What is a stop loss actually promising?

Every calculation in Unit 1 assumed one thing: that the loss really equals (entry − stop) × shares. That assumption is usually true and occasionally very false, and knowing exactly when it fails is the difference between a risk estimate and a risk fantasy.

What a stop order actually is

A stop order is a conditional instruction, not a price guarantee. It says: when the market trades at or through my trigger price, send an order. Two flavours, two different broken promises:

• Stop-market — the trigger releases a market order. You will almost certainly get filled; you have no control over the price. In a fast market that fill can be materially worse than the trigger. This is slippage.
• Stop-limit — the trigger releases a limit order. You control the price and may get no fill at all, which is the worst of both worlds: the trade went against you and you're still holding it.

There is no third option that gives you both. That's not a broker limitation; it's arithmetic about who bears the uncertainty.

The gap: where stops simply don't exist

A stop can only trigger when the market is trading. It cannot act overnight, over a weekend, or during a halt. When the market reopens at a new price, the stop fires at whatever is available there.

Worked example. 200 shares bought at $50.00, stop at $48.00. Planned risk: $2.00 × 200 = $400. The company misses on earnings after the close and the stock opens at $41.00. The stop triggers immediately and fills near the open.

• Realised loss: $9.00 × 200 = $1,800.
• That is 4.5× the planned risk — a "1% risk" trade that cost 4.5%.

Nobody did anything wrong. The stop worked exactly as designed. It simply had nothing to act on between $48.00 and $41.00, because no trades happened there — a band exactly like the shaded one below.

What this means for the sizing math

Your 1% is really "1% most of the time, and occasionally a multiple of that." Which leads to a distinct decision that sizing alone can't make: how much exposure to carry through events where prices can jump — earnings, scheduled economic releases, weekends in instruments that trade around the clock. That is a separate line in the plan, not a consequence of the stop.

Thin instruments make it worse: on a stock trading a few thousand shares a day, a stop-market can walk through several price levels before it finds enough size to fill.

In the data

The hole is two adjacent entries in the record: one session's close and the next session's open, with nothing between them. Here is Apple across the night of its July 2026 results, in five-minute bars, the finest the regular-session record keeps:

Every price between those two numbers appears nowhere, because the regular session never traded there, and finer bars do not fill it. A stop resting anywhere in that band had nothing to trigger on until the open. The gap is an absence in the record, not a fast move through it.

Try it now

1. Measure the band on the schematic above in dollars — previous close to next open — and put your finger on the prices at which a stop would have had nothing whatever to act on.
2. Now find a real one. In the five years below, look for a session whose open sits far outside the previous session's range, and Measure from the previous close to that open. Compute what a stop $2 below the prior close would actually have cost on a hypothetical 200 shares, and compare it with the planned $400.

3. Express the difference as a multiple of planned risk. Write the honest sentence: "my stop caps my loss usually, not always."