Poll: I trade only properly optimised systems

[OddTrader]

Quote from dandxg:

Or forward test on a good sim, which I believe even one should do after backtesting, atleast for a couple of weeks.

No doubt a very good Theory indeed!

 

[Mr B]

I assume that properly optimised means backtested many times with the variables subtly altered, and on a 3d chart it gave a smooth profit curve regardless of variables.

I know how to optimise but I don't (lazy).

 

[Trader666]

Do you mean SCT?

Quote from Hook N. Sinker:

The alternative to back testing is fantasize and hope.

 

[TrueStory]

Quote from Mr B:

I assume that properly optimised means backtested many times with the variables subtly altered, and on a 3d chart it gave a smooth profit curve regardless of variables.

I know how to optimise but I don't (lazy).

That is correct.

 

[Trader666]

By "properly optimised" do you mean optimization to ensure that the parameters surrounding the ones you're trading are just about as profitable?

Quote from OddTrader:

Do you trade only properly optimised systems?

 

[OddTrader]

Quote from Mr B:

I assume that properly optimised means backtested many times with the variables subtly altered, and on a 3d chart it gave a smooth profit curve regardless of variables.

I know how to optimise but I don't (lazy).

How about smooth profit curve regardless of market variations?

 

[OddTrader]

Quote from Trader666:

By "properly optimised" do you mean optimization to ensure that the parameters surrounding the ones you're trading are just about as profitable?

Perhaps that should be the minimum requirement for all systems, imo.

 

[taowave]

Quote from dandxg:

Or forward test on a good sim, which I believe even one should do after backtesting, atleast for a couple of weeks.

Dan,would you be kind enough to share your thoughts on foward testing???

Does it make a significant difference if one foward tests or simply adds the "foward" test data into the initial backtest??

I have always wondered about the virtues of foward testing

Thanks in advance

 

[vikana]

I most definitely test for parameter stability. That's the primary purpose of "optimization", imo.

How do you otherwise know if you're just chosen a lucky set of parameters that happened to look good in a particular backtest?

Test and compare to realtime (forward test). The only way to stay "alive" with systems.

 

[traderdragon2]

All my systems are highly optimized to find the "sweet spot" for parameters. You can do this without curve fitting though, and you must do this without curve fitting. Its all about system robustness. If the system behavior is erratic when you wiggle parameters, then its not a good system. All my systems make money in all markets even when using a wide set of parameters. In other words, they are robust enough to handle quite a large change in the market.