Friday 17 July 2015

Listening is better than hearing

15:37 Posted by The Thalesians (@thalesians) No comments

We all know the difference between listening and hearing. You can hear what someone says, yet can understand comparatively little. If I attend a lecture on a subject I know nothing about, I might be able to make out the words, yet ask me about the narrative of the presentation, and I'll be absolutely stumped! Furthermore, even if I understand what is being said, my interpretation could differ significantly from other people in the audience. There will be even more ambiguity, if we include body language into the mix.

So just because interpreting language is difficult, should we just ignore language? Hopefully not! Language is an important part of human civilisation and thankfully so! We simply have to live with all the ambiguities in interpreting language and use our judgement to overcome them.

When it comes to markets, numbers are clearly very important. Market prices, the values of economic variables, forecasts etc. they are often quantified with numbers. Traders have market information flowing into them. However, what about when the big news is the language of the news story itself? Traders of course digest and interpret news trying to decipher the details. For example, if we were trying to calculate the total value of the cars above, it might well help to count them. However, unless we know which type of cars they are, an estimation on their total value is likely to be poor.

There are of course many different types of news. There might be news related to economic data releases, there might be news related to politics and so on. Another important type of market news is what central banks say. Often central banks might not change their monetary policy. However, through central bank communications they might hint about future interest rate policy. We have seen this recently most notably from the BoE and also the Fed.

What central bankers say, is important for understanding markets. However, can we read central bank communications in an automated and totally unbiased way? I recently did a project for a US based company called Prattle. They read central bank communications in an automated fashion and come up with unbiased "sentiment" scores based on these. Bullish sentiment from a central bank, can be interpreted as a more hawkish central bank (so more likely to hike rates). Conversely, we'd interpret bearish sentiment as dovish (more likely to cut rates). As well as coming up with an FX trading strategy by looking at the data, I also wanted to see whether the data could be used to answer other questions about central banks.


One question I had in mind, was trying to see if there was any shift in tone by the ECB after Draghi took over. I wasn't so much concerned with individual central bank communications, more their general path. I examined Prattle's ECB index and calculated the average score during the last few months of Trichet's tenure, which covered a period of interest rate rises. I then compared this to Draghi's first few months when he cut rates. Comparing the scores, shows a slightly more dovish outlook from the ECB after Draghi took over! This fits in with market perceptions about Draghi being more dovish than Trichet.

Computers can really help us to listen, rather than simply "hear", when it comes to central bankers! If you are systematically trading the markets but ignore what central bankers say, ignore economic news and only rely on price inputs, there's a question to ask: aren't you missing a factor, that discretionary traders have known for years? Whether or not you agree, isn't it at least worth investigating?

If you'd like to read my full paper on Prattle's central bank sentiment data to systemically trade FX, drop me a message! (An abridged version of the paper can be found here)

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 interesting 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- join our Meetup.com group for more details here (Thalesians calendar below)

22 Jul - London - Paul Bilokon - Stochastic Filtering
07 Sep - Frankfurt - Saeed Amen/Yves Hilpisch/Thomas Wiecki/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
23 Sep - London - Stephen Pulman - Multi-Dimensional Sentiment Analysis

0 comments:

Post a Comment