Studying impact of new economic data on FX prices

Just over two weeks ago I listed several objectives to work on, as I set aside 5 weeks to spend on developing my trading strategies further.  As the hot weather continued in London, I sweated at my desk and made progress on these items in the last fortnight.

To Do List and Espresso Cup

In summary, I completed some decent data analysis but it wasn’t as straight forward as I had expected.

 

In this post I quickly share the second objective worked on.

The first objective was to compare the quality of the historical price data between two of my brokers. See post entitled FX Price Data Comparison: FXCM v TradeStation.

Price Reaction to News

The next objective I jumped into was completing a detailed analysis of price reaction to various news items, across various markets.  I am not sure whether I went into far too much detail here.  Needless to say, I am currently more confused than anything else, as to how to interpret and handle all that data.

Why bother with this?  My reason for doing this analysis is my long-held (personal) belief that release of new economic data introduces a new element of randomness into price, as one cannot know whether the data will be positive or negative, thus in a way nullifying the technical setup that might have occurred beforehand.  Secondly, bid-ask spreads can widen significantly during news announcements, which then work against the trader and reduce the edge that any strategy normally might have.

Using ForexFactory I downloaded a schedule of Tier-1 economic releases from 2011 till the end of last month (close to 10,000 data points) – then downloaded historical 15M data for various markets to identify the price reaction at the time of the announcements.  The ForexFactory data required significant reformatting and it was necessary to accomodate situations where more than one release was being announced at the same time (e.g. Canadian employment and US employment reports are often released simultaneously, or the ECB press conference starting at the same as the US Trade data figures). I also had to merge items together – as an example, the US employment report is often listed in the form of three seperate items – Average Hourly Earnings, Unemployment Rate and Non-Farm Employment Change.

Doing this I was able to assess the typical price reaction to any given news item (and compare that to the normal price volatility at that time of day).  This did not take into account whether the actual number matched or differed from the expected number.  The purpose was to develop an expectation of the average reaction, regardless of the actual figure.

I could then see which news items mattered most for which market.  It was necessary to create a seperate spreadsheet file for each market, with each spreadsheet file quickly snowballing to 70MB.

At this point, I actually still havn’t concluded what rules I will use for the various economic releases.  The rules should tell me when to skip a setup and/or whether to close an existing position if either occur near in time to a scheduled news item.

Here’s an example of what the analysis produces, using AUDUSD:

AUDUSD Price Reaction

All in all, the analysis wasn’t the easiest exercise, and it wasn’t helped by a lack of concentration most likely due to this heatwave we have been experiencing in London.

Recent live trades and workload

Add to that some losing live trades from my coded strategy, and difficulties in following the trades in the Reversals Trading Room with Felix (due to how I have set myself up for it rather than anything else), and it produced a couple of weeks of challenging working conditions.  In the end, my risk management and strong discipline kept the losses to manageable levels.  Some of the risk management decisions were complicated by having setups occur in heavily correlated markets at the same time – on Monday this happened with several Canadian dollar pairs, and on Tuesday with several European equity indices.

From today, I am likely to shift my focus back to another working project, with significantly less time available for trading.  Not sure where that leaves me, as I have only completed 2 of the 5 objectives thus far to progress the strategy development along, and I am cutting the working time allocated to it from 5 weeks to 2.5. Given that for now, I don’t know whether I have got an edge or not, should I continue trading the coded strategy live, simply for the sake of collecting more data?  Trading it doesn’t take much time.  In the last three months, I got around 110 live trades under my belt – this did help in helping me develop the strategy further.

In any case, it seems that the development tasks always seem to take much more time than I anticipate.

Let’s see.

 

This entry was posted in Trend-Following Strategy, Work in Progress. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.