Saturday 19 September 2015

Predicting the past

16:11 Posted by The Thalesians (@thalesians) No comments

I can predict the future. Although, if that were even remotely true, I probably wouldn't be sitting here writing a blog. Instead, I'd be lounging about on a beach somewhere. However, despite it being a totally inaccurate statement, I hope that it has encouraged you, the reader, to take a few moments to finish reading this article.

The problem with the future is that no one really knows what will happen. I remember when I was a kid. The notion of a phone having a computer in my pocket would have seemed laughable. The idea that computers would be so ubiquitous would simply be beyond my comprehension. Indeed, a similar point about the role of technology in trading was made by my good friend Pierre Lequeux speaking on a panel at the recent TradeTech FX. The difficulty with trying to predict the future doesn't prevent us from trying. Over the past weeks and months, the market has been musing about when the Fed could begin a hiking cycle, it's first in over several decades (ok, not quite, but it has been a long time since the Fed actually started a hiking cycle). What has complicated matters, is that even if the Fed does hike, what will that mean for markets?

A systematic trader does not attempt to predict the future (to an extent). I find it incredibly difficult to forecast when the systematic trading strategy that I've been running will make or lose money from day to day. This is despite the fact that the overall trajectory of returns has been upwards since I started trading my own cash just over 2 years ago. The objective is more about creating a strategy which is profitable over the long term on average.

Systematic traders rely upon patterns from the past persisting into the future, which on average have been profitable and have a good rational explanation. We can also look at the past to try to understand what happened to our trading strategy in similar situations, which might crop up in the future. The great thing about having a systematic trading strategy is that whilst we can't tell the future with certainty, we definitely know the past! I've been in the market for around a decade, and during that time I've run all manner of different systematic strategies. I can create a new trading system and then use historical data to understand how a strategy coped with the start of Fed hikes as an example, well before I even started working in currency markets.

Hence, even though I can't tell you precisely when a Fed hike will be, I can tell you how confident (or not) I would be with a trading strategy to cope with similar historical events. A discretionary trader can look back at their own P&L history to see how they coped with historical events, but unless they have been trading for a long time, they might be a bit stuck when trying to identify events before they started trading. I recently published a Thalesians paper, for example, discussing how CTA strategies have performed at the start of Fed hikes to test this idea.

Of course, there are caveats. The past is never precisely like the future. There will be unusual events that are not in our historical sample, which could impact our portfolio significantly. Indeed, this Fed hiking cycle could be very different to previous ones as an example (the Fed has never begun hikes from such a low level). At the same time, there are likely to be many historical events similar to future ones.

Whilst systematic trading isn't for everyone (just as discretionary trading isn't), the ability to look at historical data, gives us an ability to see how robust our portfolio has been to historical shocks and to check how our strategy has behaved. Even with all the mathematics that goes on behind creating a systematic trading strategy, perhaps the most important thing, is to have a modicum of common sense and understanding of risk!

If you're on the US East Coast, I'll be in Washington DC 27 Sep, NYC 29 Sep-3 Oct and Boston 5 Oct if you'd like to meet me and hear more about systematic trading! If you're in mainland Europe, I'll be in Paris 6-9 Oct at the WBS Fixed Income conference, where I'll be hosting a systematic trading workshop.

Like my writing? Have a look at my book Trading Thalesians - What the ancient world can teach us about trading today is on Palgrave Macmillan. You can order the book on Amazon. Drop me a message if you're interested in me writing something for you or creating a systematic trading strategy for you! Please also come to our regular finance talks in London, New York, Budapest, Prague, Frankfurt, Zurich & San Francisco - join our Meetup.com group for more details here (Thalesians calendar below)

21 Sep - New York - Agostino Capponi - Arbitrage-Free Pricing of XVA (Thalesians/IAQF)
23 Sep - London - Stephen Pulman - Multi-Dimensional Sentiment Analysis
01 Oct - New York - Saeed Amen - How to build a CTA? / interactive Python demo
05 Oct - Boston - Saeed Amen - Trading Thalesians Book Talk / PyThalesians Python Interactive Demo (Boston Algorithmic Trading Meetup Group)
09 Oct - Budapest - Taylor Spears - On the Sociology of CVA
14 Oct - New York - Dan Pirjol - Can one price Eurodollar futures in Black-Derman-Toy? (Thalesians/IAQF)
21 Oct - London - Robert Carver - Lessons from systematic trading

Saturday 12 September 2015

Burger parity

16:18 Posted by The Thalesians (@thalesians) No comments

I recently visited Frankfurt and Zurich to help launch Thalesians quant finance talks there (a special thanks goes to Jochen Papenbrock, Adrian Zymolka and Swati Mittal for organising the events there). Travelling from Frankfurt to Zurich, I noticed several major differences.

