forex software

Create and Test Forex Strategies

forex software

fsbpro_guide:strategy_generator

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision
Previous revision
fsbpro_guide:strategy_generator [2014/11/20 14:44] – [1.3 Generator Indicators] yavorfsbpro_guide:strategy_generator [2016/07/21 04:51] (current) – [1.2 Generator Settings] Miroslav Popov
Line 1: Line 1:
 ====== Strategy Generator ====== ====== Strategy Generator ======
  
-Strategy Generator - changes the strategy, does not change the market environment. It allows you to generate strategies fully or semi-automatically.+The Strategy Generator allows you to generate strategies fully or semi-automatically.
 Each strategy has its own generator in its tab. If you run multiple generators at the same time will work independently without interfering with each other. How much generators you can use at the same time depends on your choice and how fast you computer is. A standard computer can deal with at least few generators working at the same time. Each strategy has its own generator in its tab. If you run multiple generators at the same time will work independently without interfering with each other. How much generators you can use at the same time depends on your choice and how fast you computer is. A standard computer can deal with at least few generators working at the same time.
  
Line 108: Line 108:
 **Generator uses statistical database** the FSB Pro generator is way "smarter" than the Generator in the old FSB version. It uses statistical data and artificial intelligence (AI) to learn from the profitable strategies it creates to create even better strategies. **Generator uses statistical database** the FSB Pro generator is way "smarter" than the Generator in the old FSB version. It uses statistical data and artificial intelligence (AI) to learn from the profitable strategies it creates to create even better strategies.
  
-**Optimize strategy** this is **not** the Optimization feature of the Strategy Tab. If this option is enabled the Generator will just try to use other values and settings for the indicators in the strategy to see if it can make it a bit better. If the after the optimization completes and the strategy still is passing the Acceptance Criteria and is better than the old one, the new one will be added to the collection instead. It is a good idea to keep this option enabled since it re //This is not full optimization. You can use the Optimizer for that.//+**Optimize strategy** this is **not** the Optimization feature of the Strategy Tab. If this option is enabledthe Generator will just try to use other values and settings for the indicators in the strategy to see if it can make it a bit better. If the optimization completes and the strategy still passes the Acceptance Criteria and it is better than the old one, the new one will be shown and added to the collection. It is a good idea to keep this option enabled//This is not full optimization. You can use the Optimizer for that.//
  
 **Normalize strategy** this option is //not// meant to make a strategy more profitable but it might help.  **Normalize strategy** this option is //not// meant to make a strategy more profitable but it might help. 
Line 142: Line 142:
  
  
-The default value here is "In Sample" which means that the Generator will use all the data you provided it with. However if you go "Out of Sample" and select a different option, when you press "Start" the generator will ignore the last X% percent of the bars where X is the percentage OOS you selected. The Generator will calculate a strategy without this end portion of bars. You can see the graph after the generator finished creating a strategy on the picture in the beginning of the article, pointed by the "Out of Sample Zone" bubble. The generator did not know anything about the data in the green area. Our generator works with a "real" OOS, which means the Generator literally has no clue about the data that being trimmed from the end of the chart while generating the strategy. +The default value here is "In Sample" which means that the Generator will use all the data you provided it with. However if you go "Out of Sample" and select a different option, when you press "Start" the generator will ignore the last X% percent of the bars where X is the percentage OOS you selected. The Generator will calculate a strategy without this end portion of bars. You can see the chart after the generator finished creating a strategy on the picture in the beginning of the article, pointed by the "Out of Sample Zone" bubble. The generator did not know anything about the data in the green area. Our generator works with a "real" OOS, which means the Generator literally has no clue about the data that being trimmed from the end of the chart while generating the strategy. 
  
 The main idea of OOS is to avoid //over optimization// or //curve fitting//. Curve fitting means that a strategy is fitting given data. So fitting that if you do a little change it will start experiencing losses instead of profits. So if you are generating a strategy without using OOS there is a chance you will experience unexpected behavior when trading it - new data comes in and the strategy is over-optimized for the old data. The main idea of OOS is to avoid //over optimization// or //curve fitting//. Curve fitting means that a strategy is fitting given data. So fitting that if you do a little change it will start experiencing losses instead of profits. So if you are generating a strategy without using OOS there is a chance you will experience unexpected behavior when trading it - new data comes in and the strategy is over-optimized for the old data.
  
 The OOS is basically simulating the trading. So for example you have designed the strategy using the first X% of the data and after that you trade the strategy for the remaining Y% of the data. This is also called "future test in the past" or "testing with unknown data". The OOS is basically simulating the trading. So for example you have designed the strategy using the first X% of the data and after that you trade the strategy for the remaining Y% of the data. This is also called "future test in the past" or "testing with unknown data".
-The goal of OOS is to achieve the most smoothly raising graph in the OOS Zone. If this is not the case the strategy is not fit to be traded. On the other hand even if the graph is smoothly raising this in no way guarantees the strategy will be profitable, but it will certainly have a much bigger chance to make profits than one where the graph is fluctuating a lot. +The goal of OOS is to achieve the most smoothly raising in the OOS Zone. If this is not the case the strategy is not fit to be traded. On the other hand even if the balance line is smoothly raising this in no way guarantees the strategy will be profitable, but it will certainly have a much bigger chance to make profits than one where the balance line is fluctuating a lot.
  
  
 ==== 1.5 Collection Name ==== ==== 1.5 Collection Name ====
 +
 This field allows you to specify a collection name. The collection is a virtual "folder" where the generator will add strategies while working. This is very useful for you because you do not need to always be checking the generator for new strategies. You can instead let the Generator work on its own. When generating a strategy the generator will show the best strategy but all the strategies that make it trough the [[fsbpro_guide:acceptance_criteria|acceptance criteria]] and showed some profit but are not as good are added to this collection and you can check them out later. You can see you collection in the [[fsbpro_guide:repository_tab|Repository]]. This field allows you to specify a collection name. The collection is a virtual "folder" where the generator will add strategies while working. This is very useful for you because you do not need to always be checking the generator for new strategies. You can instead let the Generator work on its own. When generating a strategy the generator will show the best strategy but all the strategies that make it trough the [[fsbpro_guide:acceptance_criteria|acceptance criteria]] and showed some profit but are not as good are added to this collection and you can check them out later. You can see you collection in the [[fsbpro_guide:repository_tab|Repository]].