Hi all, I work as a developer for a small software company that has developing a new kind of Neural Network learning algorithm that that is being integrated into a strategy creation and testing application. The application is under construction based on the new Neural models and I was asked to get a feel for market interest and I was just wondering if the following software description sparked any interest? (Its not built yet so no I'm not trying to sell anything) "Basically it allows a user to build a trading strategy from scratch which will be automatically optimised (exit and entry) and backtested with statistical performance analysis and optional virtual forward testing on live data. Its a wizard based approach (i.e. you don't need to know anything at all about AI or stock market dynamics) you just pick suspected important technical or fundamental indicators and the system will do the rest, based on your established risk profile. A neural network that will scan the market in realtime for entry points is the result of the optimisation. These nets are highly complex and much more stable and effective than any competing products currently available." It is a real-time or EOD application and current suggested pricing model is monthly subscription (<$50 RT, <$30 EOD) So, having read the above, would you have any comments/opinions/suggestions about the product? Is this something that would interest you? If not, is there a particular reason? Is there something that could be added that would make you interested? Suggestions/opinions appreciated, Andy
Yes I agree with you, but what we have is much more advanced (by a good stretch) than the current offerings - we wouldn't be going into a saturated market unless we had something unique. I guess the only way to prove that will be at product launch but suffice to say that the AI system used to create a neural net for the user (which is a totally unique network for each user) is a major advance on current competition. Combined with the pricing model (no upfront cost) we think makes this an attractive offer. You cant get pure back-testing software (without strategy creation or AI) for less than the cost of this product so we think there is a definite advantage over the competition. I do accept the point that there are a lot of back-testing/market scanners/strategy building etc software in the market but not at this price with all these features in one... Andy
"optional virtual forward testing on live data." How you can forwardtest on live data? secondly, is this somthing similar to what safir x products for tradestation does based on fuzzy logig,..etc
The "virtual forward testing" means that you can tell the software to trade the system electronically going forward and provide you statistics on profitability. I guess it would be like a "virtual black box" system. So you can find out what profits you would have made over a time period without risking the capital (slippage, commissions and fill times are simulated based on the live data not simulated from historical data). So after strategy creation, the forward testing can be switched on and 2 months later a review of the trades, profits and losses can be undertaken to assess profitability versus expectation. Normal forward testing is also integrated btw... Thanks, Andy
As regards, the safir x question - this product would be using a completely new type of artificial learning technique which allows it to create much more complex solutions to various problems (in this case problems associated with the relationship between financial inputs and correct classification of entry points on the stock market, but it has also been used in other data spaces with high success rates such as medical and quantum mechanical predictive systems). The whole system is server-based instead of client-based so that a large rack of powerful servers does all the learning work instead of a users pc, meaning much more complex solutions are possible. The interface to this power is one of the key differences, in that it is wizard-based so you can have as much or as little knowledge of market dynamics or artificial intelligence and still use the strategy builder very effectively. If you dont want to see or tweak the advanced parameters you dont have to, you will just step by step point and click to create your strategy. The problem I had with trading software when I was trading was how difficult it was to start to use and understand, thats why this is built differently. Hope that has answered your questions, Thanks, Andy
Server based would mean to give proprietary strategies out of hand. I'm not sure if everybody is willing to do that.
I may be interested. How does it compare to Neuroshell Trader ? I also prefer to pay an upfront fee, rather than being locked into a longterm subscription that will get more expensive every year.
I don't feel very comfortable giving my proprietary system to be hosted in some server of a company I know nothing about. My answer will be: "Not likely" Good luck,