Separate entry and exit parameters optimization

Hello,

According to the strategy development process outlined at the Financial Hacker blog it's recommended to optimize entry parameters first, test the strategy, perform reality check and only then proceed with the exit parameters optimization.
So, in my case I want to optimize entry parameters with reversal exits first. Then apply a StopLoss and optimize the second time.

Blog says about the exit parameters optimization:



Questions:
1. I wonder how to optimize only exit parameters in my case and don't touch entry parameters? Does Zorro provide a mechanism to train several groups of parameters separately (i.e. optimize parameters of the 2nd group using optimal parameters of the 1st group; similar to separate training of RULES and PARAMETERS)?

2. Or should I train only entry parameters first and then entry & exit parameters together? But in my case, entry parameters in the 2nd optimization run will depend on default exit parameters. So entry parameters will be different compared to the 1st optimization run. Therefore, "don't touch entry parameters" requirement won't be fulfiled. What is recommended?

Thank you!

up

I don't know what type of algo you're trying to code, but in general the procedure is like you have stated.

First code your entry logic (with simple static exit), optimize entries (with WF and Montecarlo), add some exit logic, perform optimization (WF and Montecarlo always), compare, change exit if necessary, repeat, etc...

You must just add optimize instructions in your exit logic, keeping the optimizes in entries as well. First only entries and then the two of them.

The goal is to compare and see if your exit logic improves the ratios.

Maybe someone else can tell you something i don't know.

Thank you, brax! It makes sense.

What do you mean by static exit?

Please see 2 examples:
1st example: First optimize entry parameters + static stop=4*atr(100). Then optimize entry parameters with stop=optimize(4,2,10)*atr(100). No problem here, optimized entry parameters will be the same in both optimization cycles.

2nd example: first optimize entry parameters + reversal exit (I don't know whether we can name this exit static but it's definitely simple). Then optimize entry parameters with stop limit. There was no stop limit on the first step. So when we optimize the second time with stop, the entry parameters will be different.

Do you understand my concerns?

Hi kujo.

The thing is simple. It's not a question of entry parameters not changing, it's just a question of not altering the entry logic and focusing on the exit.

If you execute the code provided in steps 5 and 8 of the article, the entry parameters do change, despite having changed only the exit logic and not the entry one.

Every time you alter the logic (in any sense), parameters will surely change. The point is just setting an initial logic, see performance and after that, test several exit criteria and compare results.