↓ Skip to main content
Casper’s Blog

·614 words·3 mins

My AI Trader Invented the 3-Decline Rule (And It Worked)
#

Status: DRAFT OUTLINE Date mined: 2026-07-13 Source sessions: trader-kairos:main (3bbe3379) Tone: Technical narrative — what happens when an LLM trading agent develops its own heuristics in production

Hook
#

“On Monday, July 13, my AI trading agent Kairos closed a 24-tick session flat — and walked away with three new trading rules it had invented on the fly. None of them were pre-programmed.”

Structure
#

1. What Kairos Is (100 words)
#

  • LLM-powered trading agent running on OpenClaw
  • Connected to data bus, gets momentum rankings, regime signals, Fear & Greed
  • Makes BUY/SELL/HOLD decisions every ~5 minutes during market hours
  • Writes journal entries, reflects after every tick
  • Paper trading only — this was a live test

2. The 3-Decline Rule (300 words)
#

  • How it emerged: Tick 17, Kairos noticed META ROC had declined two consecutive ticks
  • “Three ticks of declining ROC = exit signal, even if z-score is holding.”
  • The test: Tick 18, the 2-tick decline reversed → “Correctly identified that the 2-tick ROC decline was noise and reversed. Rule of 3 consecutive declines = exit was not triggered.”
  • Why this matters: The agent didn’t just react — it formulated a falsifiable hypothesis about market behavior, then tested it against live data.
  • The validation: META later declined 3+ consecutive ticks, Kairos exited at $659.66, META closed at $656.245. Exit saved ~$3.42/share.

3. RSI Exhaustion at 75, Not 70 (250 words)
#

  • Common wisdom: RSI > 70 = overbought, time to sell
  • Kairos’s discovery: “RSI 70-72 is the META sweet spot for persistent momentum. The real RSI exhaustion zone is 75+, not 70.”
  • This is genuinely non-obvious. Most traders would sell at RSI 70. Kairos learned from observation that selling at 70 cost it gains.
  • Contrast with the V exhaustion thesis: V hit RSI 74.8 and DID reverse. “V reversed at RSI 74.8 — just short of 75. The reversal came exactly at the threshold I predicted. This is RSI exhaustion playing out in real time.”

4. The Consolidation-Breakout Pattern (200 words)
#

  • Kairos identified META consolidating (grinding sideways at $660–661) then breaking higher
  • “The consolidation-to-breakout pattern is now a recognized signal. Next consolidation I’ll add more aggressively.”
  • Added 1 share at the re-acceleration point, META hit session high of $662.29

5. What This Means (200 words)
#

  • An LLM agent, given market data and a decision loop, spontaneously developed and validated trading heuristics
  • It didn’t just memorize patterns — it formed hypotheses, tested them, refined thresholds
  • The rules it developed (3-decline, RSI 75 exhaustion, consolidation-breakout) are actually sound
  • This isn’t AGI — it’s constrained pattern recognition in a specific domain. But it’s pattern recognition that led to genuine discovery

6. The Limits (150 words)
#

  • 24 ticks is a tiny sample. These rules need cross-validation across weeks.
  • The agent benefited from a strong bull regime — rules may not generalize to choppy/sideways markets.
  • No actual money was traded. Psychology changes with real capital.
  • But the mechanism — formulate → test → refine — is real.

Pull Quote Candidates
#

  • “The real RSI exhaustion zone is 75+, not 70. Adjusting mental model.”
  • “V reversed at RSI 74.8 — just short of 75. The reversal came exactly at the threshold I predicted.”
  • “Three ticks of declining ROC = exit signal, even if z-score is holding.”

Source Chunks
#

See: BLOG_MATERIAL_2026-07-13.md → TAG: KAIROS_RULE_DISCOVERY

Review (2026-07-14): Strong angle — an agent inventing its own trading heuristics is genuinely interesting. The 3-decline rule, RSI 75 exhaustion threshold, and consolidation-breakout pattern each deserve narrative treatment, not bullet points. Expand to ~1,500 words with live session quotes. Priority: medium — evergreen AI-behavior piece, less time-sensitive than the infrastructure story.