Feb

26

A VeltmanThis phenomenon cannot be denied, if one simply eye-balls the daily charts over the past year. I hypothesize that as the co-relation has become more-widely observed (and embraced as an automated trading signal), it partly became self-fulfilling and re-enforcing. All simple trading ideas follow their learning curve and eventually suffer from over-crowding; this one appears to be at its pinnacle right now. There are two main propositions to consider:

1. Which precise instruments to choose: SP vs. HG, SP vs. AUD/USD, SP vs. AUD/JPY, Nikkei vs same, etc.

2. Pinpoint the causality, i.e. which leads the other; and, possibly in what time-zones and under what special circumstances.

Empirically, I have long noticed that when North-American activity is somewhat curtailed due to holidays or snow-storms, the Australian Dollar moves that precede North-American time-zone will in fact tip the US stock market direction. A very hard to swallow idea: given the enormous size of the US Equities arena. On the other hand, if one's philosophy is that markets move more on perception than they do on reality - then why not?

It's also very important to incorporate varied leverage into your model. In course of 2010, for example: trend runs in outright percentage terms were greater in currencies than SP; and twice as big in Copper than SP! If you further add the effect of widely used 100:1 Forex trading leverage or the high leverage of Copper futures - results are even more outrageous… This is where Theory of Reflexivity may enter near market crests, and those leveraged bets will cause "fundamental reality" to succumb to speculative forces.

Phil McDonnell writes:

To investigate Mr. Veltman's conjecture it might be helpful to do some counting. Looking at SPY relative to the copper ETF JJC and the Australian dollar ETF FXA showed the following coterminal 105 day correlations:

        SPY JJC FXA

SPY 100 60% 71%
JJC  100 71
FXA               100

Clearly the simultaneous relationships are strong and positive. But to trade we need a predictive relationship. Looking at the same vehicles but lagged 1 day for FXA and JJC we get the following correlations to SPY the next day.

SPY JJC 0% FXA -7%

Both correlations are insignificant. Neither yesterday's move in JJC or FXA is any help in predicting the next day's move in SPY.

Nick White comments:

I echo Dr. McD's analysis. There are far too many players joining the dots between AU index — resources stocks — AU FICC without any real (well, "real" as how most Dailyspec readers would define it) reason to.

Anatoly Veltman adds:

This was yet another in the series of testing accidents, this time from esteemed Dr. McDonnell. Where would anyone get the idea to lag Australian action by a calendar day, before impacting North American S&P? And if you switch the lag around: that's not the causality hypothesized.No surprise here: for any test to be meaningful, considerable resources are required to set up proper test. Usual handicap: quants are not precise in coding trader's idea.

I originally described this idea in connection with the previous Feb.10 snow-storm. Price action could not have been more clear: Australian Dollar strength led the eventual "surprise" intra-day upside resolution in (hesitant) North American S&P, many hours in advance. S&P's (technical) up-trend then lasted for days! But how will this ever be coded… 


Comments

WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '4456' AND comment_approved = '1' ORDER BY comment_date

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search