Provided By Jake & Bernstein @ 2Chimps.net
Good Trade To Put In Your Journal...
See Chart Description for Specif Rules and Downloadable PDF written by Jake
Green Line = January in a year that ends in 5
Blue Line = December in a year that ends in 5
Rule - Monthly Chart. (See PDF for Details)
- If We Get a Monthly Close Above January High Enter Long
- Close Trade last day of the year.
Potential Symbols: DIA , SPY , Or Leveraged Versions...
No Losing years going back to 1900!
Only time trade did not work is 2005 and there was NO ENTRY Trigger.
Link to PDF:
However I would say that there have been many "Seasonal" type trades found using Presidential cycles, Years that end in X#, Specific years in Presidential Cycle based on what term they are in. These have worked so far...and the sample rate is low. This is too small of a sample set to base an entire trading plan on, but putting it in a basket of trades like the ones mentioned above....just works.
Where I am a little taken back is when you say it's Curve Fitting. I develop systems for a living...and Jake is one of the most respected Trading System Developers of our time. We found that years that end in 5 typically end UP for the year. To capitalize off a entire years move the highest time frame is 99.9% of the time the most reliable. Therefore a Monthly Close above the January High was a common sense one to test. This Trade Setup was noticed and announced before 2005 using the Monthly close. It didn't use to be a weekly or daily close and we changed it to Monthly after 2005. That would be Curve Fitting....Unrealistic...and in my opinion unethical.
Sample set too low to base an entire system on = YES. Curve Fitting...I have to respectfully disagree.
Curve fitting doesn't have to be intentional and I am not saying it was intentional, so there is no need to mention ethics here. I understand that you simply did this as a matter of curiosity and I still did find it interesting despite flaws that you are already aware of.
Typically over optimization and curve fitting result in smaller sample sizes so they are very closely related concepts. In this case the 'years ending in 5' universe constraint has been applied to a simple entry rule. This has given you perfect expectancy but it has also reduced the sample size dramatically. However there is no real logical reason why you can't apply your entry rule to every year instead of just the years that end in 5. As a suggestion it would be interesting to see this same entry rule applied to every year over the past century. Even if that lowered the expectancy to the 55-60% range it would be a lot more meaningful and robust IMHO.