RoninAi Updates: A Look Inside Cutting Edge Ai Developments

After our last Ai update (which you can read here), we received a number of excited emails that wanted to know even more details about RoninAi’s progress. As part of our regular updates from our product architects and the rest of the development team, here’s the latest on RoninAi development. Due to overwhelming demand, we have decided to carve out some extra time on a regular basis to deliver more updates through this news feed.

In this update we have news about several subsystems in the RoninAi neural network, Ai training success, and exciting front-end and back-end developments that are close to completion. We are currently on schedule in development, and our progress thus far is allowing us to free up time to take RoninAi well beyond and include features that separate RoninAi from the pack and make it in a crypto trading tool league of its own.

Data transfer subsystem for neural network

This neural network subsystem was mentioned as something coming through the pipeline in our last update article, and the final iteration was completed just hours after we published the previous update. This element is designed for data transfer between the user interface and the neural core in training mode. Training mode is where the neural core is learning how to interpret data and approximate a response. Since the team is feeding the neural core past data, we can check the accuracy of the neural network’s results. This error margin is then lowered through changes in coefficients to reach an optimal neural network.

Consultation subsystem for neural network

While the data transfer subsystem is used in training mode, the consultation subsystem concerns itself with the consultation mode, which is used by RoninAi’s trainers to evaluate the neural network’s efficiencies in crucial areas. This is an integral part of the system allowing for seamless reaction times of live data and forecasts of specific elements and comparisons with accurate PhD level understanding of the variables in the crypto trading market.

Time series data transmission mode for training and consultation modes

Perhaps one of the most integral parts of any deep learning system is not just accurate analysis, but the capability to accurately make predictions. A time series is a sequentially indexed representation of historical data that can be used to solve classification and segmentation problems, in addition to forecasting future values of numerical properties. This allows our Ai trainers to feed the Ai data and let it discover trends over various different time periods and lengths of time, offering considerably more levels of data which leads to more accurate predictive capabilities.


As opposed to classification or regression models, which only allow one objective field per model, time series models can produce forecasts for a set of variables. This is what will allow RoninAi to produce predictive analysis on a multiple indicator and signal types for the best results for our users. It also allows our Ai to differentiate between trends related to varying sources (seasonal, fundamental analysis, social trends, etc).

The team has made excellent progress on this since the last Ai update was released, and we are very confident of remaining on schedule moving forward. Our efficient progress so far has allowed us to ensure that we are testing and tuning the finer points of the product and not have a mad dash to release it. Here are some features that are currently being worked on by the team.

Curation of pre-configured neural networks

While the capabilities of RoninAi’s more complex neural networks are excellent for more advanced traders that are looking to gain a meaningful edge, simpler tuning and options are required for RoninAi to be as useful for the less experienced and advanced trader as it is for the elite traders. Several senior members of the team are currently curating a list of elements that will go into pre-designed mode that will offer a variant of the core architecture and topology of the RoninAi system. This is one central difference between our Investor and Pro Terminals.

Selection of user settings for curated pre-configured neural networks

While the more complex mode will have a significant number of settings and signal options, the curated mode will have a select number of these options. Once the pre-configured neural networks are completed and tuned up to our standards, the selection of user settings will be undertaken by members of the team.


Source data-based autotuning module

One of the more exciting modules being developed currently is the autotuning module, which will ensure that neural core is functioning properly and optimally over time. As conditions change, the neural networks will need to be tuned to remain up to date and suited for changing markets, the autotuning is able to do this on a regular basis allowing our Ai to adjust to any market situation. Remaining a tool unrivaled in precision and functionality over time is something that the RoninAi team is very committed to, and this module is a crucial piece of that puzzle.

As more features are completed and the team continues to develop and tune the finer points of RoninAi, we will continue to give updates, including more clarity surrounding the launch of the product, which we are very excited about.

Related article: What Type of Cryptocurrency Investor Are You? [Quick Guide]