Trading Bitcoin with a Hidden Markov Model (Experiment #4)
⚠️Warning: The HMM algorithm presented in this video can actually repaint when you use the "predict" method on a data interval. However, I leave the strategy active since, in any case, it is a dry-run test, but the performance might be lower. I have updated the GitHub repository to show an example where we check the repaint.
Discover in this video how to use hidden Markov models (HMM) to create a high-performance algorithmic trading strategy on Bitcoin ₿ and other cryptocurrencies Ξ! 💰
🧠 Inspired by one of the techniques used by Jim Simmons (founder of Renaissance Technologies) and an excellent article by FrenchQuant, this approach allows for the automatic identification of different market regimes:
📈 Bull market
📉 Bear market
↔️ Range market
💡 What makes this strategy exceptional:
Results that far surpass the simple buy & hold of Bitcoin ₿
Only one adjustable parameter (substantially reduced risk of overfitting), and the results remain excellent across a wide range of this parameter's values
Two available versions: long+short (derivatives) and long-only (spot)
A complete implementation via FreqTrade accessible to all
🔍 The long+short strategy generated an impressive return: turning $10,000 into over $100,000 during the 4-year test period!
🔮 This approach could also serve as a filter for other trading strategies by indicating the optimal market regime for each of them.
⚙️ All code is available on GitHub, and FreqTrade makes deployment from backtest to live trading quite simple. You can easily move from backtesting to paper trading and then to trading with real money in a few clicks.
CHAPTERS:
1:33 - HMM model?
2:41 - Implementation of HMM in Python
4:42 - HMM strategy for FreqTrade
9:59 - Backtests and results
12:58 - Conclusions
Useful links:
GitHub repository containing all the strategies presented on this channel:
https://github.com/djienne/YOUTUBE_ST...
Unfortunately, the FrenchQuant blog post is no longer available, as the site is down, but the link is:
https://frenchquant.com/comment-creer...
Tutorial video on deploying FreqTrade on a VPS server for live 24/7 trading:
• Déploiement bot de trading crypto sur AWS....
FrenchQuant's YouTube channel (a gem):
/ @frenchquant-jz5vf
Video from FrenchQuant's Channel on an HMM model applied to the CAC40:
• Modèles de Markov cachés : Application en ...
🚀 Trade on StandX – Spot & Perpetuals, 100% decentralized, no KYC, airdrop coming.
https://standx.com/referral?code=FREQ...
⚡Trade on Lighter – Spot & Perpetuals, 100% decentralized, no KYC, and ZERO fees – [https://app.lighter.xyz/?referral=FRE...](https://app.lighter.xyz/?referral=FRE...) (I’ll give you 100% kickback with this link)
🚀 Trade on Hyperliquid – Spot & Perpetuals, 100% decentralized, no KYC, and support the channel: https://app.hyperliquid.xyz/join/FREQ...
🔗 Get €20 in cryptocurrency by creating a Robinhood Europe account through my link and depositing €20 — while also supporting the channel: https://join.robinhood.com/eu_crypto/...
Affiliate link to sign up on Binance and support the channel:
https://accounts.binance.com/register...
Very good regulated platforms in Europe for long-term investment in crypto and Bitcoin:
Earn up to $100 in BORG tokens by signing up with SwissBorg:
https://join.swissborg.com/fr/r/david...
Sign up with BitPanda:
https://www.bitpanda.com/?ref=1474370...