Monday, 14 September 2009

Save the COT Report!

U.S. regulators claim they are improving market transparency with some big changes to the Commitments of Traders report. But the changes threaten to sharply curtain the very transparency they hope to achieve.

Please help save this valuable free data by writing a comment to the U.S. Commodity Futures Trading Commission at cotchanges@cftc.gov. The deadline is Oct. 1. More info on how to do so is at the end of this post.

The COT data comes in a free weekly report issued by the U.S. Commodity Futures Trading Commission. It details trillions of dollars in derivatives positioning in over 100 major markets - everything from crude oil to the Japanese yen, gold, frozen pork bellies, wheat and the S&P 500.

Just Raw Numbers

The data is just a bunch of raw numbers that are difficult to decipher until you look back over it for many years and start seeing patterns. Analysts like Steve Briese and Larry Williams pioneered methods to help decipher the data and increase popular understanding of its usefulness.

Taking off from their work, I crunched years of data and studied the long-term charts in search of a way to trade off it alone. This led to my COTs Timer system, a trading strategy based exclusively on this fascinating data that I offer here for free as a public service.

The strategy has a 72-percent win record since December, when I started to trade the first of a new generation of trading setups. The average has been a 5.4% gain with a 3.2-wk holding period, with a 18.3% real-time cumulative return (adjusted to remove gains due to leverage and adjusted to the maximum portfolio allocation of each setup).

Now, in the wake of the financial mess we've faced for over a year, the CFTC is rightly trying to increase market transparency. That includes making some changes to the COT report to give more details about the data, announced Sept. 2. (See the CFTC's additional explanatory notes here.)

But in so doing, the commission will actually curtail the very transparency it seeks to improve. Unless we speak up.

Historic Data Vital

The CFTC will replace two of the categories of traders in the report with four new categories. The commercial traders will now be broken down into a category for producers, merchants, processors and users, plus a separate category for swap dealers. The non-commercial category (popularly known as large speculators) will be broken down into "managed money" and "other reportables." Presumably, the third category - non-reportables (a.k.a. the small traders) - will remain untouched.

The CFTC is inviting comments on whether it should continue to publish the data for the existing categories alongside the new data. This is a vital issue for anyone who cares about this data. If we no longer get the data in the existing categories, the new numbers could be difficult or impossible to evaluate for years.

The CFTC has said it would publish historic data for the new categories going back three years. That's far from being enough for most purposes of statistical research. At a bare minimum, my calculations suggest we need at least about 10 years of data - and likely much more; as much as possible, in fact. Also, the data must cover as many diverse market conditions as possible for the data to have any statistical meaning.

In fact, the existing COT data, which goes back to 1995 for most major markets, just became useful for statistical purposes in the last few years when there were finally enough data points to research.

Comparing Apples With Oranges

The CFTC may argue that a researcher could just recreate the data for the old categories by adding together the positions of the new categories. But this is cumbersome and creates significant programming difficulties because the new data is reported in a different format.

It also raises the question of whether the new categories will exactly replicate the existing ones. For example, if we add up the positions for the new "managed money" and "other reportables" categories, will we get the same figures as for the old non-commercial category? Will the small trader category remain exactly the same? Or will we be comparing apples with oranges?

Do the changes involve any other kind of redefinition of categories or reassignment of traders? Could the new categories make such changes more likely because the old data will no longer be seen as relevant? All these things would reduce or eliminate the ability to compare the old and new data.

Email the CFTC

If you think the CFTC should continue to publish the existing categories, email the CFTC at cotchanges@cftc.gov. Please write before the deadline for comments on October 1. (Mention that you are commenting on the CFTC's proposal of Sept. 2 to amend the COT report.) If you're pressed for time or not sure what to say, you can even just email them a link to this post: Save the COT Report!

Can we make a difference? Yes! The last time the CFTC reviewed the COT report, it received an unprecedented 4,659 comments from 23 countries. It was by far the most in the agency's history in response to such a notice. The response was unanimous in support of the reports. Remarks included: "Please, please do not discontinue this very valuable report," "Don't you dare," "Save the COT" and "Leave it alone you knuckleheads." The CFTC backed down from a proposal to discontinue them. If we remind the agency of the importance of this valuable data, the cause of transparency has an excellent chance of prevailing again.

19 comments:

Jez Liberty said...

Alex,
I do not understand why this change is bad for anybody except maybe people like you..

It seems like they are adding more information to the reports - which can not be a bad thing!

I dont believe that COT reports are primarily published for historical researchers and it seems like they will indeed offer more transparency and as you mention, one should be able to keep aggregating the 2 sub-categories together to come back to the historical categories.

Or am I missing something?

Alex Roslin said...

Hi Jez,

Thanks for your message. Adding more data is certainly a good thing. That's not the problem.

As I explained above, aggregating the new categories is cumbersome and creates an obstacle to research.

For example, it makes it much harder to integrate the two datasets when analyzing it through any kind of application - which is already difficult enough without this added headache. The new data is presented in a different format, adding to the difficulties.

Already, there are few enough applications out there - even in proprietary hands - that can analyze this data. The proposed change will make the data even more inaccessible.

The data's already-existing inaccessibility is why so few people have actually done any solid research showing how the data impacts or is related to market prices - which is after all one of the key points of compiling and releasing the data.

As well, there is the point that the new categories may not map completely onto the old categories. And even if they do map perfectly right now, without the old data categories being reported there will be less reason in future for the CFTC to preserve the continuity of the datasets.

You are right - the data isn't being published primarily for historical researchers. This is exactly my point. If the data's continuity isn't preserved, it will be much harder to make use of all the data until now for anything except historical research.

Regards,
Alex

Alex Roslin said...

