Why people “freak out†when markets swing
Layer What happens Why it evolved / persists
Biology A sharp loss (portfolio -5 % in minutes) triggers the fight-or-flight cascade: the amygdala fires, adrenaline and cortisol flood the body, heart rate and blood pressure jump. Rational deliberation in the pre-frontal cortex is literally down-regulated—Daniel Goleman calls this an “amygdala hijack.†In the Pleistocene a rustle in the grass had to be treated as a lion; over-reacting was safer than under-reacting. The same circuitry now misfires on flashing red candles.
Psychology We are loss-averse: a 1-dollar loss hurts about twice as much as a 1-dollar gain pleases. Prospect-theory experiments show people will pay to avoid that pain, even if the bet is favourable. In groups, herd behaviour and social contagion magnify the fear—Twitter/Telegram turbo-charge it in crypto. The bias once helped us avoid rotten berries and follow the group to water; today it pushes traders into panic-selling loops.
Environment High-frequency newsfeeds, phone alerts, and 24-hour markets keep the threat signal alive. Volatility becomes ambient noise, so the stress system rarely resets, stoking chronic anxiety disorders.
⸻
Does an AI “freak out�
Aspect Humans AI trading system
Physiology / emotion Hormones, sweating, tunnel vision. None—just floating-point arithmetic; no cortisol receptor to flood.
Decision distortion Fear narrows attention, triggers snap judgments. No feelings, but objective-function distortion can act like panic: e.g., poorly tuned sell algorithms cascading into a “flash crash.â€
Failure modes Panic-sell everything, regret later. Feedback runaway (2010 Flash Crash) or reward hacking in reinforcement-learning agents—aggressive actions that maximise the coded goal while ignoring intent.
Recovery Requires time, coping skills, maybe CBT or meds. Requires human engineers to halt trading, patch code, or add circuit-breakers; the AI itself feels nothing about the episode.
Key point:
AI systems can behave in a way that resembles mass panic—dumping huge volumes in milliseconds—yet the driver is mechanical amplification of their rules, not fear. Humans experience a subjective state (anxiety, dread) that changes physiology and judgment; silicon agents do not.
⸻
Why the difference matters
• Predictability: Human freak-outs are irregular and context-rich; the same headline might spook some traders and excite others. AI reactions are fully determined by inputs plus code—once you know the rule set, you can test every scenario (though interactions among many bots can still create surprises).
• Mitigation:
• Humans calm down via position sizing, pre-commitment (“stop lossesâ€), mindfulness, or simply stepping away from the screen.
• AI is kept in check with guardrails—kill switches, circuit-breakers, adversarial testing, and reward-shaping so the optimisation can’t “hack†its own objective.
• Hybrid stacks: The most robust desks pair machine speed with human override. Humans set broad risk limits and narrative context; algorithms execute inside those fences, emotion-free.
⸻
Take-away for traders
1. Expect the surge: Your lizard brain will perceive sudden red ink as an existential threat—design your process (alerts, pre-planned exits) knowing that a cool head is unlikely in the moment.
2. Respect the machine: An emotionless bot can still trigger chaos if its design is naive or its objective mis-aligned—watch for liquidity holes and have circuit-breakers.
3. Blend the two: Use AI for fast, data-heavy tasks and reserve human judgment for regime changes, ethical calls, and meta-decisions about what the objective should be.
In short, humans freak out because millions of years of evolution wired us to over-react to danger; AIs don’t feel a thing, but can run amok if their math meets the wrong market micro-structure. Your edge comes from understanding—and bounding—both.