The market and quant views for 2018

20171229 Year Ahead

I’ve spent of most of my career thinking about modelling markets. Typically, I develop indicators for estimating economic sentiment, or actual trading models usually with an intraday or daily time horizon. I end up using all sorts of different factors in my models. However, at the same time, I like to follow what’s happening in the market and form my own views. True, I’m not a discretionary trader risking capital on these views. At the same time, it helps to provide me with a framework on how to view markets. Invariably, I get ideas on precisely what to model, because I follow markets. Without having an idea what’s happening it can be difficult to form a hypothesis behind a model. I’m a strong believer that the best way to build a trading model is by having both a strong technical understanding (a combined knowledge of mathematics and coding) and an economic intuition about how markets behave. As a result, at this time of year, I try to write down a few thoughts about the year ahead and my expectations around them.


Year ahead thoughts are generally wrong!


Whoever you talk to in the market, they will tell you that outlook pieces are invariably wrong, and the best way to profit from them is usually to fade the move. I hate to agree with them, but the data does tend to suggest this! I recall doing a study at Nomura on the subject, looking at the currency moves during January versus their moves over the rest of the quarter. The idea was that the moves at the beginning of the year were driven by so-called year ahead trades. For those currency pairs that moved the most in January, historical data suggested that, it was best to fade them during the rest of the quarter. I’ve also done research on forecasts more broadly, which suggested that in 60% of quarters, the direction was wrong for EUR/USD for consensus forecasts. Admittedly, the sample sizes were pretty small for both of these studies and I need to update them with more recent data, but the results nevertheless seem to tally with what is general market intuition. You can read my own year ahead thoughts from last year here (was a mixed bag in terms of accuracy!) and below.


What about the dollar?


I suppose for the rest of the piece, we kind of need to ignore my initial point! For currency markets, a key view is obviously where the dollar will go. For the past year, the dollar has traded relatively weakly. EUR/USD has had a significant move through 2017, even if over the past few months its been range bound. My former boss, Jens Nordvig has an interesting recent interview on Bloomberg TV discussing why the dollar has been weak in 2017. The Eurozone economy grew quicker than expected, giving rise to the #Euroboom hashtag on Twitter (h/t @Birdyword for coining that!). Furthermore, European politics were market friendly, with Macron comfortably winning the election in France. Going into 2018, the Fed looks ready to continue the hiking cycle, potentially with another 3 or 4 hikes. The terminal rate is likely to be in the region of 2.5%, so we are getting closer to the end of the hiking cycle. The Fed have managed to normalise policy relatively smoothly without shocking the market. The ECB have started to go down the same path, with the reduction in asset purchases due to start in January 2018 (announced in October). Obviously, the ECB are at much earlier stages of their normalisation, with a lot more to go. Hence, it seems reasonable to expect at least some further upside in EUR/USD during 2018. It’s also worth noting that a lot of the USD move higher occurred at the end 2014, before the Fed had even hiked, although it had already started tapering asset purchases and ended them in October 2014, so maybe it’s now the EUR’s turn in this process? Going forward, the USD bullish effect from repatriation flows is unlikely to be huge, given most US companies are likely already holding USD offshore, rather than foreign currency (in 2004, USD strengthened during HIA).


On USD/JPY, the US presidential election helped to push it above 110 in late 2016, with the whole reflation theme which was dominating markets. For most of the year it’s been range bound, as have US Treasury yields, which is a big driver for this cross. Consensus forecasts for UST 10Y yields are around 2.88 for end 2018. I suspect that’s a bit on the optimistic side, unless US inflation begins to pick up aggressively. For 2018, I suspect that USD/JPY will find it difficult to make new highs and we could see a move lower. It’s also difficult to see BoJ expanding its stimulus at any point and we could see signs of withdrawing the stimulus maybe next year (or possibly in 2019).


