Ed Downs Member  Posts: 32 Joined: 1/1/1900 Location: Austin, Texas  User Profile | I agree 100%. TRADEnosis should be discussed, and all TN users should have as much information as possible about our work to bring Cloud A.I. to our Nirvana Customers, because we aren't "done" yet. Let me share what I know... TRADEnosis & Nirvana First, Steve Mayo is an excellent scientist. I have been meeting with Steve and others in our "mastermind" group almost every Tuesday since 2014. You've seen Steve's work on the web site he presented at BASH (I believe) 2015 called "OmniVesting", where he did extensive work on the trade data coming out of OmniVest. I didn't ask him to do that - he just thought it would add value for everyone using OmniVest. And he built all of it himself. All of you watched the presentation and know the story, but Steve truly did a rigorous dive into the A.I. Systems that were available over the past few years to build this platform. When he brought it to me, I was impressed at the simulated trade statistics he was generating. The idea of bringing this to our customers was a good one, so we collaborated, and along with the intense efforts of our staff, to pulled the TN Signals into OT. Deploying the Beta As Steve explained in his presentation, the numbers he reported were the theoretical trade performance assuming every trade could be taken. Steve didn't have an OmniTrader-like system running in the Cloud, so he couldn't measure actual trade simulations. But as soon as the Signals could be delivered to OmniTrader, we could measure the historical signals using PortSim. And, they looked great. So we decided to share the results and provide licenses to TN at an introductory price. Once it was fully implemented in OT, we started trading the Signals by way of Collective 2. When performance was less than expected, we began tweaking the OmniTrader Signal process. For a while, our portfolio simulations seemed to be providing better results and a reasonable "Alpha". But, now that we have 8-9 months of actual, live market results, we understand that we aren't there yet. What affected TN's performance this year? A change in the market? High interest rates and persistent inflation - a scenario we haven't seen since the 80's? To answer this question, Steve has been running the QScore Signals through another A.I. process to understand how the scoring could have been affected. The TN Beta started as a live in-market project based on the work that Steve did over the past few years. But I feel that I must do what I can to ensure ultimate success. So, in addition to Steve's investigations the last few months, I have been doing my own research. Moving Forward Tonight, I am in Michigan because a family member suffered a massive stroke last week. I thank God that they appear to be gradually recovering. But I am also thankful for every one of you, who took the leap of faith and signed up for TRADEnosis. And I can tell you, we definitely aren't done. Getting a Cloud A.I. Solution is important, not only for those of us who want to trade with it, but for Nirvana Systems. Having worked with our own home-grown A.I. since 1996, I can tell you that we DO have an edge; we just need to exploit it. If my family situation had not happened, an update would have been sent out by now. I don't have the details ready to publish tonight, but I think you deserve some info on what's going on, and am happy to informally explain what Steve and I are doing. An Exciting New World The A.I. Model Steve used was a "Classification" model, just one of many models that are available. There are also Long-Short Term Memory NNs (LSTMs), Recurrent Neural Networks (RNN), Random Forests, and about 10 others, including Transformers - the A.I. model that ChatGPT is based on. Steve has been investigating these models, as well as the ML (Machine Learning) platforms that are available. While he does that, I have been running my own experiments using the "Codeless" A.I. platforms that sprung up this year. These platforms enable those who want to use A.I. to upload training data, have it trained, and then pull down answers. No coding required. I recently uploaded a test data set to one of them and was very excited about what I got back. It didn't have stock market data, but it was a difficult problem to solve using any other form of numerical analysis. Before I left for Michigan, I was preparing a data set with actual measurements from OmniTrader. As soon as I have the results, however they turn out, I will publish my update along with whatever Steve has discovered, and what we think comes next. I want to add that I do think we are still ahead of the pack. It's no secret that others out there are trying to use these models to solve the market. But they don't have what we have - nearly 30 years of experience with pattern matching and the indicators that make that happen. Update Coming.. Barry told me today that I needed to look at this thread and respond. I appreciated that, because I am not in the Forums every day. But I also appreciate all of you and the value you have added for everyone over the years - in the Forums and elsewhere. I know one of you used our Genetic Algorithm to get great returns last year. You also published your results. However, you added that you put in a lot of work in to make that happen, including hundreds of runs and using many Strategies in parallel. There had to be an easier way, and I firmly believe that Cloud A.I. is the way. I didn't want to send out an update with a lot of speculation. What Steve and I are working hard to get to is a viable path to augmenting TRADEnosis, perhaps with a new model or even a new approach - whatever it takes. But I totally agree that we need to share with you how we are advancing the ball. I will endeavor to check this thread periodically and respond. I ask that you give us this week to conclude this early look at the paths we are evaluating. At a MINIMUM, I will share the results I got last week in my experiment, and the path forward. Deploying a strong A.I. capability is crucial for Nirvana Systems. There's a saying by radio host Kim Komando, "Innovate or Evaporate." So true. Nirvana needs to produce a solid success in the Cloud A.I. arena. It goes without saying, but whatever we get to from here, your license will provide you with access to it. Thanks for hanging in there with us. Bottom Line: We will continue working the problem until we solve it, and I will make sure we keep you informed as we go down the path. Sincerely,  Ed Downs |