Hi again Jez,

Another point: The CFTC proposes to supply historical data for the new categories going back just three years. That, too, has virtually no usefulness either for historical research purposes or for trading today.

Data needs to go back much further than three years when it comes only once a week for any meaningful comparisons or statistical purposes.

This isn't a question of "historical research." Data back to 1995 - when the existing data starts for most markets - isn't "historical." That's "current" data.

To suggest otherwise feeds into the popular misconception that you can reliably identify patterns in the COT data by simply eyeballing a chart going back two or three years. The fact is there isn't a heck of a lot of market information you can get from such an exercize that has any statistical validity or usefulness for trading.

Regards,
Alex

Maumaj said...

Just a brief note to let you know I've sent an email to cotchanges@cftc.go. Thanks for having notified us of the situation.
Maurice Major

Maumaj said...

Just a brief note to let you know I've sent an email to cotchanges@cftc.go. Thanks for having notified us of the situation.
Maurice Major

Alex Roslin said...

Thanks, Maurice!

Jez Liberty said...

Hi Alex,

I get your point I think but I still think that you are trying to oppose to a "greater good" improvement of the COT reports.
Without any consideration for the historical data I believe it is better to have more information/further breakdown on the COT reports. And in my opinion it does improve the transparency of the data published.

Regarding the historical data and assuming that the new sub-categories will still aggregate to the same current categories (I think this is a reasonable assumption) - if I were in your place I would just "ignore" the new changes/categories by recalculating the aggregated current values. I agree this will be more work for you and everybody else using historical data (sorry - I disagree on your definition: to me, data from the past belongs to history, hence historical data. Current data would be data from the current week. However I agree that you need more than 2-3 years to draw any sort of significant statistical analysis).
But this is a technical constraint more than anything else (ie you will have to get the new file formats, pre-process them to get back to the current format and finally apply you current existing process). In the mean time you could start building historical data on the sub-categories in order to be able to switch to using them once the amount of historical data is substantial enough to draw significant statistical analysis.

I am all against "smoke and mirrors" moves from the government and their aganecies to disguise / present the truth the way they want you to hear it; but in that instance I do not believe this is the case.

For the lack of a better metaphor, it is as if you were complaining that weather forecasts will now be made per hour instead of per half-day (assuming they could be correct ;-) Or that your electircity bill will be improved by breaking down how much you use on lights, heating, washing machine and others. I cant see any downside to these.


On a more general note regarding the COT reports (I do enjoy reading about them every week as they add another dimension to other analyses - I hope you are not perceiving my comments as a rant..) I was wondering if you thought about the rise of OTC positions over the last few years. To my mind their increasing use might (this is a fact that is often discussed on goldbug forums when trying to decipher the COT data: i.e. OTC data is not reported and would potentially skew/alter substantially what the overall positions would be compared to how they are reported in the COT reports).
Do you think that could have an impact and make the relevance of COT data less important?

Alex Roslin said...

Hi Jez,

Thanks again for your message. I think you misunderstood a little. I have no problem with the new data. Yes, it's welcome. It won't be of much use to anyone for many years, but the more data we've got the more transparent things will be. I tried to make that clear above, so sorry if it wasn't.

The point is until we've got enough of that new data - maybe seven years from now or so - the existing data needs to be published too.

It is almost completely useless to focus on the current week's data without knowing where things stand in relation to past data and without doing a whole bunch of research to come up with the right prism through which to understand it.

What I'm doing now is the result of years of studying this data and trying to make sense of it. The fact that there are so few people who've done so is testament to how hard it is to figure out.

If there's any point to publishing it at all, the CFTC shouldn't make it even harder to figure the data out. Sure, some people will figure out work-arounds to integrate both datasets - if they are indeed identical and remain so, which isn't a certainty - but the point is it's yet a major hurdle to the average person trying to figure the data out.

It will put the data until this date out of the vast majority of the general public's reach and thus make it largely inaccessible. Most charts will start with the new data.

I didn't start with any stats knowledge. I just kept examining the long-term charts of the COT data and comparing them with prices until years later I started coming up with some ideas of my own.

I think a lot less people will bother when they've only got a few years of the new data to study and when it won't be clear to most how the new and old data match up.

That's going to make all of it a lot less accessible and hurt the cause of transparency.

Regards,
Alex

Alex Roslin said...

Hi Jez,

Regarding the OTC market, I just haven't looked at any data closely enough to see what possible impacts there could be. In gold, total open interest has increased steadily since 1995 in the large spec and commercial categories (though not for the small traders) - so it's not the case that there's less volume in the COT data. And the signals remain robust through the timeframe, so I don't see any significant impacts at this point on the COT data's usefulness.

Regards,
Alex

Joseph said...

Hi Alex, I as well sent in a email. Take care

man said...

Alex,

I and several of my friends that view your great site regularly have sent e mails.

Alex Roslin said...

That's great. Thanks!

Regards,
Alex

maximus said...

email sent.....

thank you for the outstanding blog and updates....

Maximus
http://4best4worst.wordpress.com/

Alex Roslin said...

Thanks very much, Maximus.

Anonymous said...

Let's get rid of the COT.

Who needs it. Doesn't help anyway.

jj

jesuscheung said...

email sent!

etfgateway said...

Hey Alex, I sent an email asking them to preserve the original way of posting the data and just to add the new breakdowns to what they are already doing so that the data integrity is preserved.
- Kevin

Anonymous said...

If they're going to change things, how about taking the Friday close data and then making it available over the weekend. So its more timely, fresh; this can easily be done with our modern Info Tech. That 3 day lag does matter, esp on major expiration weeks.

Alex Roslin said...

Hi Anonymous,

I agree - that would be an improvement. But again, they'd need to give us at least 10 years of data before they discontinue what they're offering now.

Regards,
Alex