On sterling, it seems we’re veering away from a hard Brexit in the near term, and at the very least the talks seem to be going in a better direction and there’s going to be a transition period. The transition period is due to end in 2020/2021. A general election (if there government last till then) is due in 2022. This may increase the incentives for a “smoother” Brexit if there is an election just around the corner, with the current state of the economy in voters’ minds. Speculative positioning in GBP/USD, according to CFTC data had been very short most of the year, it’s now slightly net long, so there’s still a lot of room to build up exposure and the market hasn’t yet fully bought into the bullish GBP story. Ok, I’m not saying things are totally rosy in the UK, more a case of it being better than expected.


Populists might not be a populist as you think


Politics has again been a key factor in investors’ thoughts this year. From a markets perspective though, how populist are the economic policies of populists, when they’re in power? In the USA, Trump has managed to get through tax cuts. However, it’s not clear how “populist” these tax cuts are. Furthermore, had there been another Republican in the White House, it seems likely that tax cuts would also be a major policy. In any case, the tax cut narrative seems useful in helping equities higher in 2018! Admittedly, other Republicans might not have declined to sign TPP, but for the most part, there hasn’t been as radical a change in economic policies as there could have been (eg. massive trade barriers being erected). In the UK, Brexit will result in major change, but even there, the idea of a very hard Brexit seems less likely as mentioned earlier. Sure, stuff like “Blue” passports are going to return, but I doubt that’s going to impact GBP much… Maybe the economic aspect of populist policies are just less “popular” than the parts designed to impact social policies?


In terms of possible political risks for 2018, one place could be Italy, where we have an election on 4 March. Polls suggest that the “Five Star Movement” would be the biggest party, but only just, and would to have to rule in a coalition. Whilst, some modicum of success for the “Five Star Movement” could put pressure on EUR/USD, I think that it would be short-lived, if they were forced into a coalition. Also it’s worth noting that the Five Star Movement’s leader has noted that a referendum on Euro membership would be a “last resort”. Perhaps as a consequence of Brexit, the Euro-sceptic view has been less of a vote winner for populists in Europe. Another possible risk is a flare up of tensions in North Korea, which we shall all be watching closely. It is clearly in no one’s interests for it to escalate


The continuing march of quant, alternative data and machine learning


One thing that has been very apparent to me more broadly, is that quant is now a hot area. For most of my career it was something that purely for quant funds. However, it’s not just quant funds which are into this space as well. Increasingly, I’ve noticed interest in this area from discretionary traders. It’s not just about creating trading models to give you buy and sell signals, but about using quantitative techniques to give you more of insight into the markets. As well as more traditional quantitative trading models, like trend following or value based strategies, we also have a lot more alternative datasets. Many alternative datasets consist of text, which need to be structured in an appropriate way to extract some value from them. For example, I’ve created an index for quantifying sentiment of Fed communication using natural language processing and it has a reasonable fit to recent changes in US Treasury 10Y yields. I’ve also done a research paper for Bloomberg, looking at how you can extract FX signals by using a machine readable dataset consisting of articles written on Bloomberg News. There is some short term momentum that can be gauged for trading FX, I’ve found. The research is going to be published this week by Bloomberg. Another idea you can try is to fade extremes in news sentiment.


There’s also the whole field of extracting value from social media. I’ve for example tried to augment nonfarm payrolls forecasts with data extracted from Twitter feeds. The list of alternative datasets you can use is massive. In practice though, it takes a lot of time to go through the datasets, and just because a dataset is unusual, doesn’t necessary mean it has trading value. This is the challenge, even for large quant funds, sorting through all these datasets to find those that are truly valuable.


You also have machine learning, which is another big buzzword. In recent weeks I’ve written a lot about this on my blog. The idea is to extract signals datasets, that you might not ordinary think of. Caution does need to be used though, to avoid data mining too much. Techniques like deep learning also need a lot of data to “work”. There have been many successes in machine learning outside of finance. You’ve seen DeepMind’s AlphaZero for example using reinforcement learning to learn to play chess, and beat the best existing chess computer. We use machine learning all the time, eg. when we text on a phone and it predicts the next word, or in automated language translation. As yet, it’s not made the leap in full to finance. You also have the difficulty that financial data is not stationary, so you keep having different regimes. It’s still early days for machine learning in finance, but let’s see how it plays out in the years ahead.


