• Home
  • Forums
  • Trades
  • News
  • Calendar
  • Market
  • Brokers
  • Login
  • Join
  • 6:38pm
Menu
  • Forums
  • Trades
  • News
  • Calendar
  • Market
  • Brokers
  • Login
  • Join
  • 6:38pm
Sister Sites
  • Metals Mine
  • Energy EXCH
  • Crypto Craft

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Curve fitting vs optimization 48 replies

Curve Fitting: The Ultimate Conundrum 17 replies

EA Backtesting and Curve Fitting Concern 2 replies

(curve)-fitting and overfitting, good examples wanted 0 replies

"curve fitting" or just improving?? 5 replies

  • Platform Tech
  • /
  • Reply to Thread
  • Subscribe
  • 2
Attachments: Optimization, statistics, and avoidance of curve fitting
Exit Attachments
Tags: Optimization, statistics, and avoidance of curve fitting
Cancel

Optimization, statistics, and avoidance of curve fitting

  • Last Post
  •  
  • 1 Page 2
  • 1 Page 2
  •  
  • Post #21
  • Quote
  • Jan 9, 2021 3:06am Jan 9, 2021 3:06am
  •  remonpilip
  • Joined Nov 2019 | Status: Member | 164 Posts
I always have this philosophy when making a strategy, you only need 2 filters ,3 MAXIMUM.

which is:
ONE DIRECTIONAL filter.
One to Two VOLATILITY filters.

DIRECTIONAL filter:
>> This filter purpose is mainly to take as many trades as possible, don't waste your time too much on optimization on this filter, as its main purpose is only to bring trade signals.
>> A good Directional filter is one that takes a lot of trades and considers if the market is a more Momentum market or a Mean-Reverting market.

VOLATILITY filter:
>> A good volatility filter will make a bad Directional filter look good, because its main purpose is to not trade when the market is either experiencing a low market volatility or high market volatility.

Note: don't optimize so much on the settings,, Use common sense settings to avoid curve-fitting (i.e. 5 days 20 days ,etc.)
56 is quite random number and not common sense, unless you did the optimizing and found out that when you change 56 to say for example 50,51,52,53,54,55,56,57,58,59,60 and the profit/dd result don't vary much from 56, then you can use 56.

However for an Intraday approach, I would also consider 3 time filters:
-Hours of the Day (time allowed entry & time must exit all position) -> only trade from 8am to 3pm.
-Days of the Week (use it carefully as this filter is more vulnerable to curve-fitting)-> don't long on Wednesday & don't short on Mondays.
-Months of the Year (this is a seasonal filter, as per year you can see which months of the year is CONSISTENTLY bad month for longs or bad months for shorts)-> never long on August, never short on July.
Let ur winners run & cut ur losses short.
 
 
  • Post #22
  • Quote
  • Last Post: Edited 1:45pm Jan 9, 2021 3:40am | Edited 1:45pm
  •  ryuryu
  • Joined Apr 2020 | Status: Member | 1,870 Posts
Quoting mfurlend
Disliked
I have been curious about this subject, and I'm interested in hearing some educated opinions on the matter
Ignored
VERY interesting thread!

I came to conclusion that strategy tester is only for visual tests of the strategy to find bugs (if you are not EA seller so you need curve fitting to create "great" stats). For real trading it is better to have bullet proof approach to go.

Second idea is about that all are trying to optimize static values. But must not, must make them dynamic (like averaging distance, lot sizing, TP, SL etc). For example there is "new" thread from another "20 years of trading guru" where he is showing another grid EA. As you can see from settings there are stoch, rsi (the same time), static lot size, step. What to optimize here? Static step for averaging, stoch/rsi settings?

Third idea is that most are trying to create EA that act like human. For example using EA on higher time frames. This make no sense. EA must do what human can't. Also trying to predict/determine trend/range. This can't be achieved too and no optimization will help.

The last but not the least. Applying more filters (like do not trade on Friday, after news, at lunch time, in 2020, on riding moon etc) just add more mess.
Observer effect
 
 
  • Platform Tech
  • /
  • Optimization, statistics, and avoidance of curve fitting
  • Reply to Thread
    • 1 Page 2
    • 1 Page 2
0 traders viewing now
Top of Page
  • Facebook
  • Twitter
About FF
  • Mission
  • Products
  • User Guide
  • Media Kit
  • Blog
  • Contact
FF Products
  • Forums
  • Trades
  • Calendar
  • News
  • Market
  • Brokers
  • Trade Explorer
FF Website
  • Homepage
  • Search
  • Members
  • Report a Bug
Follow FF
  • Facebook
  • Twitter

FF Sister Sites:

  • Metals Mine
  • Energy EXCH
  • Crypto Craft

Forex Factory® is a brand of Fair Economy, Inc.

Terms of Service / ©2023