Friday, 23 November 2007

Holiday Delay

This week's Commitments of Traders report will be released Monday instead of this afternoon due to Thanksgiving. Hope American readers had a good turkey, and best wishes for a relaxing weekend. Still snowing up here in Canada. By the looks of it, it'll be not only a white Christmas, but also a white June.

17 comments:

Dave said...

Hi Alex,
Just a general question:

Do you have a S&P summary setup where it lists your trades, their dates, profit/loss, etc?

I remeber seeing something like this for the RUT or Gold, but can't find it. I'm doing some testing, and would like to test directly against your results.

Thanks!
Dave
PS: I just sent you an email if you want to reply directly to it instead.

Alex Roslin said...

Hi Dave,

Thanks for writing. I don't have that posted for the SP500. The gold and composite equity tables are linked on the "Latest Signals" page. For RUT, I created a chart linked on that page, but no table with the exact numbers, although that table did appear in my Futures & Options Trader story a couple of months back.

But you can also download my SP500 spreadsheet (see the link in the Navigation bar) and see all the trade dates that way. Remember: when bringing in the SP500 prices, be sure to use the weekly open price. (Some readers who've tried to test the numbers didn't do so and came up with differing results.)

Regards,
Alex

Dave said...

Thanks for the reply.

Cheers!
Dave

Dave said...

Hi Alex,
Sorry to be such a pain.

I setup a spreadsheet of all the S&P trades based upon your S&P setup.

I am not seeing the same results you are (I'm guessing I'm doing something wrong).

I sent the spreadsheet to your aroslin1@yahoo.ca account (the email from this blog).

In the spreadsheet I included notes about how/where I got my data, along with the calculations used (see the individual cells).

Sorry to be such a pain. But, what I'm trying to do, is match your results so I have a baseline. Once I better understand your results, I want to do some more testing to see if I can come up with other setups. And, thus, compare apples-to-apples. And make sure there aren't any glaring errors in my calcuations or assumptions.

Thanks a bunch!
Dave

Anonymous said...

Hi Alex,

I have done up a rough & dirty mind mapping of your set-up for Gold, S&P and NASdaq. If it helps your readers(like dave), perhaps I can post a link here.
That is, if you are agreeable.

Regards
Fred

Richard said...

Thanks for all of your research which you publish free of charge, I look forward to reading your next few reports.

The ETF TLT is a tradeable measure of the long term U.S. Treasury bond; and may be moving higher between now and the next Federal Reserve Meeting on December 11th; but, a market top may arrive before then.

It may be that the traders are going to be 'over committed' at the market top, and give a delayed signal to go short the bond market since:
1) they can stay in the market longer than most having seen long ago that the Fed was on a continuing path to lower interest rates.
2) they are hoping that their position in Treasuries will remain up at least until end of the year reporting.

The Stockcharts.com chart of TLT shows Friday November 23rd's trade at 93.73: a fall below 92.53 would for me be a sell signal: in other words sell the ETF TLT.

The chart shows two "so called" flights to safety out of the stock market: one on October 19th and another on November 2nd; here investors jumped the sinking ship, the Russell 2000 and took shelter in the "Treasury Bond Lifeboat".

The chart of TLT relative to VTI looks like a top has formed
1) there was a hanging man candlestick on November 12th, and
2) there was a dragonfly doji on November 21, and
3) while it moved higher, it did so on lower volume.

Other evidence that long term US Government Treasury bonds are peaking comes from the topping out of the bond ETF relative to its bond counterpart in the futures market TLT relative to $USB; notice the recent lollipop hangingman candlestick portending a market reversal is at hand; this chart shows that the players in the futures market have been getting a strong and free ride up; but, the most recent candlestick is down; this I am sure is stirring some concern for the wealthy speculators.

And the Yahoo Finance chart of Rydex's bull bond mutual fund shows oversubscribed relative to its bond counterpart -- RYCGX relative to TLT appears overbought.

The coming fall in U.S. Treasury Bond prices will be spectacular: it will have huge gaps down for a number of reasons:
1) that which goes up parabolically, comes down parabolically.
2) many bond fund managers, hedge fund managers and even speculators have bought interest rate derivatives, also known as interest rate swaps to supposedly protect their assets; but the reserves and deposits on these is practically non-existant.

Interest rate derivatives have created a bond market financial house-of-playing cards; it's as Warren Buffet warned in a BBC article that derivatives are an investment time bomb.

The coming collapse in Treasury Bonds is going to make the Citicorp-CDO subprime debacle look like child's play.

The rapid loss of value in bonds will be like that of Citicorp, with weekly gaps lower as news comes about the under-funded derivatives.

I recommend that one be long the gold ETF GLD; and use a small margin account to 'dollar cost average' in a very small amount of TLT shorts.

Alex Roslin said...

Thanks, Fred. Be my guest.

Regards,
Alex

Alex Roslin said...

Hi Dave,

I'll take a look and get back to you.

Take care,
Alex

Dave said...

Hi Alex,
Thanks a bunch for looking at my S&P trades!