I think the real risk of quant techniques, is applying them blindly without any knowledge of how they constructed. You still need to take time to understand how they work before using them in your investment process. More grey box, as opposed to black box.


Bitcoin – well I had to talk about it!


I rarely get asked by my friends outside of finance, about my thoughts about EUR/USD! The question is invariably about Bitcoin. It is somewhat ironic that we group together the likes of Bitcoin, Ethereum and Litecoin under the term of cryptocurrencies. The use case for Bitcoin is that it can be used as a store of value and also a medium of exchange. However, the massive amount of volatility and relative lack of liquidity makes it difficult to function at a currency. There’s also currently the issue around the relatively high cost of doing Bitcoin transfers, which reduces the attractiveness of using it anyway versus traditional payment transfers. We should note that other cryptocurrencies such as Ripple however do offer cheaper rates of transfers.


However, I do think longer term cryptocurrencies will be used more generally a proper “currencies” alongside fiat currencies, maybe even cryptocurrencies issued by central banks? Long term, I have no idea, which cryptocurrency will dominate. Bitcoin just has first mover advantage. As to their value, it is very difficult to quantify, given the relative lack of fundamentals (see my earlier post). One factor which does seem to work is that of trend, which is pretty universal across asset markets. There’s also the issue about institutional involvement which is likely to increase in 2018, and that should help support the price higher. On the flipside, the more institional involvement, the more risks around the impact of a Bitcoin unwind impacting the market more broadly. At present, if Bitcoin falls 25%, VIX frankly doesn’t care. The risks are if we end up seeing a lot of institutional leverage (and also retail leverage) chasing around Bitcoin, it could get messy. Whilst, supply for example of Bitcoin might be limited, the number of new cryptocurrencies is not. Ultimately what sustains the price is belief among market participants that any cryptocurrency is worth something (there is no dividend for example, unlike with stocks). I’d also note, that chatter and news about Bitcoin is totally disproportionate to the actual amount of activity in the sector.


Some level of regulation is also likely ahead. This shouldn’t really be seen as a bad thing for most of us. I’d much rather be involved in a marketplace which was regulated and safer. We’ve already perhaps unsurprisingly seen issues in the market due to lack of regulation (eg. Fortune: Coinbase Is Investigating Possible Bitcoin Cash Insider Trading by Employees). Regulation will also bring a lot more traders to the market. There are many interesting questions to be raised by the increasing uses of cryptocurrencies, notably its impact on monetary policy, which I’ll try to explore in future blogs. I think a lot of the discussion around cryptocurrenices has been far to heavily skewed towards the technical aspects of how they work, but not really their economic impact, which really needs to be explored more broadly. There’s a great article by Robert Carver on Bitcoin, which I implore you to read after this.


MiFID II – again, I had to talk about it


2018 is MiFID II time, which is nearly here. From my point of view the biggest impact will be from unbundling of research in fixed income/FX. Cuemacro is an independent provider of consulting and research for the market, so perhaps my view is skewed. Will MiFID II provide a more level playing field, given that the sell side will no longer be able to give out research for free? Perhaps too early to say. I do think though that overall, there is less likely to be quite as much consumption of research from the buy side, given it will now have to be paid for in fixed income/FX. I do think though that buy side’s budget for research could end up being spent on other external services, whether that is bespoke external research work (call me in that case!), also data to guide research, training, and not exclusively on purely generic research papers. It will also likely focus the minds of all research providers whether in the sell side or independents to elevate the standard of any research they publish.




Best of luck for 2018. Luck is pretty much the most important component for success in financial markets (mixed in with a bit of skill…!)