The AI Trade Just Gave a One-Week Course in Position Sizing — a Weekend Guide to the Rules It Taught

Last week, the AI trade gave a compressed lesson in why position sizing matters more than any forecast. On Monday, Nvidia's record $96 billion quarter sat on the tape and the AI-name complex looked unstoppable. By Thursday — three days later — the PHLX chip index had sold off 5.6 percent on a slowdown call from the very executives who buy the chips, then bounced hard the session after as dip-buyers stepped in. Record, rout, relief. Five days, three chapters. A portfolio sized for a smooth uptrend was forced to react to all three; a properly sized one treated all three as noise. That is the entire point of position sizing, and last week demonstrated it better than any textbook.

The uncomfortable truth of the move is that nothing about the underlying story changed. NVIDIA's demand is still real, the AI infrastructure buildout is still under contract, and the slowdown call was about pacing model development, not halting it. What changed was the price — violently, in both directions. That is precisely the scenario position sizing exists to absorb: when the story is unchanged but the price swings hard, sizing is the only variable you control. Here is the weekend guide to setting it up.

The four rules the week taught

Rule one: cap every single name at a size you can lose half of without a forced decision. A common retail ceiling is 5 to 10 percent of the portfolio per name. The test is simple — if your one AI stock is 15 percent of your money and it gaps down 6 percent higher, can you hold without panic? If the honest answer is no, the position is too big. Last week's 5.6 percent chip-index day is a permanent reminder that a single name in this complex can move that far in one session.

Rule two: cap any single theme, and AI counts as one theme. This is the rule that catches most retail investors, because their exposure is usually hidden. You might hold Nvidia directly, a semiconductor ETF in one account, and a broad tech fund in another — none of which looks concentrated on its own. Together they are one AI bet, and last week they all moved together. A reasonable theme cap is a level where a fifth loss would be uncomfortable but survivable. The point is that AI is a single risk event regardless of how many tickers you hold to express it.

Rule three: set your stop-loss as a fixed rule in advance, not as a reaction. The slowdown-call rout is the perfect example. If your stop was a written-down 15 percent below cost, then the 5.6 percent chip-index move either triggered it (and you acted by rule) or did not (and you held by rule). Either way you did not make the decision in the middle of a red day full of headlines, which is exactly when bad decisions get made.

Rule four: rebalance back to the caps on a schedule, not on emotion. The run to records is the quiet killer — a position that was 5 percent becomes 11 percent on its own, and the portfolio drifts out of its own rules without a single trade. The relief bounce is the natural moment to bring an outsized winner back to cap. A monthly or quarterly check that compares actual to target and trims the drift is how winners stop eating the portfolio.

How to run the rules across every brokerage in one sheet

Every rule above depends on one thing: seeing your true exposure across every account at once. That is exactly what InvestSheet removes — it syncs positions from Robinhood, Fidelity, Schwab, and 35+ other brokerages into a single Google Sheet, so each name and each theme totals across all of them with live formulas:

The workflow is three columns in one sheet. First, the value of each direct holding, from =IVS_BROKERAGE("value", "TICKER"). Second, the weight of the same tickers inside any fund you own, so a semiconductor ETF's Nvidia weight adds to the direct Nvidia line. Third, the total of each against your cap. The discipline of sizing only works if the sheet sees everything at once — the whole reason the AI theme is dangerous is that it hides across accounts, and the whole reason a single sheet fixes it is that it makes the hiding impossible.

The weekend is the right time to do this, because there is no tape. No flashing prices, no reason to rationalize a concentration because it went up today. The AI trade proved last week that it can move 5 percent either way in a session on headlines with no change in fundamentals. The investor who knows their true exposure in one sheet is the one who can hold through the record the rout, and the bounce unchanged — because the position was sized for all three from the start.

Frequently asked questions