The Rules
$25,000 starting capital. Three AI signal sources: Trade Ideas Holly AI, TrendSpider's AI-detected patterns, and a custom sentiment analysis bot monitoring financial Twitter/X. Every trade must be triggered by at least one AI signal. No overriding the signals based on "gut feeling." Maximum risk: 1% of account per trade ($250). Duration: 30 trading days.
Week 1: The Honeymoon
Account balance: $25,000 → $26,340 (+5.4%). The AI signals were hitting. Holly identified a momentum breakout in PLTR that we caught for a 3.2% gain. TrendSpider flagged a bull flag in NVDA that played out perfectly. Confidence was high — maybe too high.
Week 2: Reality Check
Account balance: $26,340 → $25,180 (-4.4%). A string of false breakouts in choppy market conditions. The AI signals were technically correct — patterns formed — but the follow-through wasn't there. Three consecutive losers on Holly signals in a range-bound SPY market. The sentiment bot generated a buy signal on a stock that gapped up on news, only to reverse sharply. Risk management saved us: no single loss exceeded $250.
Week 3: Adaptation
Account balance: $25,180 → $26,890 (+6.8%). We noticed the AI signals worked best in trending markets and worst in choppy/range-bound conditions. Adding a market regime filter (only taking signals when SPY is above its 20 EMA and VIX is below 20) dramatically improved win rate. This wasn't in the original rules — it was a human insight applied to AI signals.
Week 4: The Grind
Account balance: $26,890 → $27,420 (+2.0%). Disciplined execution. Took 14 trades, won 9. The AI signals combined with our market regime filter produced a 64% win rate with a 1.5:1 reward-to-risk ratio. Not spectacular, but consistently profitable.
30-Day Results
| Metric | Result |
|---|---|
| Starting balance | $25,000 |
| Ending balance | $27,420 |
| Net return | +9.68% |
| Total trades | 47 |
| Win rate | 59.6% |
| Avg winner | +$412 |
| Avg loser | -$218 |
| Max drawdown | -$1,160 (4.4%) |
The Takeaway
AI signals provide genuine edge — but not a magic one. The raw AI signals without human filtering produced mediocre results (Week 2). AI signals combined with market regime analysis and disciplined risk management produced consistent profitability. The AI handles the pattern recognition; you handle the context, risk management, and psychological discipline. That combination works. Neither alone is sufficient.
