I fell down the crypto rabbit hole — and came back with an arbitrage bot in my pocket 🧩
Hey! It’s been a while since my last post — the last one was all the way back in November 2024.
Today I want to share what I’ve been up to — where I disappeared, what I’ve been building, and just how deep I dove into the crypto underworld :)
Where it all started
At first, I wanted to get into crypto the “classic” way — technical analysis and traditional trading. Honestly, it felt overwhelming: levels, indicators, candlestick patterns — there was so much info my head was spinning. (Now it’s not that scary, and I’m slowly exploring that side too.)
But then something shifted. I randomly tracked a couple of wallets on-chain — turns out they belonged to guys I vaguely knew. They made insane gains on shitcoins on Solana: put in $100–200, cashed out over $10k just a couple of months later.
Maybe they just got lucky — I didn’t know their strategies. But I saw the raw facts: wallet data, trade dates, profits. And that hit me. Seeing results like that, clearly not just luck, was the trigger. I knew I had to dig deeper.
So instead of playing it safe with CEXes and charts, I dove straight into the real action: DeFi and Solana. Why Solana? 🚀 Blazing-fast transactions, dirt-cheap fees, and insane shitcoin activity — perfect conditions for running micro-trades and observing liquidity behavior in real time. Exactly the kind of playground I needed to explore fast and learn by doing.
I started with manual analysis
- I went through hundreds, if not thousands, of wallets on Solana using DexScreener, GMGN, GeckoTerminal, and other tools.
- I watched how people traded, what strategies they used, how they moved liquidity.
- I even intentionally jumped into obvious scam coins and shady strategies — not for profit, but to understand firsthand how those mechanics work.
This experiment was fully intentional — I treated any lost money as an investment in education.
That said, I never went in big — only with small test trades.
And honestly, it paid off 100%.
Then things got more interesting — I dove into automation.
- For copy trading on CEXs
- For copy trading on DEXs
- For algorithmic trading integrated with both DEX and CEX platforms like Binance, Bitget, WhiteBit, Raydium, and others
Copy trading turned out to be surprisingly useful: I found a few wallets that were consistently in the green. Set up a bot to follow them (with smaller amounts) — and ended up making between $100 and $650 a few times.
Algorithmic trading was a letdown. Backtests showed +100–200% gains, but in reality the numbers were much lower — and I had to constantly tweak the bot to fit market trends. Whether the market was trending up, down, or going sideways, it always needed adjustments. Otherwise, it started trading against the market and bleeding money. Hedging didn’t help much either.
So I hit pause on algo trading for a while.
But just copying trades wasn’t enough — I wanted to understand how these strategies worked under the hood.
And that’s when I came across a case that really caught my attention…
Crypto Arbitrage
I came across a wallet where every transaction was a chain of 2–3–4 swaps within a single atomic call.
Same token in and out — usually SOL or USDC — but the output was always slightly higher than the input.
Each chain earned just $0.10–$0.40 on average, but these ran 10–15 times an hour. I did the math: around $7–12/hour consistently, sometimes even $2–7 profit per single tx. Scaled to monthly volume — that’s $10–15k/month, fully on autopilot.
At first, I couldn’t understand how multiple swaps could happen in one transaction. Normally, when I swap tokens, it’s just a single route — and there’s no public UI that allows batching multiple swaps.
So I Googled, dug through forums, and annoyed ChatGPT with endless questions.
The answer: Multihop Swaps via CPI (cross-program invocation) using a custom smart contract.
That was the unlock. You can write a smart contract that executes a sequence of instructions — swaps, token mints, transfers, even LP creation — all bundled into a single atomic transaction. If the full route is profitable, it executes. If not, it fails and rolls back. Efficient and safe.
That’s when I realized — it’s time to build my own bot. 🚀
What’s next?
I tapped back into my dev roots—before stepping into Head of Product roles, I was coding full-time for 6–7 years. So I decided to dive back in.
I picked up some Rust to build smart contracts for Solana, and put together a custom bot in TypeScript with configurable logic and RPC proxying.
Right now, I have a working bot that connects to an existing smart contract and is already showing positive returns. For now, I’m using an external CPI-enabled contract for swaps—but I’m also building my own.
My custom contract already supports four of the most liquid pools on Solana: Raydium AMM, Raydium CPMM, Raydium CLMM, and Meteora DLMM.
- Full strategy autonomy and customizability using my own smart contract.
- Longer-term: building a full liquidity aggregator (think: a simplified Jupiter). I’ve already built a parser and data deserializer for Raydium in Go—the core logic works, and I just need to expand support to more DEXs.
📈 At the moment, everything runs on a local Mac using free RPCs—no infra costs. After three months of grinding and development, the bot finally started generating profit with virtually zero capital investment.
Key takeaways
- Understanding DeFi architecture and low-level transaction mechanics is a must if you’re serious about trading.
- Manual wallet analysis gives you insights you’ll never get from YouTube videos or public Discord channels.
- Multihop swaps via CPI are insanely powerful for arbitrage and micro-trading.
- Automation removes emotions and lets you test hypotheses way faster.
In upcoming posts, I’ll dive deeper: how I analyzed strategies, share screenshots of my results, what I ran into while building the bot, how it actually works, the challenges of writing a smart contract—and where this all goes next.
Still building. Still exploring.
Grateful for the journey—and excited to share it with you.