If you have any questions, or if you are looking at my setup saying "wtf was he thinking. ;-)", don't be afraid to point out the obvious.

I'd rather look like an idiot for 5min, and get something corrected.

Fred -- I'd be very interested in seeing your setups. (Thanks Alex!) If you want to email them, feel free to send them to seionage[at]mailcity.com .

Right now, I'm trying to read and learn all I can about the COT. I'm amazed at the various opinions/articles there are on these reports.

Cheers!
Dave

Fred said...

Hi Alex,

Here goes:
http://rough-notes.blogspot.com/

Perhaps your reader will find it helpful as a mind-map to your COTS system.

Cheers!
Fred

Alex Roslin said...

Hi Dave,

I've just taken a closer look at your data series, and all matches up with mine - except for three trades that explain the differences between our results. (I've already written back Dave personally, but I wanted to post a version here too in case it might help other readers with similar questions.)

Firstly, the two trades that have different results are due to the trades that start and stop in Dec. 2001. Due to holidays, I've found you frequently have to adjust the COTs data when comparing to price data. In this case, due to the Christmas holiday, the COTs report that should have been dated Dec. 25, 2001, was in fact dated the following Friday, Dec. 28, 2001. Then there was no COTs report dated the following week at all, this time due to New Year's.

I assume the Dec. 28, 2001, COTs report would not have come out without a three- or four-day delay, especially because of New Year's, so my testing program executed that trade with the 1172.51 price for the week of Jan. 7, 2002 (the next weekly open). I don't think that COTs report would have been available in time for execution on the open of Dec. 31, 2001, which is the date for the trade in your table. Thus, you and I would have slightly different results for the previous and ensuing trades.

The third discrepancy results from the fact that you don't count the very first trade. (Dave didn't count the first trade because it starts on the first date in the column, and because this is the beginning of the data, it can't be known whether this is merely a continuation of a previous signal or not.)

While this reasoning makes sense, to me it didn't matter whether the previous signal was a buy or sell. It's still a signal, either way. Think of it as a new or renewed signal. In any event, the first trade was the biggest loss that setup has seen, so it certainly didn't sway the results for the better!

Regards,
Alex

Dave said...

Hi Alex,
Thanks again for the reply.

I made the changes to my S&P Trades worksheet. I'm still trying to resolve a few of the differences.

If you get a chance, can you check to see what I'm doing wrong?

Your results:

Cot Profit 520.7%
Index Profit 266.6%
Winning/Loosing Trades 32/13

My Setup (Date range 7/11/1995 to 5/1/2007 )
Cot Profit 400.99% ($100 to $500.99)
Index Profit 168.69% ($560.34 to $1,505.57 )
Winning/Loosing Trades 33/13

So, it looks like I'm off ~100% in my profits, and I have 1 more trade.

Did I miss something obvious?

I resent you the spreadsheet, but I'm also posting here in case someone else can benefit from this thread.

Thanks!
Dave

Alex Roslin said...

Hi Dave,

Thanks again for your message. Firstly, let me say congratulations for crunching the data as you've done. I think this kind of analysis really opens your eyes to the possibilities and challenges of trading and investing, whether with COTs or any other means.

To answer your questions:

1) The results on my "Latest Signals" table are not expressed as a percentage profit, but as the return starting from a baseline of 100. (See the notes for the spreadsheet.) I've just clarified the notes for this table to explain this better.

2) The remaining small discrepancy in our results is due to the fact that my results for the SP500 setup end on May 1, 2007, which is a few weeks into a new trade. (See the "Last Update" column on the "Latest Signals" page to see the date for which the setup results are good.) In your case, your results appear to end with the last trade, which is on March 20, 2007.

Mind you, for purposes of the statistical confidence testing, I rely only on completed trades.

Regards,
Alex

Dave said...

Thanks for the clarification.

This should provide a baseline for comparison against other setups.

Cheers!
Dave

Alex Roslin said...

And thanks to you for sticking with it and bringing a source of possible confusion to my attention. It's always great to see someone else corroborate the interesting results this data can yield.

Take care,
Alex

Dave said...

Hi Alex,
Just fyi...I may have found a better profit setup for the S&P 500.

I sent you the setup earlier today, and am just posting this coment in case my email got filted as spam.

No need to approve this comment, just wanted to let you know. I'm interested in your thoughts about the setup.

Cheers!
Dave

Alex Roslin said...

Hi Dave,

Just took a first preliminary look. Looks like a winner. Congratulations on your discovery, and thanks for sharing it. I was planning to retest all the setups starting early next year, as part of a regular exercize of reoptimizing them with the new data, but your discovery for the SP500 makes me think I should redo that one sooner. I'd like to see if I can find something even better.

For the DIYers out there, Dave's setup works by fading the small traders when their net % of open interest position is 1.1 or more standard deviations from its 15-week moving average. Like the 17-week setup I'm using, Dave flipped to a bullish signal with the Sept. 25 COTs report, and it's still on a buy signal right now.

The SP500 setup was the first one I developed, back when I was still working out the best ways to backtest for the ideal parameters.

Regards,
Alex