Search…
SingularityDAO’s Artificial Intelligence End of Year Updates
First Half 2022 Goals
Greetings Singularitarians,
As we come to the end of 2021, it seems fitting to provide an update on our AI development over the past few months and also look forward to what is to come in the early stages of 2022. The majority of 2021 has been about building the platform to host DynaSets, as well as the Beta of DynaSets themselves. While all of this has been going on, our AI team has been working hard preparing data pipelines, scraping years worth of data from all over the internet, designing tools to clean that data. The data then prepared for consumption by AI Algorithms and the training of our first AI market predictors.
While the Beta launching this week will use only offline signal generating algorithms which will be utilized by human traders, the data that is generated during these two months of testing will be invaluable in the training of the more advanced AI’s set to be autonomously deployed as part of the public launch of DynaSets in early 2022.
Critical to our success, the data engineering team have spent the past few months building infrastructure to collect vast amounts of information from all over the blockchain and beyond, every day so much data is gathered that they are currently upgrading systems to be able to handle 200gb+ of raw crypto data per day. Humans and Algorithms cannot directly understand raw data however and so the team is also working on “Data Transformation”
A good example of how this is currently done, is how a site like TradingView translates raw price data into candlestick charts, the types of which we all like to draw triangles on, and pretend that people know which way Bitcoin is going to go! The team is creating various indicators that can then be coded in such a way that they are able to be consumed by AI algorithms and/or Humans as needed, analysed and then used to output signals for our DAM to act upon. Outside of pricing data, other types of data also being collected are from Social Media like Twitter, reddit and Telegram, to name but a few.
Portfolio Management modelling is another central point of focus. and something that Ben Goertzel himself is quite heavily involved in, the goal being to create something of a “safe haven”, a place to park funds when market conditions are such that it would be unwise to execute risky trading strategies. We are very happy with the current results of this ongoing effort, but thorough testing and validation is of course required before anything can be implemented into DynaSets.
Let’s take a look at the AI Path, from what we have already started working on in 2021 to some of plans going in to 2022.
AI Path: End of 2021
- 1.Extensive data engineering: The team has been working hard alongside new partners SoftServe to create highly efficient and resilient data pipelines to provide the required information to our ‘in training’ AI algorithms
- Actively collecting the data of a significant part of the total crypto ecosystem
- CEX (order book data, trade executions for all primary exchanges)
- DEX (liquidity pools, trade executions for all major and many minor pairings)
- Additional Metrics (market positioning, market metrics, indexes)
2. Extensive work on data transformational tools / analysis
- Understanding the LOB and extracting consistent features for ML modelling and assisting decision-making within the trading team
- Research on novel ways of modelling trade execution data
3. However, Human-in-the-Loop (HIL) is the go-forward strategy: At least for the foreseeable future, Humans will be instrumental in overseeing the functionality of all our AI’s. Just like with self driving cars, Human-in-the-loop is a must have for the next few years
Beta State (December 15):
- 1.Large Scale ML of portfolios:
- SNET Collaboration: Price Prediction (Using the latest ML and HPC)
- Partners: Crypto market studies, and HPC vendor partnerships
2. Staffing up
- New Algorithmic-focused staff: We will continue to hire the best and brightest minds in the AI and Quantitative Analysis space
- Partners: Professional services in mathematics: Building advanced Artificial Intelligence takes a wide range of skills, while we continue to grow our team, we will also continue to bring on new partners with the skills and experience required to ensure we continue to move forward at pace
AI PATH: Early 2022
GA v1 State (Q1):
- 1.Trading Agents
- Poulin & DeepMind: Chris is working on a currently confidential project that will help us with AI based trading in the new year
- Trading Bots
- Internal + External Staff: With aid from certain strategic partners, work will begin on full automation of the platform, with the end goal that our human traders will eventually be able to sit back and watch the bots create Alpha day to day, while being ready to make the “Big Plays” that an AI agent does not see in the ‘big picture’.
2. Other Vendor Solutions
- External vendor assessments: What are the cause and effect relationships in crypto trades? When is an agent performing due to luck? These are things that rigorous mathematical analysis can answer. And for this, we partner with universities and affiliated research organizations.
- Enterprise Sales: We will begin to assess certain vendors and consider selling access to some of the tools we are building. Proceeds from this would go to furthering R&D as well as rewarding DAO members
Future State (Q2):
Integration with SNET AI Marketplace
- Continued integration and expansion of financial AI offerings on the SingularityNET Marketplace: SingularityDAO was created to be the DeFi arm of the SingularityNET Ecosystem, we have already begun creation of the first AI’s that will be hosted on the SingularityNET marketplace. We will continue to work on these and create more in preparation for AGIX on ADA Marketplace, at which point SDAO will be ready to start making calls to SNET AI Agents
Process Mining (“HIL” auditing)
- Internal + External Staff: With assistance from strategic partners, we will thoroughly stress-test, bug-test and security test BOTH our human traders and our Human-in-the-loop trading bots before full deployment into future versions of DynaSets.
Closing Thoughts
The above roadmap is a very high level overview of just a small part of what the AI team has been, and will be, working on in the coming months. As work progresses we will begin a deep dive series into all things AI, similar to the DynaSet Deep Dive that we have done these past few weeks. We know you are all incredibly excited to learn more and we also can’t wait to share it! Just as 2021 was an amazing year for the growth of SingularityDAO, 2022 is shaping up to be even more impressive! We look forward to you all accompanying us on this journey!
Stay up to date on all things ‘Artificial Intelligence’ and more on our social channels.
Copy link