Excellent article and study as always. Beside the robustness of your study I like the way you bring your message across and your writing style very interactive. As a quick inquiry do you mind charing the average win rate of the system, I know it is not important from a performance results bit I just wanna compare it to my other systems. For small retail traders like us sometimes win rate and max time to recover can play a major in the implementation of the syatem from a mental perspective Have a lovely day and keep them coming
Excellent article and study as always. Beside the robustness of your study I like the way you bring your message across and your writing style very interactive.
As a quick inquiry do you mind charing the average win rate of the system, I know it is not important from a performance results bit I just wanna compare it to my other systems.
For small retail traders like us sometimes win rate and max time to recover can play a major in the implementation of the syatem from a mental perspective
Hi N. Sorry I took forever to respond. Thanks for your comment. I missed the position level analytics in the post. I am coding an QC'ing the analytics in detail and will make my next post about it. Probably I should include these analytics in this post itself. The win rate of the system with 10 position cap and momentum entry ranking is 46%, for a total of 678 positions in 20 years. Median trade duration is 56 bars, mean is 73 bars, and longest trade ever is 376 bars, for London Cocoa, exiting May 2024. The longest drawdown was 595 days (425 bars) from April 2006 to Nov 2007. There were 4 drawdowns longer than 300 days.
Impressive coverage of many trading strategy considerations. Thank you! Working on similar concepts for individual equities. Limiting entries is a must as the number of “markets” is too big to take all entries. Finding similar results. Have a few more ideas to chase down now too! Thanks again.
Thanks Vic. I'm glad you found the article useful. Definitely... for individual stocks there are some other challenges. It's hard to beat the S&P 500 or Nasdaq 100 index (when normalized by volatility) with trend following on individual stocks. So far "no-joy" for me on that front, but I'm exploring it again using RealTest tool. Plus mean reversion systems, etc.
Did think of something earlier today… I think you have trailing stop (5x ATR) as initial stop. The whole nature of the system changed for me when I incorporated a separate initial stop (which doesn’t change). Something between 2 or 3 ATR. The reason it changes everything is because the tight stop means much larger positions. Basically, you end up with more losers, but the winners overcome that hurdle and then some (from their sheer size). In other words, you hit the initial stop 2, 3, or 4 times before the trend takes off. But the position size is 2-3 times larger so when the trend gets going the position quickly covers those initial loses. Another comment, at least for stocks, 10x ATR as a trailing stop worked better than 5x. This probably only “works” if you also have a tight initial stop. (I haven’t tested this, just trying to picture it in my mind)
Awesome I'll test that. With position count limits the position sizes will be the roughly the same regardless of the # of ATRs to stop. But adding the tight initial stop will change how the strategy behaves. I'll test that out and let you know the results. There's a huge number of rabbit holes we can go with this strategy.
Plenty of rabbit holes for sure. The 10x trailing stop did make for consistent improvements over 5x in the stuff I tested. There are always exceptions and a few did win at 5x, but just a few. A 2x initial stop means more activity, but its effect on position size makes the biggest difference. How this works out... using derivatives you have the leverage necessary to put on a much more diversified portfolio (more concurrent active positions). When you're using equities this becomes a challenge as the 2x initial stop implies decent size for positions and therefore limits total number of concurrent trades and you definitely can't take all the trades. This can probably be managed through filtering for a very limited universe. Something like top X by volume or performance over past ?... or as simple as N100 constituents. In my tinkering I found the initial stop and 10x to be two big factors. Other than that, there does seem to be a couple sweet spots as far as lookbacks for breakout entries. Hope you find this useful.
Excellent article and study as always. Beside the robustness of your study I like the way you bring your message across and your writing style very interactive. As a quick inquiry do you mind charing the average win rate of the system, I know it is not important from a performance results bit I just wanna compare it to my other systems. For small retail traders like us sometimes win rate and max time to recover can play a major in the implementation of the syatem from a mental perspective Have a lovely day and keep them coming
Excellent article and study as always. Beside the robustness of your study I like the way you bring your message across and your writing style very interactive.
As a quick inquiry do you mind charing the average win rate of the system, I know it is not important from a performance results bit I just wanna compare it to my other systems.
For small retail traders like us sometimes win rate and max time to recover can play a major in the implementation of the syatem from a mental perspective
Have a lovely day and keep them coming
@carlos mata
Hi N. Sorry I took forever to respond. Thanks for your comment. I missed the position level analytics in the post. I am coding an QC'ing the analytics in detail and will make my next post about it. Probably I should include these analytics in this post itself. The win rate of the system with 10 position cap and momentum entry ranking is 46%, for a total of 678 positions in 20 years. Median trade duration is 56 bars, mean is 73 bars, and longest trade ever is 376 bars, for London Cocoa, exiting May 2024. The longest drawdown was 595 days (425 bars) from April 2006 to Nov 2007. There were 4 drawdowns longer than 300 days.
Impressive coverage of many trading strategy considerations. Thank you! Working on similar concepts for individual equities. Limiting entries is a must as the number of “markets” is too big to take all entries. Finding similar results. Have a few more ideas to chase down now too! Thanks again.
Thanks Vic. I'm glad you found the article useful. Definitely... for individual stocks there are some other challenges. It's hard to beat the S&P 500 or Nasdaq 100 index (when normalized by volatility) with trend following on individual stocks. So far "no-joy" for me on that front, but I'm exploring it again using RealTest tool. Plus mean reversion systems, etc.
Amazing article!
Thanks! I hope this article addresses your comments from the previous article: https://tradingstrategies.substack.com/p/setting-up-a-portfolio-of-futures
Did think of something earlier today… I think you have trailing stop (5x ATR) as initial stop. The whole nature of the system changed for me when I incorporated a separate initial stop (which doesn’t change). Something between 2 or 3 ATR. The reason it changes everything is because the tight stop means much larger positions. Basically, you end up with more losers, but the winners overcome that hurdle and then some (from their sheer size). In other words, you hit the initial stop 2, 3, or 4 times before the trend takes off. But the position size is 2-3 times larger so when the trend gets going the position quickly covers those initial loses. Another comment, at least for stocks, 10x ATR as a trailing stop worked better than 5x. This probably only “works” if you also have a tight initial stop. (I haven’t tested this, just trying to picture it in my mind)
Awesome I'll test that. With position count limits the position sizes will be the roughly the same regardless of the # of ATRs to stop. But adding the tight initial stop will change how the strategy behaves. I'll test that out and let you know the results. There's a huge number of rabbit holes we can go with this strategy.
Plenty of rabbit holes for sure. The 10x trailing stop did make for consistent improvements over 5x in the stuff I tested. There are always exceptions and a few did win at 5x, but just a few. A 2x initial stop means more activity, but its effect on position size makes the biggest difference. How this works out... using derivatives you have the leverage necessary to put on a much more diversified portfolio (more concurrent active positions). When you're using equities this becomes a challenge as the 2x initial stop implies decent size for positions and therefore limits total number of concurrent trades and you definitely can't take all the trades. This can probably be managed through filtering for a very limited universe. Something like top X by volume or performance over past ?... or as simple as N100 constituents. In my tinkering I found the initial stop and 10x to be two big factors. Other than that, there does seem to be a couple sweet spots as far as lookbacks for breakout entries. Hope you find this useful.