Wednesday, 2 May 2012

Nasdaq Signals Gained 1,800%, While Market Fell 70%

My new trading setup for the Nasdaq-100 Index gained 1,800 percent since 1999, while the index itself lost 70 percent, according to my backtesting results using detrended price data. (Scroll down to the "Backtesting Results Table.")

And my new setup for the Nikkei went up 680 percent, while Japan's stock index fell 40 percent.

These are some of the new results I've posted for four new trading setups that give me signals based on the free weekly Commitments of Traders reports. These reports are issued by the U.S. Commodity Futures Trading Commission and reveal trillions of dollars of trader positioning in 150-odd markets - everything from gold to crude oil, the Canadian dollar and S&P 500 futures and options.

I've been developing my COTs Timer trading strategy based on the COT data for several years and using myself as a guinea pig to trade my signals in real time since 2007. I've just gone through a major updating process and have been improving my backtesting process based on my experiences.

I've posted the backtesting results for the Nasdaq-100 (which are based on the Nasdaq-100 mini COT data), the Nikkei, the benchmark BKX U.S. Bank Index (which are based on the three-month Eurodollar contract - the liquidity measure, not the currency) and the 30-year Treasury bond.

I've also just updated my latest signals table to include signals and data from all of these new setups. See my FAQs page for more details on my backtesting process.

Thanks for your suggestions for new markets to test using the COT data. I plan to work on silver and the Euro/USD cross next.

Your comments and suggestions are welcome. I'll also post a new sample spreadsheet for my S&P 500 setup soon on my DIY page. Thanks for your patience, support and kind words through this process.

Good luck the rest of the week, and see you back here after Friday's COT release.


jeff said...

1. How does your stops work. For example the Nasdaq has a 12% stop - does that mean if current price falls under the entry by 12% then to close the trade?

2. Also the signal line entries, I take is the output of your regression test.

3. How do you delay the COT data to see determine the optimal entry point into the market? Is it automated or do you offset the data by 1 week at a time until it is optimized?

jeff said...

Did you ever look at commercials from the perspective of
-swap dealers
- swap dealers spreading

Or were they not significant?

Alex Roslin said...

Hi Jeff,

Thanks for your messages. Yes, if the price falls below my stop, I close the trade. The signal lines do come from the backtesting, as do the trade delays. My spreadsheets and testing app are set up in such a way that I can see the results from the various trade delays. I don't consider setups with greater than two weeks of trade delays because, while potentially robust statistically, I found my real-time trading results showed there was a significant negative correlation between the results and the average trade delay of the signals in each setup.

For your second comment, I haven't looked at the disaggregated COT data because it doesn't go back nearly enough to be useful for backtesting or reliable trading signals in my opinion. We'll probably need many more years of trading data, including a good mix of varying market conditions, before that data could be useful for robust signals - at least the way I interpret the data.