Tuesday, 10 June 2008

New S&P 500 Setup Outperformed Market by 2.4 Times Since '03; Currently On a Buy

Is the Commitments of Traders data becoming less relevant? Are too many people following it and arbitraging away the edge it once provided? Some have suggested that's the case. I disagree. The data, which tells us how major traders are positioned in 100-odd markets like gold, crude oil, the NASDAQ 100 and frozen orange juice, is still far too little-known for it to have much impact on trading volume. Also, few people have actually done any testing to see how best to trade with it. The data in its raw form as published by the Commodity Futures Trading Commission is pretty much useless for trading purposes. It's easy to test this for yourself. Plug the data into an Excel spreadsheet and you'll see that the week-to-week changes in trader positioning correlate poorly with market fluctuations.

That's why I'm pretty excited to report the specs for my new trading setup for the S&P 500 based on this data. My research found dozens of setups for the S&P 500 that outperformed the market by large margins since both 1995 (the start of the data) and 2003. Note that outperformance since 2003 is especially noteworthy because most of that period has seen a clear uptrend for the S&P 500.

The S&P 500 setup I finally settled on draws on a new approach I hadn't tested for before - using the total open interest (i.e., the total number of long futures and options contracts plus the total number of short contracts). The setup combines the signals of two setups - one in which I fade, or trade opposite to, the small trader total open interest; and a second in which I trade on the same side as the "smart money" commercial trader net number of contracts as a percentage of the total open interest. The idea is to trade alongside the commercial traders, who in this market and the specific timeframe I chose, have the best information on market dynamics, while doing the opposite of the "dumb money" crowd (that's, unfortunately, the little guy in this case!) If the two signals don't agree, as happened half the time, I stay in cash.

Amazingly, the setup's compound annual growth outperformed the S&P 500 by 46 percent since 2003 while being out of the market 51 percent of the time. That's including a 0.2-percent charge per trade to account for commissions and slippage. Take a look at the newly revamped main table linked on my Latest Signals page for all the details on this new setup and a bunch of new measures I'm using to check all my setups. Btw, the new setup has been on a "buy" signal since the April 15 COTs report (with the entry on the open of April 21).

Note that I've created a new measure to evaluate setups called the Walk-Around Test, which tests each of the setup parameter values in order to see how stable or spiky the setup is in relation to "neighbouring" setups and help reduce the risk of data-mining. That table is still a work in progress, and I plan to fill in many of those blank cells soon. As well, I've temporarily removed some of my older, less robust setups that I haven't had a chance to update - particularly the agriculturals. When I get around to revising those, I'll stick them back in, and eventually I plan to add a few new ones too, including some other currencies.

If you're especially adventurous, see my S&P 500 Sample Spreadsheet and DIY Page to download your own copy of the new setup in Excel or Google Docs. Have fun!

TAGS: Commitments of Traders, COT, Dow Jones industrial average, NASDAQ 100, NDX, data mining, Commodity Futures Trading Commission, CFTC, S&P 500

No comments: