indicators_boost.png We have released new versions of our programs, which include several improvements, but the main update is the significant expansion of our technical indicators database. The number of indicators on our platform has more than doubled and now stands at around 160. With the help of AI, we have quickly entered the niche of leading global platforms, securing a strong competitive advantage in innovation and development. We have already matched the codebase for indicators with top platforms and are ready to significantly expand it in the near future, thanks to AI. This will allow us to offer even more tools for analysis and market operations. Previously, developing the first half of these indicators required considerable efforts from our users, who created them as part of open source projects. Their work laid the foundation for training the AI, which now generates new indicators fully adapted to our platform. During the process, it became clear that the AI initially had difficulties with our indicators API. This led us to revisit and simplify the internal code architecture, which not only helped the AI but also improved the overall flexibility of the system. Interestingly, the AI itself suggested optimal solutions, based on the best programming practices. The process of developing indicators is similar to what we described in our article about creating connectors using AI. We are happy to note that all indicators remain in open access, and we would greatly appreciate your feedback. You can leave your feedback in our chat or on GitHub to help us improve the product even further. Unfortunately, we’ve encountered some limitations in using AI tools. Claude is not available for public access or through an API, so we cannot provide these capabilities directly to our clients. However, we are hopeful that the upcoming GPT-5 will address the competition and give a much-needed \"kick\" to others in terms of speed of development and accessibility of technologies.
Or how AI can save you hours of work, but not from the necessity of verifying every step. claudeai-vs-chatgpt.jpg September 2024. We, the StockSharp team, are actively using AI to write connectors for crypto exchanges. But let me warn you — if you are reading this in 2025 or later, all of this might already be outdated. If you’re from the future, welcome to the past! And don’t forget to check if our methods are still relevant. Our journey with AI started with ChatGPT 3.5, which, frankly, could not write even a simple trading strategy, let alone a crypto exchange connector. However, with the arrival of ChatGPT 4.0 and Claude Sonnet 3.5, things drastically changed. Now, AI can write complex code modules, though with caveats: you still have to intervene, clarify, and fix bugs, which has become a normal part of our process. Step 1. Starting a project in Claude.ai Before we start writing a new connector, the first thing we do is create a project in Claude.ai. This is not just a chat that will forget everything once you close it. The project allows you to save everything you upload: code, documents, comments. It’s like Custom GPT, where AI \"learns\" from your examples and instructions, not just answering questions. project.png Project creation interface in Claude.ai. Here, all data and examples are saved, allowing you to maintain the context of the work. Claude.ai is like your personal developer, who understands some things, but without your supervision, can write something that might make your hair stand on end. So keep your documentation and vigilance close. Step 2. Copying existing code To avoid reinventing the wheel every time, we base our work on an already existing connector, for example, for Coinbase. We copy the project structure and adapt all the key classes and methods for the new exchange. This is just the beginning — the real fun starts now. Step 3. Adapting the WebSocket client Now it’s time to set up the WebSocket client for the new exchange. For example, let\u0027s take GateIO Spot WebSocket v4. Claude.ai helps generate code, but this is where our ongoing battle with inaccuracies begins. The AI often generates code that theoretically seems correct, but in practice, needs to be supplemented and fixed. 1.png Original WebSocket client code for one of the exchanges, generated by Claude.ai. The AI suggests expanding the data types and events. When it comes to authentication, AI often makes mistakes, and we have to manually add the missing elements. Claude can forget even basic things like authorization before subscribing to WebSocket channels. 2.png Claude.ai made a mistake in implementing WebSocket authentication. We corrected the code and added proper authentication before connecting. We also rewrite the code to make it more flexible by passing the WebSocket address as a parameter. This allows us to easily adapt to various exchanges and their APIs. 3.png Example code where the WebSocket address is passed as a parameter, making the architecture more flexible. Step 4. Setting up the REST client and adapter After setting up the WebSocket client, we move on to the REST client. Claude.ai generates basic requests, but we have to manually clarify whether they are correct. Each exchange\u0027s API differs, and you need to be careful not to miss important details in the documentation. Once the client is ready, we move on to the adapter for StockSharp, where the main magic happens — data processing and bringing them to a common format. Here, AI helps generate classes, but every step must be double-checked. This is especially important when working with JSON data transmitted via the exchange’s API. 5.png Generated domain model classes. We manually supplemented them with attributes and corrected the data types. Claude can offer interesting solutions, but you need to add attributes like JsonProperty to correctly process exchange data. If you don\u0027t, be prepared for surprises like incorrect data in reports. 6.png We added JsonProperty attributes for proper work with JSON data. Processing trade data is another important step where AI doesn’t always manage on its own. For example, trade data and their processing can lead to errors. 7.png Claude fixes code for processing trade data transmitted via WebSocket and REST API. Step 5. Optimization and new methods Claude.ai does a decent job of writing basic methods like working with candlesticks and trade data. However, its suggestions require optimization. We often refactor logic into separate classes for better structure. 8.png Optimized methods for working with candlesticks and trade data via WebSocket, refactored into separate classes. We also have to refine methods for working with trade operations via WebSocket. Even though AI helps automate such processes, the code still requires final polishing. 9.png Claude.ai added methods for trade operations in SocketClient, but they had to be optimized and corrected for errors. Step 6. Testing — the code doesn’t always work on the first try When the code is ready, it\u0027s time to test it. This is where AI comes to the rescue again, but all of its suggestions need to be thoroughly checked because things often don’t work correctly the first time. For example, when testing the exchange’s API, we encounter errors in instrument and position requests. 