TradingView is a brilliant product. I use it. Most serious traders use it. It has hundreds of indicators, script-based custom studies, and multi-timeframe layouts built in. So why did I spend several weeks building an entirely custom algorithmic terminal?
Because TradingView couldn't do what I wanted in the way I wanted it.
What I actually wanted
I wanted a system that:
- Evaluated 8 different signal dimensions simultaneously and produced a single weighted confidence score
- Filtered every short-timeframe signal through a 4H macro trend rule — automatically suppressing counter-trend entries
- Generated human-readable reasoning for the current indicator state on every single asset I was watching
- Connected to a live AI model (Gemini or Claude) to add commentary to that reasoning
- Ran in a browser tab on any device, with zero server I needed to maintain
TradingView's Pine Script can do parts of 1. None of 2 with automatic suppression. None of 3 dynamically. None of 4. And it's a hosted service — I can't run it self-hosted with my own data pipeline.
The real motivation: learning by building
There was a second motivation I won't pretend doesn't exist: I wanted to understand exactly how every indicator is calculated. Not at the conceptual level — I knew what RSI measured. But at the mathematical level. Why does Wilder use exponential smoothing specifically for RSI? What exactly is the "true range" in ATR, and what does accounting for candle gaps mean in practice? How is ADX actually derived from +DI and -DI?
The only way to really know is to implement it from scratch, verify the outputs against known-correct values (TradingView's indicator readings), and debug the discrepancies.
What I learned
Most JavaScript indicator implementations I found online were wrong. Not slightly off — wrong enough to produce meaningfully different signals from professional platforms. The most common mistake in RSI: using a simple moving average for the initial gain/loss smoothing instead of Wilder's proprietary smoothing method. This produces different RSI readings that diverge from TradingView, Bloomberg Terminal, and MT4/MT5.
Building the terminal forced me to get every one of these calculations right. And the result is a tool that produces indicator values that match professional-grade platforms, running entirely in the browser with React and no server-side computation.
---
Mustafa Kamal Hossain
Founder & Principal Engineer at Manfi. Passionate about Laravel, SaaS architecture, and high-performance engineering.