Is Strategy Building Engine heavily biased on EURUSD generation ?
6件の返信
こんにちは,
I started to use STQ/X quite recently and I find it really a fantastic tool for Testing and Optimizing strategies. As far as Building new strategies is concerned, instead, I am still struggling a lot to find good strategies in Forex that are not based on EURUSD. Due to these difficulties, I have tried to run a bit more structured analysis to understand whether these difficulties could be related to the building config I used (which is the standard Forex breakout that you can find the App) or to the building engine itself, that has likely (I guess) been developed on EURUSD. To do so, I applied my building and testing workflow (which is highly based on the one that is described in the Algo trading course) to generate more than 30 million strategies with 2 different approaches:
1. Same config (the standard one I mentioned above) to different currencies
2. A randomly chosen config (changed every 5000 accepted raw strategies) to one of the currencies where I struggled more with standard config (USDCHF) The result (see attached xls) basically shows that:
a) The differences between EURUSD and the other currencies using the standard config are really significant. You can have a 9/1 raw generation efficiency ratio (EURUSD vs EURGBP) and more than 1000/1 ratio in the overall End2End efficiency (EURUSD vs USDCHF)!!!
b) With the randomly chosen config the result gets even worse than using the standard one (the average raw generation efficiency on USDCHF goes from 3.5% with the standard config to 2.1% with the random one; the E2E goes from 1100/1 to 1600/1) This may suggest that the root cause could be the building engine itself and not the configuration. Of course, this is not a scientific study, but I strongly suggest Strategy Quant to engineer and run a more comprehensive study about the subject, that is in my opinion of absolute interest for whoever wants to use the tool to effectively build new strategies.
In attachment you can find:
The workflow and all builder configs used The xls with the stats. I am also running a short study to evaluate the quality of the generated strategies. I will keep you posted!
0
こんにちは、,
what you show does not surprise me. Other users have similar experience (not completely the same though) with some forex pairs. Some better some worse but overall the forex market is what it is. Most traders focus on forex because it is very easy to start with. Tiny accounts, forex brokers, super leverage, you name it. But even if something is easy to begin with does not necessarily mean it is also the best option for algo-trading.
There are literally thousands of markets around the world, contracts and financial instruments that wait for exploration. Forex is a super-tine fraction of them and to me it seems to be the most difficult for algo-trading as well.
You can find very good strategies in a fraction of time required for those forex pairs for markets like indexes, stocks, certain commodities etc.
It is also important to mention that real power of SQX and the very best results comes with the trader’s mindset – trading ideas implemented by creating custom templates and applying kind of ‘idea first’ approach. This way you can create systems that cannot be simply generated automatically in the default regime
0
I agree, for the same reasons I moved to commodities and indices always in cfd, and finding strategies is easier. The forex market is very efficient
0
Thank you Tomas and all for your comments and kind suggestions. I will apply them all immediately.
Nevertheless, as a matter of fact, the Algo trading course is totally focused on Forex, so I believe that if it is not the optimal market to develop strategies automatically and/or using different templates can be crucial to succeed, you should be more clear about this and maybe dedicate a traning session to templates and other markets as well.
0
Hi Massimo, try to watch SQX webinar on youtube in the official channel. There are videos where they talk about GOLD and NASDAQ, maybe you can take a cue.
0
Thank you for your suggestion. I will definetely will !
0
6件の返信を表示中 - 1 - 6件目 (全6件中)