10.png Claude.ai suggested code for working with REST API Gate.io, but it needed to be tested and refined. Step 7. Data conversion for spot and derivatives Claude.ai helps generate classes for working with spot and derivatives, but again — be prepared to manually adjust methods and data types. For example, working with derivatives on Gate.io requires additional checks and adjustments. 11.png Claude.ai suggested the Extensions class to support data conversion between spot and derivatives on Gate.io. We also adapt SpotAdapter to correctly process exchange data using the conversion methods suggested by Claude. 12.png SpotAdapter was rewritten with changes in HttpClient and SocketClient, using methods from Extensions. Step 8. Futures adapter and error corrections Claude.ai helps write methods for working with futures, but as before, careful checking of logic and fixing errors is required. For example, when building the order book data, we encountered an issue with exception handling. 13.png Generated HttpClient for working with futures via REST API Gate.io. Claude also sometimes makes mistakes with data handling in the futures adapter. We have to manually adjust the methods for correct order book processing. 14.png Fixed futures adapter with correct order book processing logic. As a result, after corrections, AI proposed an optimized method for working with order books and data recovery. 15.png Optimized methods for data recovery and order book processing based on Claude.ai’s suggestions. GitHub Copilot — overhyped toy or real helper? GitHub Copilot, which Microsoft and GitHub actively promote as a tool of the future, is actually more of an interesting toy than a full-fledged developer assistant at its current stage. Copilot helps write needed sections of code, but expecting groundbreaking results from it is premature — its strength lies in refining and suggesting code improvements. However, its key advantage is its tight integration with the development environment, making it convenient for quick fixes and completing template code, which it handles much better than writing complex logic modules. copilot.gif AI: an enhancer for professionals and a barrier for beginners If you’re a beginner developer just diving into the world of programming, you’re, to put it mildly, out of luck. Currently, AI won’t boost your efficiency in any significant way. Moreover, trusting AI with complex tasks can get you even more tangled in code and problems it generates. It’s a different story for experienced programmers. For professionals, AI becomes a powerful tool that amplifies their capabilities: it speeds up development, offers solutions, and enriches knowledge of various technologies and libraries. In such a tandem, AI helps you focus on key project aspects, leaving the routine to it. However, as we have already mentioned, AI hits the weakest spot - beginner developers, making the gap between them and professionals even wider. Experienced programmers can quickly identify AI errors, correct them, and continue working, while a novice will simply drown in these problems. Ironically, at a time when AI was conceived as a help for everyone, it currently only increases this gap. But we should hope that in the future, AI will become more independent in programming, and then this gap may not only stop growing but disappear altogether, leveling the playing field for everyone, regardless of their level of expertise.
ai_integration.png Artificial intelligence is gradually permeating all spheres, including algorithmic trading. We plan to release various AI-based features, as this trend cannot be ignored. This article discusses how AI is developing in the field of algo trading and the changes it brings. Impact of AI on Algorithmic Trading 1. Reduced Need for Connector Developers With the development of AI, the need for connector developers is decreasing. Basic connectors, such as those for cryptocurrency exchanges, are easier and faster to create through AI with proper model fine-tuning. If AI learns to adapt in real-time, it can automatically track errors and protocol updates, promptly adjusting the code. Screenshot of the process, showing how a connector adapts to a changed crypto exchange protocol: b4794999143c21f9e62c2ba5e8828b46.png One of the key advantages of AI is the significant acceleration of development. In our company, we use AI to create connectors, which reduces their development time by several orders of magnitude. Previously, writing and testing a new connector could take several weeks, but now it takes only a few days. This demonstrates how AI can transform processes, increasing efficiency and reducing costs. We will write a separate article on our approach to developing connectors through AI. 2. Strategy Writing Almost all developers engaged in custom strategy writing may disappear as a class. AI can generate individual strategies faster and more accurately than a human. In algorithmic trading, strategy creators are often not professional programmers (which we understand, offering a solution based on {{entity:product:9}}). Therefore, replacing this direction with AI is already possible. Users will be able to receive personalized solutions tailored to their trading style and preferences. 25df797d550c1bb0caaa0ff99de54b37.png 3. Disappearance of Popular Open-Source Projects As AI develops, the need for popular open-source projects based on ready-made robots or strategies will decrease. It will be easier for people to ask AI to create a suitable solution for their needs and the software they are familiar with, rather than searching for a ready-made one and trying to adapt it. 4. Strategy Rental Services and Purchase of Ready-Made Robots Creating personal strategies through AI will become easier than purchasing ready-made solutions or renting strategies. Users will be able to generate their strategies considering their personal preferences and market specifics. 5. Level of Programming Knowledge and Its Necessity The question of the necessity of learning programming is becoming increasingly relevant. AI can already generate code better than many average programmers. This raises the question of why spend years learning when a machine can perform the same tasks faster and more efficiently. 6. Disappearance of Juniors in Algorithmic Trading With the development of AI, juniors (entry-level developers) in algorithmic trading may disappear. Simple tasks that were previously given to newcomers for training are now performed by AI. This makes entering the profession more challenging, and tasks for seniors are becoming increasingly complex. no_job.png The Future of AI in Algorithmic Trading In the future, AI is expected to become increasingly integrated into algorithmic trading, offering more advanced and tailored solutions for each user. Tracking protocol changes, writing unique strategies, and creating new connectors will all become commonplace. The role of humans in this process will be to control and adapt AI for specific needs and to solve complex tasks that remain beyond the capabilities of machines. ai_algorithmic_trading.png Conclusion In conclusion, it should be noted that the use of AI in programming opens up many opportunities but also requires adaptation and new approaches to human resource management. We are on the brink of a new era, and our future depends on how we can integrate these technologies into our daily work.