One difference is the language. I can vaguely understand Hochdeutsch, as it is spoken in Frankfurt and the rest of Germany, owing to a few years studying German, during a period of my life when I could be called young (and importantly, my mind has not yet quite expunged all the German I learnt then). By contrast, my ears have some trouble understanding Schweizerdeutsch, the dialect of German spoken in Switzerland, simply because I am not used to it.

Another difference is the geography. Frankfurt to my untrained eye is relatively flat, making it pretty easy to get around on foot. In Zurich, the hilly terrain makes it somewhat more challenging to navigate to certain neighbourhood, such as the area around ETH, where I gave my Thalesians talk and where I had the opportunity to meet Nassim Taleb, through a combination of randomness and Twitter!

However, the difference I'd like to focus on is price. Perhaps, it is not an understatement, to note that Zurich and Switzerland in general feels expensive for visitors. One of the easiest ways to test this is to use the Big Mac index, maintained by the Economist, which gives you the relative cost of a Big Mac in a number of different countries. The idea is that you can work out how overvalued (or undervalued) a certain currency by assessing the price of similar goods in different countries. In practice, you would aim to choose a basket of goods, rather than a single item, like a Big Mac in such an index, which is referred to as a Purchasing Parity Power index,. As an aside, how much do this the burger costs at the top, which I had in Zurich (clue: it was rather more expensive than I would have liked...)? Using the Big Mac index, we note that in Switzerland, a Big Mac costs 6.82 USD. By contrast in the UK, it is 4.51 USD and in the Eurozone it is 4.05 USD.

We are now going to use this information to work on a little problem. Let's say, we had 5 USD and we wanted to share out Big Macs in Zurich, London and Frankfurt, as prizes. Whilst, in London and Frankfurt, we could give people a whole Big Mac and still have change. However, in Zurich, we would end up giving out less a full Big Mac. This would be kind of unfair, a prize in Zurich would be far less.

Yet, this is exactly what people can do when they invest, based on notional amounts of different assets. However, the "risk" that such an approach buys is unequal. A dollar worth of bonds is less risky than a dollar of equities. If you have a fund which runs a multi-asset strategy, trading all manner of different assets, if they scale purely by notional, rather than risk, the portfolio could have considerably different exposures to those intended.

There has been much in the press recently, talking about the recent performance of risk parity, which basically attempts to allocate in this fashion based on risk, rather than purely notional. Of course any strategy is not going to outperform all the time. What has been lost in the argument is why risk parity is used, namely that assets do not all have the same risk profiles. For example for trend following funds, if they had no method to adjust for the risk differences in their portfolios, they would end up being dominated by the highest volatility assets, like crude oil, whilst having little exposure to lower risk assets like bonds.

So yes, strategies might not always outperform all the time. But it is important to consider, what the alternatives are. In the meantime, parity is just as important for burgers as it is for risk...

If you're on the US East Coast, I'll be in Washington DC 27 Sep, NYC 29 Sep-3 Oct and Boston 5 Oct if you'd like to meet me and hear more about systematic trading! If you're in mainland Europe, I'll be in Paris 6-9 Oct at the WBS Fixed Income conference, where I'll be hosting a systematic trading workshop.

