Welcome to Nofinity

In this video, the creator tests 432 AI-driven trading strategies head-to-head to find the best one. A custom backtesting platform was built, pulling data from multiple sources (e.g., Alpaca, Binance). The goal was not only to identify the winner but to understand why it works – and how to adapt it for personal use.

• 432 strategy variants across different crypto assets (BTC, ETH, DOGE)
• Popular methods tested included:
  • EMA Cross
  • MACD Trend
  • SuperTrend
  • Bollinger Breakout
  • Volume Breakout
• Timeframe: 9 years (since 2017) on 4-hour candles
• Starting capital: $10,000

The winning strategy was a Donchian Channel Breakout on Ethereum (ETH).

How it works:

• Buy when the closing price exceeds the highest high of the last 38 candles (≈6.3 days)
• Take profit: +6%
• Stop loss: −3%
• Traded on 4-hour candles

Performance:

• Win rate: 41.3%
• Average win: +5.72%
• Average loss: −3.13%
• Total return: +24.9x (from $10,000 to over $544,000)
• Particularly strong in downtrends – crypto crashes are faster and cleaner than rallies

The creator emphasizes: Backtesting is not a silver bullet. He explains why many results are unreliable and how to avoid pitfalls.

1. Avoid Overfitting Bias

• Overfitting: A strategy is too closely tuned to past data – it works only on that data, not in the future.
• Solution: Out-of-sample testing – use 70% of data for training, 30% for testing. Consistent results across both indicate robustness.

2. Look Beyond Returns

• Max Drawdown: A 70% gain with 60% drawdown can be worse than 30% gain with 20% drawdown.
• Sharpe Ratio: Measures risk-adjusted return – higher is better.
• Trade frequency: Many small trades vs. few large ones – each strategy has a unique risk profile.

3. Discipline in Backtesting

• The platform is more pessimistic than TradingView: it accounts for larger spreads and worse execution.
• Small differences (0.1% per trade) compound significantly over 758 trades.
• No strategy is perfect – the best results come from combining AI with your own judgment.

The creator provides a GitHub repository with the backtesting engine code for free (link in description). With it, you can:

• Develop your own strategies using Claude or GPT
• Use historical data from Binance or Alpaca
• Compare and tweak strategies
• Export as PineScript and test in TradingView

Additional tips:

• Test across multiple asset classes: If a strategy works only on one asset, it may be overfit.
• Adjust position sizing: Use a volatility-based approach (e.g., GARCH model) to manage risk.
• Ask Claude/GPT for improvements: Upload the CSV data and let the AI analyze and optimize the strategy.

• AI + human judgment outperforms pure AI or pure manual trading.
• Backtesting is an essential tool to avoid trading blindly.
• The creator announces a $20,000 live challenge – the next step after theory.

Summary follows...

Summary follows...

Summary follows...

• Sentiment echoes a new bull market with strong upward price action, especially for Bitcoin, which is approaching the $80,000 region.
• The weekly trend reversal is confirmed as price has broken above the 200-day moving average.
• Caution is advised, however, as price enters a resistance zone between $80,000 and $82,000. A retest of the $60,000 area is possible.
• Analysts expect consolidation before a potential further move higher.

• Zcash (ZEC): Strong upward momentum; focus on breakouts after a sweep of a base/low. Potential price target: over $1,000.
• Pump.fun: Waiting for consolidation to find a long entry.
• ENA: Potential entry after a pullback into the $13-14 zone.
• Hyperliquid (HYPE): Waiting for a retest of the $70 level, then looking for a reclaim and continuation of the uptrend.
• Memecoins (Robin Hood Ecosystem): The rotation is extremely fast. Currently, Cash Cat is the favorite, but the strategy is to buy dips after a breakout.

• Caution: A pure uptrend is not guaranteed. The trader employs a long-term strategy (accumulation) combined with short-term momentum trades.
• Risk Management: Stop-loss is essential. Every trade is defined with a clear invalidation (e.g., a close below a previous low).
• Focus: Currently focusing on a small basket of 5-6 assets to avoid information overload.
• Portfolio Building: The best entries are during periods of calm and stability, not after a strong pump. Bitcoin is seen as a long-term asset (target: $500,000).

