17 歲開發者花 2.5 年獨力打造免費一站式加密貨幣分析應用 AiTraderView
I'm 17 and built a free all-in-one crypto analysis app alone over 2.5 years
17 歲開發者花 2.5 年獨力打造免費加密貨幣分析網頁應用 AiTraderView，整合圖表、加密貨幣篩選器及投資組合工具。平台提供 50+ 項指標、AI Market Regime 評分、清算熱圖，以及加密貨幣新聞和 AI 生成的數據摘要，支援 59 種語言，並可在手機瀏覽器使用。它只作分析，不執行交易或提供財務建議。
It started as a small private script. I wanted to follow my own crypto trades without opening ten different websites, so I wrote something that pulled the data together for me. I'm 17, and over the past 2.5 years that script slowly grew into AiTraderView, a free web app that I built completely on my own, alongside school and everything else.
What really drove me was that I wanted to understand the movement behind the price. Existing tools showed me charts, but not why the price was moving, so I built my own.
What it is today
Three tools in one app:
- Crypto Cockpit: live charts with drawing tools, 50+ indicators, an AI Market Regime score and a liquidation heatmap
- Crypto Screener: scans 30+ coins on the 1h, 4h and 1d timeframes, so you can find setups without opening dozens of charts
- Crypto Portfolio: live prices, profit and loss, risk/reward and ETF flows
On top of that, I publish crypto news and AI-generated summaries of crypto data, so everything a trader needs is in one place.
The indicators
Most indicators on other platforms are built only on OHLCV data (open, high, low, close, volume). That makes them price-based, and they tend to lag. Many of mine combine several data sources for a coin and turn them into one overall picture with mathematical calculations. I also turned methods from well-known crypto trading books into mathematical formulas.
The stack
The backend is Python and Flask, the frontend is JavaScript, HTML and CSS, and the website runs on WordPress. I also set up and tuned the VPS server, made the videos and wrote the knowledge base myself. The app is available in 59 languages, and it works on mobile in the browser.
What it does not do
AiTraderView is an analysis tool only. It doesn't execute trades and it doesn't give financial advice. I want traders to make their own decisions with better information.
What I learned
Building this alone taught me more than any course could. These were the hardest parts:
- Writing the right prompts. I built a lot of AiTraderView with the help of Claude AI, and the hardest skill was figuring out how to explain exactly what I wanted so that Claude actually did it. Vague prompts gave vague results. Clear, specific ones saved me weeks. In my experience, ChatGPT and other AIs can't match the power of Claude.
- Learning the theory. I had to read and understand crypto trading theorybooks, and work out how indicators really work before I could build them as formulas.
- Finding the right data. Figuring out which data is available for a coin, and which of it is actually relevant for analysis, took a lot of trial and error.
- Tuning the VPS server. Keeping a live app fast and stable meant learning server tuning from scratch.
- Marketing. This was the hardest of all. Building the product was easier than getting anyone to look at it, because nobody takes a beginner seriously. That's one reason I'm sharing this story.
What's next
On my roadmap: price and indicator alerts, backtesting, 100+ indicators, more coins, a mobile app and optional trading with your own API keys, and I'm launching on Product Hunt today: https://www.producthunt.com/products/aitraderview
It's free to try at https://aitraderview.com. I'd love to hear what data you still have to look up elsewhere before making a trade.
來源：DEV Community · dev.to