The largest open-source crypto data initiative in the industry to date. The mission is to map and store financial and metadata into its atomic structure from the inception of the events to the future, storing a high resolution of events and financial data for greater data and finance applications. Atomic Granularity on crypto markets enables the new era of quantitative finance, financial engineering & HFT.
Using machine learning, Collider calculates high-quality metrics and complex insights, delivering them in real-time. It empowers users the opportunity to collect more superior insights, advanced metrics to build solutions for their data and execution of applications in both centralised and decentralised worlds.
The current inefficiency in the crypto space is data. Traditional market exchanges have an obligation to provide clean and reliable data, while crypto markets are heavily unregulated. This means that the data is unreliable.
Fusion is a direct follow up to our open crypto data initiative, L3 Atom. A milestone project which addressed the issue of unregulated crypto data. Project L3 Atom solved these issues and provides a next-generation data stream in terms of reliability, granularity, and complexity — Fusion builds on this by enabling the augmentation of data via machine learning and provides the tools to do so in a lucrative, open-source manner. While Atom alone does have the potential to achieve data supremacy in crypto trading by creating a foundation for quants, traders, and financial institutions to build various trade plans and strategies, this data can be further processed into meaningful, contextual information. This is the focus of Project Fusion, an open platform for the enhancement and utilization of crypto data.
The following is a brief explanatory video outlining the project’s scope and functionality:
In the current world of trading, traders’ perspectives on the market vary, each uses unique indicators to inform their decisions on-the-fly. Fusion is a multitudinous collection of modules, both user created and auto-generated, facilitating the next generation of crypto quant development and high frequency trading with its distinguished insights. Other modules produce various risk management metrics and insights for financial analysts.
While our L3 Atom project has addressed many of the issues in cryptocurrency relating to the availability, reliability, and granularity of data — further augmentation is required to transition this raw data into supreme information with which traders and investors can draw context and act accordingly to get ahead of the market.
Open-source contributors have little incentive beyond passion to contribute to their respective communities. While there are those who persist in creating with little reward, there are a multitude of outstanding developers who are simply unable to provide such services without remuneration. With monetary incentives, developers would be able to dedicate more of their time and energy into their projects, a prospect which is conducive with thriving software ecosystems.
Fusion is designed with advanced users in mind, but still facilitates traders and analysts who do not come from a programming background by providing a block-programming style, flowchart-based system for making use of modules.
Fusion is a multi-modular solution which utilizes machine learning and deep reinforcement learning to improve modules outputs, machine learning will be responsible for producing various indicators based on the patterns in the data and deep reinforcement learning will be used to find various advanced indicators that are dynamic, complex, and complicated to discover. Fusion/Collider would allow traders to gain access to higher complexity levels of data and indicators enabling faster and more certain. In an address to this, we have developed a service which allows a user to enhance data in such a way that meets their specific needs by either creating their own custom modules or installing modules created by others. These modules can be created and integrated with one another seamlessly within our integrated development environment (IDE).
One example of such a module is GDA’s own Automated Chart Pattern Recognition Infrastructure (ACPRI) — which you can read more about here. In short, the ACPRI module leverages machine learning techniques such as deep learning to automate the process of technical analysis, while also factoring in deeper insights such as liquidity. This module will be available within the Collider ecosystem and will be interoperable with future modules, allowing users to ‘plug-and-play’ this toolset with their own, and those created by other users to ensemble programs built from interconnected modules, while still allowing for deep customization beyond this building block system.
In Crypto, data is everything. A reliable, fast and qualitative raw data will lead to many more research and trading opportunities but having access to a high level of information extracted from raw data in a fast, reliable manner could open up more advanced use cases to a diverse set of professionals and academics.
Fusion runs on data streamed directly from our very own L3 Atom open data initiative, which addresses many of the existing issues of the crypto space. In essence, L3 Atom offers reliable, granular, and modular crypto data at speeds which facilitate complex use cases such as high frequency trading (HFT). The full whitepaper can be found here.
L3 Atom is the world’s first Open Crypto Data Initiative. As a data provider, this initiative will enable users to connect to an institutional-grade, reliable, and fast stream of financial data across a wide range of exchanges, including decentralized protocols. All the data collection, storage, and streaming will be handled by this initiative. The data that we provide will also be reliable and accurate at an unprecedented level in the crypto industry.