• Treasury Bond buybacks (Scott Bessant) are driving Bitcoin and Gold as non-yielding assets.
• Equities (stocks) are correcting, however, due to higher interest rate expectations and valuation issues.
• Bitcoin ETFs show strong inflows, supporting demand.
• MicroStrategy did not buy this week, suggesting a natural upswing without artificial demand.

• The market shows real strength, but entering a resistance zone requires discipline and stops.
• Momentum trades (breakouts) and dips can be profitable.
• Important: No blind buying at highs; wait for good entries with clear risk parameters.

Bitcoin surged 29% in one month, kissing the $80,000 mark. Many investors who were waiting for a drop to $40,000 are now caught in the "40K trap." This analysis dives into the psychology behind the paralysis and the on-chain data that tells a different story.

🔍 Key Points

• Anchoring Bias: Bears fixated on $40K as their target, unable to revise their thesis despite rising prices.
• Cognitive Dissonance: Publicly reversing a position feels like a character flaw, yet failing fast is a true sign of intelligence.
• On-Chain Reality:
  • ETF inflows nearing $2 billion in a single week – institutional smart money is buying aggressively.
  • Short liquidations at record highs; bears are being squeezed out.
  • Old hands to new hands: 81,000 BTC moved from long-term to short-term holders; 75% of short-term holders are now in profit.
  • Bitcoin Capitulation Index mostly green – the bottom is likely in.
• Market Structure Change: No traditional blow-off top; the cycle is shallower and shorter. Old superstitions (e.g., "October crash") no longer apply.

🧠 Psychological Traps

• Lizard Brain: Fear of being wrong leads to decision paralysis.
• Tribalism: Bears stick together, rejecting data that would exile them from their tribe.
• Paralysis by Analysis: Too much data that only confirms existing bias.

💡 Solutions

• DCA on Steroids: A rules-based model that tells you when and how much to buy – emotion-free. Example: Instead of 2 BTC via simple DCA, the model yields 3.1 BTC with higher ROI.
• Investor Profiler: A free survey that identifies your blind spots and matches you with a guru investor type.

📈 Outlook

• Bitcoin could reach $130,000 to $150,000 this cycle.
• Only 15 million BTC exist vs. 68 million millionaires – supply is scarce.
• Those who don't enter now risk watching others get rich for the next three years.

Bottom Line: Bears are trapped. Admitting you were wrong and pivoting fast is the smart move. The market has changed – old patterns no longer hold. 🚀

• The first pressurized espresso machine by Achille Gaggia (1940s) reached 8–10 bar – not due to scientific testing, but because of the ergonomics of the spring used.
• The E61 by Faema (1961) simply copied the peak pressure of lever machines. No sensory or scientific studies prove why 9 bar was chosen.
• Illy's book claims 9 bar emerged from "trial and error" and "customer satisfaction" – the author calls this reasoning highly questionable.

• Darcy's Law states: Higher pressure → higher flow. However, this does not apply linearly for espresso.
• Studies (Baldini & Petrecca, 1993) show: Flow becomes less efficient from 5–7 bar, not only at 9 bar.
• A new paper by Polish physicists (2026) proves: The most efficient flow occurs at 3–6 bar – depending on roast and grind size.
• 9 bar is far below the flow peak and sensorially almost indistinguishable from 7 or 10 bar.

• Porosity (void spaces) and permeability (ease of flow) change during extraction:
  • Fines migration clogs pores.
  • Erosion increases porosity at the puck's bottom.
  • Tortuosity (winding paths) and poroelasticity (elastic deformation) impede flow.
• These effects accumulate: Higher pressure compresses the puck further without improving flow.

• The author tested dozens of shots with different roasts and grinders:
  • Dark roast: Flow peak at 5–6 bar
  • Light roast: Peak at 3–4 bar
  • 9 bar was never near the peak.
• Crema forms already at 3 bar – 9 bar is unnecessary.
• Higher pressure (8–12 bar) significantly increases the risk of channeling, reducing consistency.

• Flow is the key criterion, not pressure.
• Recommendation: Try shots at 5–6 bar – they are often smoother, more consistent, and taste better.
• The 9-bar belief is a historical accident, not a scientifically founded standard.
• Author's quote: "The higher the pressure, the harder the shot becomes consistent."

• The video cites several scientific papers (e.g., Baldini & Petrecca 1993, the 2026 study).
• Open-source data from the Polish physicists are linked.
• Recommended reading: James‘ book about pressure and espresso.