When you buy a stock on a gut feeling, somebody sells it to you. Who is that... and what are they seeing that you're not?

Daniel confesses a forty-second lunch-break trade, and Maya walks him across the table: into the probability machine that actually sets prices. How serious desks strip the story out of a chart and label the market's weather instead. Why trends really are sticky, and why the same math that proves it admits that a few weeks out, everything is a coin toss. Why the beautiful backtest on your trading app is a mirage, what testing looks like when it stops cheating, and why the most famous buy signal in finance got demoted to a permission slip. Then the strangest part: the whole apparatus is leaking down to ordinary people as open-source AI trading firms — researchers, skeptics, compliance officers and executors, all robots, all working for whoever sets them up. Which sounds like a money printer, until you meet the three-in-the-morning version.

By the end you'll know which game you're playing — the story or the probabilities — and the three disciplines worth stealing from the machines even if you never trade again.

CHAPTERS00:00 — Cold open: the forty-second sandwich trade02:42 — What this episode asks03:36 — Same screen, different game05:23 — The wet finger and the barometer06:43 — States: labeling the market's weather09:16 — The market has no memory12:00 — The ledger: counting what follows what13:14 — Trends are sticky (the uncle is right)14:07 — From odds to a position: signal and sizing15:43 — The coin-toss horizon17:03 — The arbitrary five percent18:26 — Hidden regimes: the babysitter21:18 — The strategy gauntlet, run naked24:03 — The momentum objection26:46 — The five-hundred-percent mirage27:37 — The exam with the answer sheet30:06 — Walk-forward: the wall down the middle of time31:20 — Shuffled timelines: sequence risk33:27 — Demoting the golden cross35:55 — The robot firm's org chart39:16 — The engine and the road40:32 — Sizing by math: the Kelly brake42:05 — Friction and the veto42:59 — The circuit breaker44:05 — The myth of hands-off48:21 — The boring answer that wins50:41 — When everyone owns the machine52:19 — Callback: the chip stock53:30 — The model to leave with

Core models:

• Two games, one field: Humans see stories; machines see measurable states.
• States, not stories: Classify regimes before predicting outcomes.
• No memory: Models care more about the current state than the path there.
• The ledger: Historical transitions become probabilities.
• Conviction is a number: Position size follows edge, not excitement.
• Coin-toss horizon: Forecast confidence decays with time.
• Let data define regimes: Avoid arbitrary human thresholds.
• The gauntlet: Test strategies across markets, regimes and decades.
• The mirage: Look-ahead bias and overfitting create fake edges.
• Test without cheating: Walk-forward and Monte Carlo expose fragility.
• Strategy ≠ system: Sizing, costs and drawdown rules keep you alive.
• Not hands-off: Humans still detect structural change and operational failure.
• The boring answer: Indexing wins often; quant thinking improves every bet.

Further reading

• Malkiel — A Random Walk Down Wall Street
• Taleb — Fooled by Randomness
• Duke — Thinking in Bets
• Poundstone — Fortune’s Formula
• Zuckerman — The Man Who Solved the Market
• Bogle — The Little Book of Common Sense Investing

Daniel Reed and Maya Chen are fictional hosts performed with synthetic voices. Each episode is produced with AI-assisted research, writing, and audio tools, under human editorial direction and review.