Sizing Against Average Daily Volume

A stock trades $40m a day. That number tells you almost nothing about how big you can get, because "$40m traded" includes every market maker, every index rebalance and every other participant on both sides — and you are not entitled to be more than a small slice of it.

The rule of thumb, and why it is a rule of thumb

Desks commonly cap participation at 10–20% of average daily volume (ADV) for a position they expect to build or unwind over a single session, and lower — 5–10% — if the exit needs to happen without leaking information about intent. Above roughly 20% of ADV in a session, you are no longer a price taker; your own order starts to move the tape, and the price you model on your screen stops being the price you actually get.

The number is a rule of thumb because ADV itself is not stable. It is usually computed as a 20- or 30-day trailing average, which means it silently includes volume spikes from a merger rumor or an index reconstitution that will not repeat. A stock that "trades $40m a day" on a trailing average built mostly from two abnormal days might really trade $15m a day in ordinary conditions — and that is the number your exit needs to survive.

Worked example

You want to build a position in a name with 30-day ADV of $40m. Desk policy: 15% max participation to build, 10% to exit if you need to leave discreetly.

• Build ceiling: $0.15 \times 40{,}000{,}000 = 6{,}000{,}000$. You could build $6m in a single session at policy participation.
• But check the trailing average for spikes: excluding two days when the stock traded $140m on an index announcement, the other 28 days average $18m. That is the volume you will actually face on a normal day going forward.
• Recomputed exit ceiling on the honest number: $0.10 \times 18{,}000{,}000 = 1{,}800{,}000$.

If the risk-budget ceiling from your stop distance said you could hold $4m, the liquidity ceiling still caps you at $1.8m — less than half — because the risk-budget number does not know or care whether you can get out.

Where this bites

The failure mode is not building too much in one day — most execution systems will slow you down automatically. It is holding a position sized off headline ADV and then discovering, weeks later when you need to leave, that the volume behind that number was never really there. Recheck ADV for the name you are actually in whenever it has had a news event, and size the exit off the volume you would see on the worst plausible day, not the average one.