Welcome to NexusFi: the best trading community on the planet, with over 200,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- discounts are available after registering.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Does anyone have experience using AlgoWizard? It is a UI to create strategies and generate EasyLanguage code. I am trying to use it. It seems clean and straightforward. However I get odd, no, or inconsistent results. If it is good then I will invest more time. If the tool is not good than I do not want to waste more time.
I started using AlgoWizard when I took the free Strategy Lab course in mid July, so I don't have a lot of experience with it yet. For me there was a lot of learning to do before I figured out how to generate decent strategies. The main thing I found was to create the AlgoWizard template with just the entry and exit code, along with a RandomCondition in the triggers, and let SQX build the best strategies. It has been said that the daily timeframe is best to begin building strategies with, but I found that I can build a template with AlgoWizard and have it run with data on multiple timeframes, and that the 5m and 15m timeframes result in more nimble, higher-returning strategies (at least with @MES.D). I use the AlgoWizard templates with the custom projects to build, thoroughly backtest with out-of-sample data and rough market data, as well as validate with Monte Carlo Tests, and let them run at night, so after a couple of nights I have a large selection of robust strategies to pick through. Although I am still new to SQX, I have already built what I think are decent strategies that I would never have been able to find or backtest manually in order to have enough confidence to trade them.
Based on what you described above about the process, it seems their approach is use data-mining to curve-fit a strategy (and its parameter set) to the historical data, to come up with a decent looking equity curve. Then use Monte Carlo or out-of-sample result to confirm or validate the strategy. It does not matter, or does not start off with, any known logic for the strategy or indicator combination used. Is that how you would describe it, based on your experience? Thanks for any feedback.
That pretty much describes it. The strategies are built using a subset of historical data (in-sample) and then backtested using the rest of the historical data (out-of-sample) to make sure the strategies are robust. Multiple strategies are found that provide good results (I guess there really is more than one way to skin a cat) and then you can paper trade and eventually pick what you feel comfortable trading live. Robust strategies for one symbol could be used to trade other similar stocks/futures if they test out well.
Based on the posts from Ihale67, he seems to have found AlgoWizard working well for him so far. I wonder what you may have done differently from what he did, when you said the template was not working out for you. If it works, I would be interested as well. Thanks.
Visiting back this old thread ... it seems nowadays there is no need to learn how to use this tool to generate trading code for strategies. I can use ChatGPT (the paid version) to write codes for me. It does come back with some syntax errors sometimes, but usually easy to spot, fix and verify.
Trading: Primarily Energy but also a little Equities, Fixed Income, Metals, U308 and Crypto.
Frequency: Many times daily
Duration: Never
Posts: 5,241 since Dec 2013
Thanks Given: 4,585
Thanks Received: 10,531
You mean things like "If Close > Highest(High, 10) then ... "which of course is impossible. I've also had issues with it trying to use reserved words as variable names and reference functions that don't exist! Remember it's not actually thinking, it's just regurgitating, so I don't think it can give you something that hasn't been tried by and published by somebody else.
I don't expect any AI bot to invent anything for me. I have my own trading ideas, and then some; but I want to use the bot as a coding assistant to write initial trading codes for my ideas. I expect to have to debug and clean up the code myself. That can still save me time overall. Thanks for your feedback. Bill
You've identified the fundamental limitation - LLMs pattern-match rather than understand market logic. The "Close > Highest(High, 10)" impossibility is a perfect example of syntactically correct but logically flawed code generation.
Reserved word collisions and phantom functions are common pitfalls when AI tools train on mixed documentation versions. They excel at boilerplate structure but struggle with platform-specific nuances and edge cases that experienced developers catch instantly.
AlgoWizard's structured approach may actually produce more reliable code for complex strategies since it enforces logical constraints. ChatGPT works best for simple utility functions or when you already know the solution and need rapid prototyping.
Have you experimented with using ChatGPT to generate test cases for your strategies rather than the strategies themselves?
Please leave feedback here. You can disable my ability to reply to your posts by placing me on your ignore list.
Fi provides educational information on a best-effort basis only. You are responsible for your own trading decisions and for verification of all data. This message is not trading advice.
