Finding spread trading opportunities
- Thread starter Gambit
- Start date
Hi Guys,
last years i studied very much the opportunities of spread trading on the commodities markets. I spoke about my researces there.
I would like to share publically a powerpoint file but i don't know if it's allowed so for these of you who are interested i'm available to send it privately. It's in italian, but if it's needed i'll prepare it in english (ehm...my english
) with pleasure.
I'm relatively new here and i don't want to give you a bad impression. I'm just an indipendent trader and a researcher, i don't sell anything.
last years i studied very much the opportunities of spread trading on the commodities markets. I spoke about my researces there.
I would like to share publically a powerpoint file but i don't know if it's allowed so for these of you who are interested i'm available to send it privately. It's in italian, but if it's needed i'll prepare it in english (ehm...my english
) with pleasure. I'm relatively new here and i don't want to give you a bad impression. I'm just an indipendent trader and a researcher, i don't sell anything.
First, this is a great thread that so far has stayed on topic and been active. Kudos Gambit.
Secondly always hear the oft repeated saying that "spreads trend better than outrights".
Is there any evidence to back that up? I have traded spreads and outrights, and cannot tell any difference in "trendiness" between the two.
Secondly always hear the oft repeated saying that "spreads trend better than outrights".
Is there any evidence to back that up? I have traded spreads and outrights, and cannot tell any difference in "trendiness" between the two.
Thanks clacy. Your comment made me think about the concept of trending time series. I cannot say I have evidence of trendiness or lack of it. I don't know if it sone of those old myths passed that get passed around or not like "the trend is your friend" or "losers average losers."
What I did find was a statistical measure which can quantify the persistence of different time series.
http://www.analytics-magazine.org/j...-hurst-exponent-predictability-of-time-series
I've heard some other posters on ET talk about the Hurst exponent. I'll quote from the article:
"The Hurst exponent is a useful statistical method for inferring the properties of a time series without making assumptions about stationarity. It is most useful when used in conjunction with other techniques, and has been applied in a wide range of industries. For example the Hurst exponent is paired with technical indicators to make decisions about trading securities in financial markets; and it is used extensively in the healthcare industry, where it is paired with machine-learning techniques to monitor EEG signals. The Hurst exponent can even be applied in ecology, where it is used to model populations."
Mr. Mansukani has included some useful citations in his article. It is definitely worth a read.
Here are some links for further research:
http://www.bearcave.com/misl/misl_tech/wavelets/hurst/index.html
http://www.r-bloggers.com/exploring-the-market-with-hurst/
http://www.bearcave.com/misl/misl_tech/wavelets/hurst/index.html
From a cursory glance, it seems that the Hurst exponent is no magic bullet. I found at least 2 or 3 implementations of hurst so if one can download time series data into an analysis package, the test can be implemented. However, there are doubts about its accuracy and predictive power. The test itself is an approximation so there is some variance associated with the exponent.
I myself would use it on a securities that I know have some fundamental link (ex: RDS A vs RDS B). Perhaps Hurst is best used as a ranking tool. One could trade the securities that show the highest and lowest Hurst as well as checking for the Hurst exponent over time.
Since most of us are not mathematicians or quants, I would advise traders (myself included) to backtest all of these measures. Cointegration seems to be generally accepted as tool. Correlation is widely accepted. I'm not so sure about Hurst.
If any quants would like to give their professional opinion, it would be appreciated.
What I did find was a statistical measure which can quantify the persistence of different time series.
http://www.analytics-magazine.org/j...-hurst-exponent-predictability-of-time-series
I've heard some other posters on ET talk about the Hurst exponent. I'll quote from the article:
"The Hurst exponent is a useful statistical method for inferring the properties of a time series without making assumptions about stationarity. It is most useful when used in conjunction with other techniques, and has been applied in a wide range of industries. For example the Hurst exponent is paired with technical indicators to make decisions about trading securities in financial markets; and it is used extensively in the healthcare industry, where it is paired with machine-learning techniques to monitor EEG signals. The Hurst exponent can even be applied in ecology, where it is used to model populations."
Mr. Mansukani has included some useful citations in his article. It is definitely worth a read.
Here are some links for further research:
http://www.bearcave.com/misl/misl_tech/wavelets/hurst/index.html
http://www.r-bloggers.com/exploring-the-market-with-hurst/
http://www.bearcave.com/misl/misl_tech/wavelets/hurst/index.html
From a cursory glance, it seems that the Hurst exponent is no magic bullet. I found at least 2 or 3 implementations of hurst so if one can download time series data into an analysis package, the test can be implemented. However, there are doubts about its accuracy and predictive power. The test itself is an approximation so there is some variance associated with the exponent.
I myself would use it on a securities that I know have some fundamental link (ex: RDS A vs RDS B). Perhaps Hurst is best used as a ranking tool. One could trade the securities that show the highest and lowest Hurst as well as checking for the Hurst exponent over time.
Since most of us are not mathematicians or quants, I would advise traders (myself included) to backtest all of these measures. Cointegration seems to be generally accepted as tool. Correlation is widely accepted. I'm not so sure about Hurst.
If any quants would like to give their professional opinion, it would be appreciated.
Last edited:
I hope to don't have made too many mistakes 
Being long (or short) two consecutive butterflies is called a condor.
Example: Consider 2 butterflies (+1H -2M +1U and +1M -2U +1Z). Going LONG both butterflies results in being LONG the following CONDOR: +1H -1M -1U +1Z.
A spread of one butterfly against another butterfly is called a box (or double butterfly).
Example: Consider the same 2 butterflies (+1H -2M +1U and +1M -2U +1Z). Going LONG the first (HMU) fly and SHORT the second (MUZ) fly results in being LONG the following BOX: +1H -3M +3U -1Z.
Just to be complete: There are different ways to put on / unwind a condor or a box, and using the butterflies isn't necessary the most efficient. For example consider going long a condor by buying the HM spread and selling the UZ spread.
Similarly, you can enter into (or unwind) a box by buying the HZ spread and selling 3 MU spreads.
That said, these positions remain expensive to trade, and are generally not used by retail traders.
Hope this helps.
In my researches i discovered that trends of the spreads are longer and more acknowledgeable. In the pdf file i shared this difference is showed at the page 13 (weekly trends of Corn and Wheat appear to be very erratic) and at page 14 (weekly trends of the spread, in yellow the uptrends in red the downtrends).