The latest articles on DEV Community by The_Bozgun (@bozgunber2506). https://dev.to/bozgunber2506 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3898825%2F1f5c6ef0-a699-45c4-8297-7defc4a20ea8.png https://dev.to/bozgunber2506 en The_Bozgun Sun, 26 Apr 2026 13:58:00 +0000 https://dev.to/bozgunber2506/how-i-built-a-real-time-crypto-arbitrage-dashboard-with-fastapi-websockets-20e https://dev.to/bozgunber2506/how-i-built-a-real-time-crypto-arbitrage-dashboard-with-fastapi-websockets-20e <h2> The Problem </h2> <p>Crypto arbitrage — buying on one exchange where the price is lower and selling on another where it's higher — sounds simple. The hard part is spotting the opportunity before it disappears. Prices move in milliseconds. Manual checking is useless.</p> <p>So I built <strong>EasyArbitHub</strong>: a real-time arbitrage intelligence dashboard that monitors 6 major exchanges simultaneously, calculates net profit margins with slippage factored in, and fires Telegram alerts when actionable opportunities arise.</p> <h2> The Stack </h2> <ul> <li> <strong>FastAPI</strong> — async REST API, fast enough to handle concurrent exchange polling</li> <li> <strong>WebSocket / polling</strong> — live price feeds from Binance, OKX, Bybit, KuCoin, Gate.io, Kraken via CCXT</li> <li> <strong>PostgreSQL</strong> — storing subscribers, alert history</li> <li> <strong>Docker</strong> — containerized, runs the same everywhere</li> <li> <strong>AWS EC2</strong> — production deployment</li> <li> <strong>Next.js</strong> — dashboard frontend with real-time chart updates</li> <li> <strong>Telegram Bot API</strong> — instant alerts to subscribers</li> </ul> <h2> How It Works </h2> <p>Every 15 seconds, the system fetches tickers from all 6 exchanges in parallel. For each asset, it finds the best buy and sell price across exchanges:<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight python"><code><span class="n">gross_pct</span> <span class="o">=</span> <span class="p">((</span><span class="n">best_sell</span> <span class="o">-</span> <span class="n">best_buy</span><span class="p">)</span> <span class="o">/</span> <span class="n">best_buy</span><span class="p">)</span> <span class="o">*</span> <span class="mi">100</span> <span class="n">net_pct</span> <span class="o">=</span> <span class="n">gross_pct</span> <span class="o">-</span> <span class="p">(</span><span class="n">buy_fee</span> <span class="o">+</span> <span class="n">sell_fee</span><span class="p">)</span> <span class="o">*</span> <span class="mi">100</span> </code></pre> </div> <p>Opportunities are scored based on spread size, liquidity depth, and slippage. High-scoring ones trigger Telegram notifications — even when the browser is closed.</p> <p>The Interesting Part: Slippage<br> Raw spread is misleading. A 0.5% spread on a thinly traded pair disappears the moment you place an order. So I factor in order book depth:</p> <p>slip = input_amount * 2 / order_depth<br> score -= min(15, slip * 200)</p> <p>This penalizes opportunities that look good on paper but can't handle real trade volume.</p> <p>What I Learned<br> CCXT is great but exchange APIs are inconsistent — always normalize before comparing<br> Telegram bots are easy to build and users love real-time alerts<br> Cooldown logic matters: without a 15-minute cooldown, the same opportunity triggers 10+ alerts<br> Try It<br> Live demo (UI in TR): easyarbithub.com</p> <p>I'm a freelance backend &amp; DevOps developer building production systems for SaaS and FinTech teams: thebozgun.com</p> python fastapi websocket devops