DEV Community

aymeric leon
aymeric leon

Posted on

Building a 100% Transparent Crypto Trading Bot Backtested Over 6 Years (2020-2026)

#python #algorithms #crypto #backend

Most crypto signal services online are pure marketing illusions: they show +500% leverage screenshots, delete losing trades, and sell fake promises.

As a developer and quantitative trader, I decided to build the exact opposite: a 100% transparent, automated crypto signal system backtested on 6 years of tick data (2020–2026) where every single trade—including every loss—is archived in real-time.

Here is the breakdown of how the architecture works, the quantitative engines, and the realistic expectations.

1. The Core Engines

Rather than relying on a single indicator, the system splits capital across 5 distinct quantitative engines:

  1. Funding Rate Carry (Market-Neutral): Captures perp funding rate differentials while holding spot. 84.2% win rate with an average net yield of +0.57% per trade.
  2. Relative Strength (Bullish Regime): Active only when BTC > 200-day MA. Scans 40 altcoins for top relative strength. 48.3% win rate with +2.83% expectancy per signal.
  3. 4H Momentum (Bearish Regime): Active only when BTC < 200-day MA. Ranks short-term 4-hour momentum to capture quick swings during choppy markets.
  4. Channel Breakouts & Volatility Expansion: Trend-following rules with strict volatility expansion triggers.

2. The Regime Filter (Why Silence Saved -70.9%)

The most important rule in quantitative trading is knowing when NOT to trade.

When Bitcoin trades below its 200-day moving average, directional breakout models get chopped to pieces. By enforcing a strict regime filter:

  • In 2022, the system remained completely silent during the market crash, avoiding a -70.9% drawdown.
  • In 2026, it saved -39.4% during market rangebound phases.

Silence is not a bug; it is a feature.

3. Radical Transparency & Real-Time Archiving

To eliminate trust issues:

  • No deleted posts: Every entry, stop-loss, and take-profit level is timestamped.
  • Fixed 3-day max hold: Every trade exits at 3 days regardless of noise, preventing bag-holding.
  • Realistic expectancy: 48.3% win rate for directional trades. Rentability comes from positive risk-reward asymmetry (cutting losses at -2% to -3% while letting wins run to +10%+).

Open Dashboard & Live Results

You can inspect the full 6-year backtests, live performance, and daily logs on our public dashboard:

👉 Live Site: crypto-signals-bot-site.signalytics.workers.dev

👉 Public Telegram Journal: t.me/ProSignauxPublic

Top comments (0)