Like my writing? Have a look at my book Trading Thalesians - What the ancient world can teach us about trading today is on Palgrave Macmillan. You can order the book on Amazon. Drop me a message if you're interested in me writing something for you or creating a systematic trading strategy for you! Please also come to our regular finance talks in London, New York, Budapest, Prague, Frankfurt, Zurich & San Francisco - join our Meetup.com group for more details here (Thalesians calendar below)

21 Sep - New York - Agostino Capponi - Arbitrage-Free Pricing of XVA (Thalesians/IAQF)
23 Sep - London - Stephen Pulman - Multi-Dimensional Sentiment Analysis
01 Oct - New York - Saeed Amen - How to build a CTA? / interactive Python demo
05 Oct - Boston - Saeed Amen - Trading Thalesians Book Talk / PyThalesians Python Interactive Demo (Boston Algorithmic Trading Meetup Group)
09 Oct - Budapest - Taylor Spears - On the Sociology of CVA
14 Oct - New York - Dan Pirjol - Can one price Eurodollar futures in Black-Derman-Toy? (Thalesians/IAQF)
21 Oct - London - Robert Carver - Lessons from systematic trading

Tuesday 1 September 2015

Fed (mountain) hikes

14:54 Posted by The Thalesians (@thalesians) No comments


The market has been awash with speculation about when the Fed might hike, continually watching for clues from Fed communications. The recent symposium of central bankers at Jackson Hole, organised by the Kansas Fed, presented the market with another opportunity to search for clues. Comments at Jackson Hole struck a hawkish tone. Fed vice chair Fischer suggested that inflation did not need to pick up for a Fed hike to happen. If we consider the start of previous Fed hiking cycles, whilst growth and employment levels might have been similar, for the most part, inflation was a lot higher than it is at present. For a wrap up of what happened at Jackson Hole, I'd recommend have a look at Sam Ro's write up at Business Insider. If you're interested in understanding central bank communications from a systematic viewpoint I'd have a look at my recent Thalesians quant paper on the topic for Prattle, where I show how patterns in central bank sentiment gauged from communications can be used to trade FX.

Whilst the market has been obsessing about when the Fed is likely to hike, perhaps a more pertinent question is what the market will do, when a hike actually happens? A recent tweet by Emanuel Derman got me thinking precisely about this point. (Incidentally if you don't follow him on Twitter you should do!). Derman suggested that:

Raising rates 25 bp is actually going to be a massive non-event. There will be a brief positive reaction by markets, then back to business - Emanuel Derman @EmanuelDerman

By the time the Fed actually does hike, the market will have had such a long lead time, that a considerable amount of market adjustment to the hike will have already occurred. Let's consider the dollar rally from July 2014 till just before January's ECB meeting. The Fed had not hiked during any of this period, nor had the ECB actually announced full blown QE. However, the market was primed to expect both these scenarios. The market was essentially adjusting to expectations, rather than significant changes in monetary policy.

Whilst the market were busy discussing Jackson Hole over the past few days, I was busy hiking in the Alps. The process of the mountain hike can very much be seen similar to the way the market is approaching the Fed hike. We're currently in that phase, getting closer and closer to the top. When the Fed finally does hike, it'll be like getting close to the mountain peak. Most of the hard work of climbing will be done by then, and the market will be moving on to other issues.

If you're on the US East Coast, I'll be in Washington DC 27 Sep, NYC 29 Sep-3 Oct and Boston 5 Oct if you'd like to meet me and hear more about systematic trading! If you're in mainland Europe, I'll be in Frankfurt 7 Sep, Zurich 8 Sep and Paris 6-9 Oct.

Like my writing? Have a look at my book Trading Thalesians - What the ancient world can teach us about trading today is on Palgrave Macmillan. You can order the book on Amazon. Drop me a message if you're interested in me writing something for you or creating a systematic trading strategy for you! Please also come to our regular finance talks in London, New York, Budapest, Prague, Frankfurt, Zurich & San Francisco - join our Meetup.com group for more details here (Thalesians calendar below)

07 Sep - Frankfurt - Saeed Amen/Jochen Papenbrock/Miguel Vaz/Adrian Zymolka - Quant Evening (Thalesians/Quant Finance Group Germany)
08 Sep - Zurich - Saeed Amen - How to build a CTA? / interactive Python demo
10 Sep - San Francisco - Steven Pav - Portfolio Inference and Portfolio Overfit
21 Sep - New York - Agostino Capponi - Arbitrage-Free Pricing of XVA (Thalesians/IAQF)
23 Sep - London - Stephen Pulman - Multi-Dimensional Sentiment Analysis
01 Oct - New York - Saeed Amen - How to build a CTA? / interactive Python demo
14 Oct - New York - Dan Pirjol - Can one price Eurodollar futures in Black-Derman-Toy? (Thalesians/IAQF)
21 Oct - London - Robert Carver - Lessons from systematic trading