Friday, 22 August 2008

Trader Data Smiles on Equities & Commodities; Dollar Done?

A seachange of sorts has just happened in trader positioning in derivatives markets, as reporting in today's weekly data release from the Commodity Futures Trading Commission. Equities are suddenly on much firmer ground, at least on a short-term basis going forward, while commodities aren't doing too badly either. Take a look at my latest signals page table for how my trading setups based on this fascinating government data are shaping up. Here are some highlights:

- My setups for the S&P 500 and Russell 2000 have flipped to the bullish column with today's data, with execution for the open of Monday's trading. The S&P 500 setup was in cash for the past six weeks, while RUT went into bearish mode this week. (I had reported erroneously last week that the S&P 500 setup was set to remain in cash at least until mid-September, but I didn't notice a bearish signal in the data for the week of July 29 that takes effect Monday.) The S&P 500 signal, incidentally, will only be in effect for a single week. This one-week bullish blip is based on the commercial traders turning quite bullish in the week of July 29, before flipping back to bearish since then. Since this setup remains in cash unless both of its signals agree, it goes back to cash the week of Sept. 1. Under my risk-control rule (see my "How It Works" page for more on that), I will be following the Russell 2000 signal (since my BKX Bank Index setup is also bullish) but ignoring the S&P 500 signal (because my Dow industrials signal is bearish).

- A bunch of new developments in commodities: heating oil and gold both go bearish as of Monday, but four of my six highly related commodities setups will still be bullish Monday (copper, platinum, crude and silver). Thus, I will sell my long gold bullion position but not go short. I'll also put on a long copper position, as called for by my setup for that market. Incidentally, I've been re-examining the copper data in my continuing efforts to refine my setups and this afternoon found a possible replacement setup that combines two of my best signals for the commercial traders and small traders. That new setup, which I could end up adopting after further testing, was in cash for the past four weeks and also just flipped to bullish for next week's open.

- The U.S. dollar index has staged an impressive rally in the past few weeks, briefly piercing above the downtrend line it has formed since early 2006. But my chart shows that breakout now failing, and the Commitments of Traders data looks pretty grim as well. The commercial traders in dollar futures are now more net short than any time since Nov. 2005, just as the greenback peaked. In relative terms compared to past data, they're 2.28 standard deviations below the moving average I use in this setup. I'd say that's a good bet for a new downleg.

- Lots has been said about the bullishness of Treasuries of late and how they've disconnected from inflation expectations. I've certainly noticed that in the commitments data. (See, for example, this post on how the Fed Funds contract positioning forecast weeks ago that Bernanke was highly unlikely to hike rates any time soon.) In Friday's data, the Fed Funds positioning remains quite bullish, albeit somewhat less so than in mid-July. Positioning remains one standard deviation above the average I use for my setup in that market, which means the market believes the Fed Fund rate, if anything, faces continued downward pressure. However, positioning in the 10-year Treasury is now heading off in its own direction. The commercial traders have sharply reduced their net long position in the past three weeks as a percentage of the total open interest. This week, they sit barely a hair from causing my 10-year setup to flip to bearish (meaning the yield would go up).

New markets update: A whack of new ETFs has launched to cover sectors in the commitments data that until now had to be traded using derivatives. These include cocoa, coffee, sugar, lumber, cotton, copper and platinum (both long and short). What's especially interesting about the first five on that list is they have only moderate to zero correlation with each other and with any other market in the basket for which I have signals. Between 1995 and 2007, the strongest correlations were -67% for cotton and BKX, -60% for cotton and SPX, and 58% for cotton and coffee. The results show these new markets present a big opportunity for adding diversification to an overall portfolio.

I also did a study comparing the volatility of the new markets and my existing ones. I typically like to be in the most volatile markets, as they present the highest potential for returns. Here's a list of the most volatile markets (in order): copper, heating oil, natural gas, crude oil, NDX, platinum, silver, coffee, gold, Russell 2000, sugar, BKX Bank Index, cotton, Nikkei, SPX, Dow industrials, lumber and cocoa.

Hope you have a relaxing weekend. See you early next week for my portfolio update and possibly more announcements about new setups.

TAGS: S&P 500, Russell 2000, Bernanke, Fed, interest rates, gold, silver, U.S. dollar, crude, NASDAQ 100, copper, platinum, heating oil, Dow industrials, Treasuries, Fed Funds, BKX, Bank Index, NDX, COT, Commitments of Traders, market timing, trading system development, CFTC, Commodity Futures Trading Commission, COTs Timer


Anonymous said...

I know you write about climate change and I just saw this website which thought might interest you ( ).

Did not know this would be possible. They should put weather in COT too!

Alex Roslin said...

Hi Anonymous,

Thanks for your message. That's pretty interesting. I understand the frozen orange juice contract that is in the COT data is seen as a good proxy for Florida weather, but I haven't looked into it myself yet. There is also a new ETN for carbon, symbol GRN.


Anonymous said...

Hi Alex,

I've been looking at your system and blog for a while and I find it very interesting.

I am currently doing some research on trading sytems for a paper I have to write for school (I am in my final year at HEC, in Paris). I was therefore trying to replicate your calculations on S&P500.

I managed to get the signals perfectly, but now, I am trying to backtest the PnL that such positions would have generated, and I can't really match your calculations, as your S&P500 futures are very different from those I was able to download (e.g. they start around 500 in 1995 on your spreadsheet, whereas I see them always above 850). Could you please explain your process on that particular point or where you got the data from?

Thanks a lot in advance.

Alex Roslin said...

Hi Delphine,

Thanks for writing. I don't use futures prices, but rather the cash prices, since I'm trading this system with cash-price-based securities, not futures. The SPX data comes from Yahoo Finance and is the weekly open price:^GSPC&a=02&b=3&c=1995&d=07&e=28&f=2008&g=w


Alex Roslin said...

Hi again Delphine,

Please send along your paper when you're done.