L3 will serve as the primary data source for the Fusion system and will provide basic level indicators which users can utilize for their own needs. These basic indicators include LOB data, financial market data and administrative data. Some advanced metrics will also be available from GDA such as AOPV, Long Short Ratio, Taker Volume, Contract/Index Basis and WebSocket error data. There is also some customisable data such as NFR, EOI, Fund Flow Ratio, ADL/Insurance Fund Health, GLP. These customisable feeds will depend on the users’ needs and they can choose whether or not they need these specific metrics.
Modules are the core of Fusion. Analogous to programming libraries and packages, modules are interoperable components serving their own independent functions, which can be plugged together to provide an ensemble of features depending on the needs of the user.
Modules in Fusion have capabilities which enable the next generation of crypto quant development and high frequency trading with its distinguished insights. Some modules are effective in producing various high level risk management metrics and insights for financial analysts. Basic, advanced traders and financial analysts can select a variety of tailored pre-built modules for their specific needs. With Fusion, developers have the opportunity to update, modify, and share modules that can further specialize the level of information according to the user’s needs. The main goal of project Fusion is to provide high level metrics and indicators such that users can have an insightful high-level context for their particular use cases.
Collider will serve as the brains for the Fusion infrastructure and will provide users with an opportunity to collect more superior metrics on top of the standard indicators which L3 provides. Collider takes data and creates general indicator modules which will be packages of metrics for general specific use cases. For example, different packages for trading, financial analytics and risk management will be available. In addition to this, users can harness the power of the lightweight machine-learning architecture to produce their own metric packages which can be used for their own personal analysis or trading strategies. Collider will also automatically iterate modules via evolutionary algorithms and deep reinforcement learning, improving the efficacy of metric modules over time such that they evolve alongside the ever-changing data and context involved.
In order to develop and use these modules, Reactor, a bespoke Integrated Development Environment (IDE) has been developed with both developers and end users in mind. Reactor allows developers to traditionally program modules with the full extent and depth of customizability and functionality of the Python language. While less technically inclined users may create and combine modules via our drag and drop block programming interface, similar to the modular programming interfaces in software tools such as Unreal Engine or Simulink.
Once created or generated, modules will be available from the marketplace. Developers can optionally monetize their creations either through direct purchase or revenue-based fees, though many modules will be made freely available in true open-source fashion.
The Fusion environment is integrated with our next-gen strategy testing environment, where users can deploy their strategies, metrics, and other creations into a virtual exchange, complete with an order book, user interface, users, API connectivity and other standard exchange features. Using this environment, users may test strategies and modules without the potential repercussions of testing new software with real money. More information on our back testing environment can be found here.
Also included within the Fusion environment is our data feed dashboard powered by OpenSearch Dashboards. This dashboard is custom designed for traders, with a heavy emphasis on quant workflow.
Fusion has various use cases with respect to diverse users. Advanced traders can utilize modules of Fusion to look at various complex metrics and indicators such as liquidity clusters to build efficient and time saving trade strategies. Primary traders can use image processing modules powered by deep learning like our Automatic Chart Pattern Recognition Infrastructure (ACPRI) that can identify and classify some complex trade patterns for technical analysis purposes. As retail traders are more interested in principal and basic data, Fusion also has some modules that can provide some basic predictions and visualizations of price movements. Retail traders can benefit from the readability, simplicity and understandability of the information and visualizations produced by Fusion.
Fusion also provides the chance and flexibility for the developers to create various tools such as advanced filters, trade strategies, unique metrics, trading bots. These Module developers can optionally monetize their creations by selling their modules or establishing systems by which they can receive percentage returns from investments or systems leveraging their work. Financial analysts can use components of Fusion to manage risk via various high-level indicators. Fusion has the capability to turn a basic and raw data into an advanced insightful dataset.
Project Fusion sets the stage for a new era of crypto. Leveraging our own L3 Atomic data, and the open-source community to develop and utilize modular metrics and programs will cultivate a community conducive to a prosperous revolution within the DeFi space. Stay tuned for more information as we continue along our development journey by subscribing to our Medium publication.