throughout the day back to 2007 for G10 crosses and 2009 for most EM crosses (RUB back to 2015 and CNH since Sept 2015).
Figure 3 illustrates the average mid-market return for investors who buy USD at 12AM EST against EUR, CHF and GBP, with the 2007-16 time trend removed and carry not included. Figure 4 illustrates the same returns for EUR/GBP, EUR/NOK and EUR/SEK.
The spike in USD versus European crosses during the London session, and subsequent reversal during the NY session, is worth between 2-3bp in each direction, if timed correctly. This margin potentially exceeds spread costs in many cases for these liquid pairs. EUR-crosses follow a less clearly defined pattern, sinking slightly in the London session and rising in the NY session, most likely staying within spread costs.
But while EURUSD was the volatile FX pair of chioice when Europe was blowing up every other month in the 2011-2013 period, now it is all about the USDJPY. Here, by contrast, the dollar-yen pair tends to rise throughout the NY and London trading sessions, meaning that EUR/JPY is really what investors should own in the NY afternoon in the majors. The swing higher in EUR/JPY from 11-5PM EST, and back down from 5PM to 11AM EST, is worth 3bp in each direction on a cross that often trades 1bp wide (Figure 5). The swing on AUD/JPY and NZD/JPY is even more pronounced, falling 4bp and 5bp respectively from 5-11PM EST and rising the rest of the day (Figure 6).
DB offers several explanations this phenomenon. First, yen hedging demand may be driving those crosses lower into the Tokyo Fix. Second, investors may be bidding up these crosses into 5PM EST to capture carry, and then selling off these crosses after carry has been credited to their accounts. The latter reason may be worth 1bp but is unlikely to be dictating price action throughout the day.
Moreover, the dollar tends to perform against all major Asian currencies during the Tokyo session. Figure 7 shows a spike in USD/Asia after 5PM EST following the usual NY session dollar weakness. This pattern has been particularly strong for USD/CNH since the turmoil in August 2015, although clearly the short time frame makes this conclusion provisional at best. The euro performs well against CE3 early in the morning (e.g. CET open) then falls back, although the swing is only 2-3bp, inside spread costs (Figure 8).
But what is most likely happening is that with algos now representing such a dominant portion of FX trading, it means recurring, "self-learning" patterns are all that matters, and as more algos adapt to these recurring trends, the more pronounced they become. As such the spike in the USDJPY around the Japan open has become almost a joke among the FX trading community and yet it persists. Likewise for some of the other observations in this report.
Which doesn't mean one can't profit from frontrunning the frontrunners, aka the math PhD-modeled algorithms. Just do what they do day after day, but do it first.
Then again, as Gartman will be the first to admit, "trades work until they don't." And all that may take for these self-fulfilling patterns to fail is for them to be shown in the open, and for everyone to rush in, trying to trade ahead.
Will these chart be sufficient to end the easy pattern game in FX trading? Perhaps, at which point it will be simply time to do the opposite... until that fails too, and we are back at square one.