Good point.
But still, would you buy a trading system that makes a new high after 9 years, even if the equity curve is relatively smooth and the drawdown is under 30%, especially if you are about to retire?
Seriously.
Probably not for a system which takes 9 years to made a new high ( I missed that 9 year part) =)
DanceWithTheTrend, you did get to the end of a tunnel. Iām not sure how, because you did not enter it.
Assuming system A has the same performance after 9 years that system B, system A makes just 2 equity peaks during 9 years, system B makes equity peak every 3 weeks.
Statistically over time system A will make equity peak at 3 years and 6 years (highest number of occurrences +/- 1 year). Also statistically both equity peaks value will be around 50% (+/- 25%) of capital after 9 years.
The point is that first equity peak will put system A years upfront of equity of system B. Statistically system A will be always front runner of system B.
I donāt want to discuss if someone want to run this model, if those 2 equity peaks in model A will happen or those every 3 weeks equity peaks in model B is happen. The point is there is no black or white or being right or wrong. The point is to understand the issue.
The point is that no trader wants to trade a system that needs 5, 6, 7, 8 or 9 years (or moreā¦) to make a new equity high, even if its drawdown is more than reasonable and its return on investment quite good.
This important time element is often overlooked by traders, here on C2 and elsewhere. Most of the time these traders just look at the % yearly return and the maximum drawdown of the system.
The point is that in trading you cannot guarantee 2 equity peaks every 3 years in system A and new equity peak every 3 weeks in system B. And it is much easier to control system B performance than system A performance.
True.
But once again, would you purchase a trading system that can remain up to 9 years in drawdown mode before making a new equity high, even if this system has reasonable drawdown and a good ROI?
Sure. Point me to system than makes equity peak every 3 weeks over few years.
If there is performance guarantee, you always will choose system A, after first equity peak switch to system B and after 1-2 years (depends when equity peak in system A did happen), switch back to system A to catch 2nd equity peak. And after 2nd equity peak of system A, you switch to system B to finish 9 years period.
Thatās not exactly the issue here Marekj, I am simply replying to the original poster who wrote, and I quote : āI think systems making new all time time with multiple years of trading would be a lot more significant than systems making 90 days high with limited track recordsā.
And by the way, good systems with a long track record were once systems with a limited track record.
1 Like
Yes. To make picture full ā¦
Every good system has in past limited tract record. But only small percentage of systems with limited track record became good systems.
This is no both ways freeway.
I assume you spent the last few years analyzing hundreds and hundreds of C2 systems to come up with that conclusion, of course.
Anyway.
Nice discussions here. In case if you missed my original points, systems with only a short term track record has a big unknown into its future performance and the risks of trading them is pretty high due to the lack of track records to demonstrate its robustness. Even with nice looking backtested simulation results, there are chances that these systems were overfitted in optimization. Some over-optimized systems would do well for a short period before they collapse. And these systems have not stood the test of time with various market conditions out-of-sample.
Systems with a long term record out-of-sample have at least gone through the tests of many different market conditions (bullish, bearish, trending, choppy, high volatility, low volatility etc etc). This was the basic criteria I would use for evaluating a system but it was not the ONLY criteria. There are many other factors I would use to thoroughly evaluate these systems. I have been developing and trading my own systems for over 15 years. All my systems would need to be able to survive well with at least 10 to 15 years of data in my backtest. Hope this clears up my point.
Again this is not necessarily true all the time.
A 3 month old system with 2000 trades has a much much better track record than a 10 year old system with only 17 trades, regardless of its performance.
You can do it yourself in 5 min. Just go to The Grid, setup your filter and start increasing Strategy Age.
The Grid only includes the systems that are currently active, not those that are no longer in service.
Statistically there is a reason why systems are no longer in service. They failed.
1 Like
I wish you luck if you believe a 3 month old system with 2000 trades and good track records is a good one to consider trading. Cheers
2 Likes
I suggest you read Perry J. Kaufman book āTrading system and methodsā, if you have not already done soā¦
You will see that āyoungā systems with a large number of trades are statistically more meaningful than āoldā systems with only a few trades.
Anybody can optimize a 25 year old garbage system (that produces letās say 25 trades) and make it look very very good.
But try to optimize a system that generates 2000 trades in 3 months, itās impossible.
The number of trades is much much more meaningful than the āageā of the system.
I have read Kaufman book and literally hundreds of trading books through these years. Yes, we can create systems which generate these amount of trades and yet looking good within a short period. IMHO, the number trade samples and duration of system in the markets covering a variety of markets conditions are equally important. Out-of-sample walk-forward testing and stress testings are also keys to evaluate the robustness of a system plus a number of other measurement criteria. Feel free to maintain your presumptions and hope you do well with your trading. Good luck.
1 Like
Yes, but only if the system is good to begin with.
Bad systems, on the other hand, can remain profitable after 100 or even 150 trades. Beyond that level the trader will clearly see that the trading signals generated by these systems are no better than random calls.
In other words the equity curve becomes negative in the long term due to transaction cost, spread, slippage, etcā¦
Thanks, although luck has very little to do with profitable trading.
