My review of StrategyQuant from strategyquant.com after 10 days of trial.
I stumble upon Strategy Quant software while looking for articles about optimizing strategies in NT7.
It wasn't what I was originally looking for but the software features really intrigued me.
Here's a quick description of what the software do from their own website.
"StrategyQuant is a powerful strategy development platform that uses machine learning techniques and genetic programming to automatically generate new automated trading systems for any market or timeframe"
Seeing that this software offers a free 2 weeks trial I decided to give it a go.
My first impression of the software was very good. The interface looks a bit dated but other than it's pretty intuitive and settings are all pretty much self explanatory. Anyone with experience with automated strategy development shouldn't have much trouble using this software.
Since I didn't want to miss a minute of the 14 days trial I decided to give it a go as soon as I received my trial key.
I wanted to test the software on ES and CL. Already I could see the first limitation of the software. SQ seems to be primarily developed for Forex traders. The only way to import futures historical data was to import continuous contracts. SQ doesn't have any notions of contract expiry and roll over. Anyway I decided to import a few years worth of 1 minute historical data for both CL and ES.
After my data was imported I decided to start SQ on ES first.
To do so you need to select your instrument, your timeframe as well as many other options, like which building block (indicators, conditional rules....) you want to include or exclude. (The list of settings is pretty big and I would recommend looking at SQ documentation to get a better understanding)
Once all set and nearly as soon as you press the start button you can start seeing strategies being generated and back tested.
After several hours of calculation I was expecting to be able to have a few strategies worth testing further. At least that's the theory.
Sadly after nearly 10 days of crunching numbers almost non stop for either CL or ES I couldn't find a strategy that will run successful in Ninjatrader on a previously untested time-frame. Worse, if tested on the same time-frame I'd often get very different equity curves.
I have also found out that Strategy Quant doesn't have a support email. Instead you are invited to post your questions in a forum. I posted questions related to historical data in their support forum on the day I started the trial and they have not yet been answered. Even worse after seeing my post unanswered for several days I decided to post a message in their pre-sale section (thinking I will get more attention) the message was simply deleted the next day.
In short here's what I liked and disliked about the product:
Easy to use UI
Lots of options
Lots of building blocks
Generated code (for NT at least) is very easy to read
Compatible with not only NT7 but also Metatrader and Tradestation (I only tested it with NT)
The application seems to be updated fairly regularly (a few times a year)
2 weeks free trial
Difference between SQ backtesting and NT backtesting (sometimes huge)
Too FX centric
Very small community
Getting a successful strategies is tougher than it looks (I personally couldn't succeed)
Extremely limited support (no support email, unanswered questions in the forums).
Would I recommend this program? Considering the lack of support I wouldn't. If my questions were answered on time maybe I would have been able to progress further and maybe come with some successful strategies.
That being said I'd definitely invite anyone who is curious to try the program for 2 weeks.
Hope you've found this short review helpful.
Last edited by Christophe; April 17th, 2015 at 08:53 AM.
The following 5 users say Thank You to Christophe for this post:
thank you for the review, it is valuable to get users opinions.
I have to admit that we were struggling with support recently, we are a small software company. But we are working on improving our support.
As for the differences in tests, the reason is usually data or settings differences, they can have big impact. But we will investigate the case if users send us a sample, it is our aim to make backtesting as reliable as possible.
Trying the program - I think it is a new approach to strategies development that most automatic traders here haven't tried yet, so it is interesting only from this point of view. And it is one of the way to build a successful trading, we have success stories among our users.
But it is not easy to succeed, as it is not easy to succeed in trading in general.
StrategyQuant is a tool that can save thousands of strategy development hours, but it is not a holy grail.
Disclaimer - I'm the main architect of the program, so I'm obviously biased
We have tried several programs that do claim to find trading systems for you. My company invested about 30K in related software. We have found out that Price Action Lab creates more robust systems and we have upgraded since to all new versions. Support is excellent and they have answered all questions to our satisfaction.
BTW StrategyQuant is almost the same with Adaptrade. Only minor differences that we could find. We are also in the process of testing tssb but no significant results are available yet.