more articles by

Alex Roslin

Click to enlarge Click to enlarge


COTs Flip to "Bullish" on Gold

By Alex Roslin      Printer Friendly Version
Jun 18 2007 10:37AM

The “smart money� commercial traders have given a “bullish� signal for gold and gold-related stocks, according to my trading system based on the Commitments of Traders report.

Friday’s report - based on data as of last Tuesday, June 12 - showed the commercial hedgers have flipped to a historic extreme of bullishness that in the past has tended to lead to a profitable buy signal for the Gold Bugs Index (HUI), the USERX Gold Fund and the Canadian iUnits gold exchange-traded fund (symbol XGD). (See table below for more details.)

According to my best trading setups for HUI, USERX and XGD, it was most profitable to execute the trade on the next weekly open - that is, Monday, June 18.

Meanwhile, gold itself flipped to “bullish� in the COTs report issued June 1, according to my new setup for this market. This trade was also best executed with no trade delay - that is, on the next weekly open (in this case, Monday, June 4).

The weekly COTs reports are issued free Friday afternoons by the Commodity Futures Trading Commission. They detail trillions in futures and options holdings in 100 markets - from gold to gas, the S&P 500 and frozen pork bellies.

My trading system based on the COTs reports is the first one solely based on this valuable government data. I’ve been working lately on refining my trading setups for some markets in order to find ones that have given superior past profits and can be statistically validated at a high confidence level. I think I’ve made some interesting breakthroughs recently.

You’ll notice new results in the table below for my new setups for gold, silver, USERX and HUI. I validated the first three of these setups at the 95-percent confidence level with the Student’s T-test. (In other words, 95 percent of the time, the average trade can be expected to be profitable.) The T-test is often used to validate trading systems and other data like survey results.

As for my best setup for the HUI Gold Bugs Index, I’ve validated it at the 90-percent confidence level - which is less than the ideal 95-percent level I’d like to see in my setups. One limitation is that the HUI data goes back only to 1996, so there is less historic data than for the other setups.

As for my XGD and copper setups, I can’t yet give a confidence level because there are less than 10 trades in each, which is too few for validation purposes. In other words, we can’t say with confidence how past results would translate into future profits. I’m still working on refining the HUI, XGD and copper setups further in hopes of finding ones with better statistical validity. Stay tuned.

See my blog for more details and for signals in the equities and other markets.



New signal 1

Rene-wed signal 2

COTs Timer Ratio 3

Existing signal (signal date) 4

COTs system profit 5

Index profit 6  

COTs vs. Index profit 7

Larg-est draw-down 8 

Traders to watch 9

Gold 10










Silver 11









Small Traders

US Gold (USERX) 12










Gold Bugs Index (HUI) 13




Bullish (12-Jun-07)






TSE Gold (XGD.TO) 14










Copper (high grade) 15









Large Specs


1. See my blog for important notes on how I trade new signals: Click the “How It Works� link in the right-hand column.

2. A “renewed� signal is when a market is already on a buy or sell signal, and traders again register an extreme net trading position in the same direction. I don’t usually act on such signals, unless I don’t already have a trade on. I haven’t calculated the profitability of entering trades on renewed signals.

3. The COTs Timer Ratio is the latest reading of the bullishness or bearishness of the group of traders that has historically been most profitable in this market. Unless otherwise indicated, a reading of 1 or more means a buy signal for the commercial traders or a sell for the large specs and small traders. A reading of -1 or less means a sell for the commercials or a buy for the large specs and small traders. A reading between 1 and -1 means no new or renewed signal. The ratio is based on the traders’ net percentage-of-open-interest position compared to the position’s moving average divided by two standard deviations.

4. In parentheses are the dates of the COTs report that gave this signal. Note that my profit/loss calculations were based solely on taking trades on the first or second weekly open after the signal was given. (Click the “Profit/Loss Results� link on my blog to see the trade delays for each setup.) Variations in the trade delay resulted in inferior past results or losses.

5. Past return using the signals of my COTs Timer system, starting from a baseline 100. This is the theoretical return from buying the security on a buy signal and shorting it on a sell signal. The return doesn't include commissions, slippage or other costs.

6. Past return from buying-and-holding the underlying cash market, starting from a baseline of 100.

7. Ratio of the COTs Timer return versus the underlying cash market return.

8. Largest drawdown the setup experienced during a trading signal since the beginning of the data. This was not necessarily the loss at the end of the trade. I use this figure to calculate my maximum portfolio allocation for the setup based on my 2-percent risk threshold of total assets for any one trade.

9. The group of traders that had the best historic returns in each market. My signals are given when this group reaches specific extreme levels of bullishness or bearishness. Unless otherwise noted, my system trades in the same direction as the commercials and fades - or trades opposite to - the large speculators and small traders.

10. Results are based on NYMEX gold and were last updated for the week of May 15, 2007. I plan to update the results for each setup at least once a year.

11. Results are based on NYMEX silver and were last updated for the week of May 15, 2007.

12. Signals for the U.S. Global Investors Funds U.S. Gold Fund (symbol USERX) are based on a setup correlated to the gold COTs data. Results were last updated for the week of June 12, 2007.

13. Signals for the HUI Gold Bugs Index are based on a setup correlated to the gold COTs data. Results were last updated for the week of July 12, 2007.

14. Signals for the S&P/TSE Canadian Gold iUnits ETF (symbol XGD.TO) are based on a setup correlated to the gold COTs data. XGD.TO price data available only since March 2001. Results were last updated Feb. 27, 2007.

15. Results are based on NYMEX copper and were last updated May 17, 2007.


Alex Roslin
Monday, June 18, 2007




Disclaimer: This report isn’t meant as financial advice or a recommendation to buy or sell any security. My system isn’t for everyone and involves substantial risk, including large drawdowns in some trades. Past results are no guarantee of future profits. I’m not a certified financial advisor. While I consider my information to be reliable and accurate, I make no guarantees. Please do your own homework before trading.

Alex Roslin’s COTs Precious Metals Review -