About ChartScout

The Problem

You can't watch every chart. Not across 1,000+ pairs, multiple timeframes, 24 hours a day. Pattern detection sounds simple until you try to do it at scale, in real-time, across different exchanges with different data formats.

I wanted a crypto pattern scanner that would alert me when a pattern is forming. Real-time crypto pattern alerts, not delayed notifications that arrive after the move already happened. I had wanted this since 2017. In 2024, I finally decided to build it.

Who's Behind This

I'm Stjepan Ivanović, founder of ChartScout. I've been trading crypto since 2013. My first Bitcoin cost around $200. Four bull markets later, I'm still here.

Back in 2013, I registered on Mt.Gox, BTC-e, and Bitstamp. The only account still active from that era is Bitstamp. Mt.Gox collapsed. BTC-e got seized by the FBI in 2017. If you remember those exchanges, you know the kind of market we were dealing with back then.

Trading before 2017 was completely different. No USDT. Every trading pair was denominated in BTC. Namecoin/BTC, Litecoin/BTC, everything. All the volume was in Bitcoin. BTC-e was my go-to because they had USD pairs, which meant I could exit to fiat when I needed to.

I also traded on the first on-chain exchanges: IDEX, EtherDelta, ForkDelta. Centralized order books with on-chain settlement. Primitive by today's standards, but that was DeFi before anyone called it DeFi.

After BTC-e got shut down, I moved to Bittrex. A few months later Binance launched and I switched over. Been there since.

I've invested in around 70 crypto projects over the years. First investment was Monaco through Bittrex, which later rebranded to Crypto.com. Some worked out. Many didn't. That's how it goes.

My trading style started with scalping on 1-5 minute charts. Later combined it with swing trading. When you're serious about trading, it takes over your life. I was spending 12-15 hours a day watching charts. Many nights I would wake up twice just to check the markets. It becomes an obsession.

In 2021, I built a Python-based NFT trading bot for OpenSea. Did well with it. 2023-2024 were heavy on airdrops and meme coins. The market keeps shifting. What worked last cycle rarely works the same way in the next one.

The Team

ChartScout isn't a solo project. Behind it is Balkan Tech Solutions, our development company with a full engineering team in India and contractors across Croatia.

I handle the trading logic. Every pattern, every parameter, every edge case comes from years of watching charts. The team handles the engineering that makes it work at scale.

Why I Built This

Before starting ChartScout, I bought access to every competitor I could find. TrendSpider, altFINS, and others.

Some tools were too complex. TrendSpider has powerful features, but the learning curve is steep. I wanted something I could set up in minutes, not hours. Other tools like altFINS looked promising but didn't give me any actual edge when I used them in practice. The patterns they detected either came too late or weren't accurate enough to act on.

I couldn't find a tool that did what I needed: detect patterns early, across many pairs, with alerts fast enough to actually trade on. I was tired of waking up at night to check charts. I wanted a 24/7 crypto chart scanner that would watch for me and wake me up only when something was actually happening. So I decided to build it.

What It Actually Took

The original plan was to build this for myself. About six months in, I decided to make it public. That decision changed everything.

We're now 15 months into this project. The first pattern detection script alone took six months, from October 2024 to March 2025. Six months for a single pattern to work reliably on live markets.

So we went a different direction. The pattern detection itself uses manual logic combined with algorithms like RANSAC Regressor. For filtering and cleaning the data, we use ML models like SVM, Isolation Forest, and LOF. Thousands of test results analyzed, hundreds of manual parameter tweaks. Every adjustment suggested by me based on trading experience, executed by our team based on data.

Creating a script that works on one timeframe for one symbol is one thing. Creating a universal script that works across different timeframes, all symbols, and multiple exchanges with different data formats is a different problem entirely. We moved a mountain to solve it.