Case: size three trades

When several ideas compete for one account, run each through the same steps and write the numbers down before looking at the chart again.

Nora's rules: risk at most 1% per trade, costs included (100 dollars on 10,000), and no single position worth more than 40% of the account (4,000 dollars).

1. Risk per unit: entry − stop + costs per unit, with the stop where the idea is wrong.
2. Units by risk: 100 ÷ risk per unit, rounded down.
3. Units by cap: 4,000 ÷ entry price, rounded down.
4. Take the smaller of the two.
5. Recheck: the loss at the stop, the costs, and the total for all open trades together.

The routine treats a boring fund and an exciting coin exactly alike. That's the point: excitement isn't a number in it, and charts have a way of making every idea feel bigger than its stop.

Spark Coin swings wildly, so the stop where Nora's idea is wrong sits far away: entry 2.40, stop 2.04, 15% lower. Add 0.06 of costs per coin and each coin risks 2.40 − 2.04 + 0.06 = 0.42.

By risk: 100 ÷ 0.42 = 238.1, rounded down to 238 coins. By cap: 4,000 ÷ 2.40 = 1,666 coins. The smaller is 238, so the risk limit binds. The position is worth only 238 × 2.40 = 571.20, and the loss at the stop is 238 × 0.42 = 99.96.

Nora was disappointed: 571 dollars in the idea she liked most. But excitement isn't in the formula. A wide stop means each coin carries a lot of risk, so fewer coins carry the same 100 dollars. That small position is the formula doing its job: the wilder the instrument, the less of it you hold.

Spark Coin also trades around the clock, so it can jump past its stop at any hour.

Nora's first plan put 3,000 dollars into each idea. It looks fair, but it is fair only in money invested, not in money at risk.

Spark Coin at 2.40 gives 3,000 ÷ 2.40 = 1,250 coins. Each risks 0.42 to the stop with costs, so the loss at the stop is 1,250 × 0.42 = 525 dollars: 5.25% of the account on one trade, more than five times her 1% rule.

The same 3,000 in Harbor 20 at 112.00 buys 26 shares; each risks 0.84, so the stop costs 26 × 0.84 = 21.84. In Northwind at 48.00 it buys 62 shares, risking about 62 × 1.55 = 96.10.

So the even split would have risked 525, 21.84 and 96.10 dollars: one trade carrying almost all the danger. Sizing by risk evens out what each idea can cost; the amounts invested then come out uneven, and that's fine.

Harbor 20 moves slowly, so the stop where the idea fails is close: entry 112.00, stop 111.20, costs 0.04. Each share risks 0.80 + 0.04 = 0.84.

By risk: 100 ÷ 0.84 = 119 shares. But 119 × 112.00 = 13,328, more than the whole account. By cap: 4,000 ÷ 112.00 = 35.7, rounded down to 35 shares, worth 3,920. The smaller is 35, so here the cap binds.

The risk then falls to 35 × 0.84 = 29.40, well under 1%. That's allowed: there is no prize for using the full 1% on every trade, and taking less risk is never a problem.

Why have a cap at all? Without it, close stops on slow instruments would push you to put most of the account, or borrowed money, into one position. A gap or a surprise would then hit far more than the stop suggests.

Costs are already inside the risk per unit, but it's worth seeing them on their own, as a share of what you hold.

Nora's three positions: Northwind 64 × 0.05 = 3.20 on 3,072 of value, about 0.1%. Harbor 20 35 × 0.04 = 1.40 on 3,920, about 0.04%. Spark Coin 238 × 0.06 = 14.28 on just 571.20: 14.28 ÷ 571.20 = 2.5%. Together 18.88, most of it on the smallest position.

So per dollar, the coin is by far the most expensive to trade, because its spread is wide relative to its price. The price must move 2.5% in her favor just to cover the round trip, before any gain. Module 1 showed how such costs eat small moves.

When you compare ideas, put the costs next to the value. A trade that looks the same on the chart can start much further behind.

Northwind is the ordinary idea: 1.55 a share, 64 shares, 99.20 at the stop. But the dataset's event column says it reports results before the next open. A report can make the price gap past the stop, and the stop then sells at the first price there is.

Work out the loss for a bad opening: (entry − opening price + costs) × units. If it opens 5% below the entry, at 45.60: 64 × (48.00 − 45.60 + 0.05) = 64 × 2.45 = 156.80. If it opens 8% lower, at 44.16: 64 × 3.89 = 248.96, two and a half times the plan.

What Nora can do: hold fewer shares into the report, enter only after the report is out, or knowingly accept the bigger possible loss. What doesn't help: moving the stop further away. The stop is the plan for an ordinary day; the size is the plan for a bad one.

"1% per trade" limits each trade, not the account. When several trades are open at once, a bad day can fill all their stops, and the losses add up.

Nora's three positions: 99.20 on Northwind + 29.40 on Harbor 20 + 99.96 on Spark Coin = 228.56, about 2.29% of 10,000 if all three stops fill. A gap on Northwind's report could make it more.

Trades also tend to go wrong together more often than you'd hope: a bad day for the whole market can push several of them to their stops at once.

So it's worth knowing the total before the bad day, not during it. Add up the loss at the stop for every open trade, and keep that sum in view whenever a new idea comes up.

Because risks add up, Marc keeps a second rule: all open trades together may risk at most 3% of the account. Nora's three trades use 228.56, about 2.29%, inside the limit before any gap.

A fourth idea risking a full 100 would bring the total to 328.56, that is 3.29%, over the limit. The fix is the same as always: a smaller size (at most 300 − 228.56 = 71.44 of risk), or wait until a trade closes. Never squeeze it in by moving stops.

How it fits together: every idea runs through one routine (risk per unit with costs, units by risk and by cap, the smaller wins), then the totals, costs and gaps get checked. Next module: can you lose often and still come out ahead?