What does it look like when a central bank pimp slaps a market?

H/T to Rooster360 for beginning this conversation…

We have long been keen watchers of the complex relationships between Asian economies and the US Treasury curve. Early readers of this blog will remember more than a few rants about the Yen Carry Trade. In the past we have been vocal about the evolving correlation between the Yen and the 30 Year Bond. For most of this year this relationship has been firmly positive when measured from a price correlation standpoint.

Correlation is important because the volatility (or stability) of correlation represents the health of a market. Individual prices may go up, prices may go down, but if correlation stays stable then everything makes sense. Dealers can hedge their risk more easily when they can be assured that their hedges will actually perform as advertised. When correlation itself becomes volatile, liquidity decreases: market makers and dealers cannot easily absorb their normal position sizes, marginally increasing the chances of a spike in overall price volatility.

That is why we watch discrete changes in correlation structure VERY CLOSELY. The historically stable relationship between the Yen and US 30 Year Bond underwent a seismic shift in the days leading up to May 8th when the BOJ released their latest minutes from their April 2015 meeting on monetary policy:

rolling_corr_blue

Keeping true to the classic theme of Central Bank intervention, the real move happened before the release. We are not trying to make a moral statement on the rightness/wrongness of CB’s releasing info to select parties to “prime” or “prepare” the market’s for intervention (although we certainly wish we were on that list, FOMC are you listening? f$%k Hilsenrath, quants move markets more than ever).

Rather, we are merely pointing to the “smoking gun” from statistical standpoint. The chart above is looking at rolling 10 day correlations, meaning each datapoint mainly reflects information about the past, rather than the current day’s price action. We can clearly see the correlation start to collapse well before the release of minutes on May 8th, meaning the daily price correlation was so wildly different from previous day’s that it harpooned, in real-time, a backward-looking indicator like rolling correlation. In other words, the market knew some craziness was about to happen and traders were battening down the hatches on risk exposure into the meeting.

What is interesting to note is that the long term fundamental equilibrium between Bonds and the Yen remained in tact despite the breakdown in correlation. In the weeks following the meeting, traders slapped the Yen back down into line with the Treasury Bond. This highlighted the important difference between correlation and cointegration. Correlation measures the strength of the linear relationship between co-ocurring datapoints. Cointegration refers to a long-term equilibrium relationship between two potentially nonstationary random variables. If the two variables deviate from this equilibrium, the pair’s error correction model will expose each variable to forces that bring the relationship back in line. In this case we can experience situations in which the two variables actually move in opposite directions each day, producing a negative correlation while maintaining a tradable pair in the sense of statistical arbitrage convergence trade.

For example, note how the Yen and Bond converge despite a collapse in daily correlation:

price_chart

Oftentimes, discrete changes in correlation structure represent periods replete with trading opportunity. Having been both market makers and market takers, we can safely say that the only time takers reliably make money is when makers are getting annihilated. Shocks to correlation are symptoms of stress within the market microstructure. Dealers and market makers utilize statistical relationships to offload the risk in their accumulated inventory, and in general they are highly exposed to rises/falls in correlation levels. We will continue to watch developments in the bond closely, as we have said in the past, the bubble to watch is not the stock market

Want to learn how to mine social data sources like Google Trends, StockTwits, Twitter, and Estimize? Make sure to download our book Intro to Social Data for Traders


Enter your email address to follow this blog and receive notifications of new posts by email.


follow_st

Lead image licensed under CC BY-SA  2.0 from WEF

1 reply »

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s