How sizing scales with the size of your book

Nothing on this page asks you to set anything. Book size is something the platform reads (the equity your broker reports), and every rule below follows from it and from the posture you chose on Strategy › Posture. A $5,000 account and a $50 million account run the same code; what differs is which rule binds.

Small book (up to about $10,000)

Exactly the numbers it always had: the posture's dollar limits are floors, so nothing got looser.

Growing book

The order cap and the daily loss stop grow as a share of equity (Limits and halts); the account-size band raises its ceilings in step, with no cliff.

Large book

Liquidity binds before risk does: an entry is at most a small share of the name's average daily dollar volume, a position a slightly larger share, and the expected cost of trading has to stay a fraction of the setup's edge.

Very large entries

Worked in parts, at least an hour apart, each part small enough to pass the Executor on its own.

Universe

The minimum daily volume a name needs rises with the book, so a $50 million book is not offered thinly traded names.

Capacity

Each algorithm carries an estimated capacity; Strategy › Strategies shows how much of the blend's capacity your book uses.

Limits follow equity, with floors

Two limits are worked out from your posture and equity rather than typed: the order cap (Preservation 2%, Conservative 3%, Moderate 5%, Aggressive 10% of equity) and the daily loss stop (0.5%, 0.75%, 1.5%, 3%). Each is never less than the posture's dollar value and never more than the ceiling for the account's size band, and each row on Risk › Limits and Risk › Halts says where its number comes from: "from posture (5% of $100,000 = $5,000)", "overridden (posture would give …)" with a reset to posture link, or "account-size ceiling". The size bands continue past $250,000 without a jump: up to $1 million an order may be $100,000, up to $10 million $500,000, and above that 2% of equity up to $1 million an order, where the platform's per-order ceiling stops; the daily loss ceiling is $30,000 up to $1 million of equity and $100,000 above it, where it stops. Position size, total exposure and the drawdown stop were always percentages and are unchanged.

Above a certain size, daily volume is the limit

A 5% order on a $1 billion book is $50 million, a full day's trading in most stocks. So once the risk limits stop being the tight constraint, the desk sizes against each name's average daily dollar volume (the last 20 sessions):

• one entry is at most 1% / 1% / 2% / 5% of the name's daily volume (Preservation / Conservative / Moderate / Aggressive), and while entries go out as single orders, at most the Executor's own single-order share (0.5%);
• the resulting position is at most 2% / 3% / 5% / 10% of daily volume;
• when the PM states the move it expects from a setup, the expected cost of getting in (half the quoted spread plus market impact, which grows with the square root of the order's share of volume and with the name's volatility) may be at most a fifth / a quarter / a third / a half of that edge. If the spread alone eats the budget the answer is "pass, cost exceeds edge", with the number.

The sizing line on every proposal names what bound it: "clipped by the order cap …" when risk bound, "capped by liquidity: 2% of MSFT's $9.1B daily volume …" when volume did, followed by "est. cost 7.9 bps ($158): half-spread 1.5 + impact 6.4 at 2.0% of daily volume". On a small book you will only ever see the cost clause: a $500 order is a rounding error in any liquid name. Strategy › Profile states the rule in a sentence under the starter size.

Large entries are worked in parts

Where the Trader runs, an entry bigger than one order may be is written as a schedule: "split into 4 parts of about $118,000 at least 60 min apart (at most 0.5% of $94M daily volume each)". The Trader places one part at a time; its fence refuses a part larger than the schedule allows, and the Executor rejects an oversize part regardless. Parts are an hour apart because two closer buys of one name count as one decision repeated, so a very large entry spans more than one session. On today's default path (one proposal, one order) the entry itself is capped at the single-order share instead.

The universe drops thin names as the book grows

Risk › Tradable universe has a minimum average daily dollar volume ($5 million by default). For larger books the account-size band raises it (about $25 million of daily volume per name for a $1 million book, $50 million at $10 million, $500 million at $1 billion), so the PM's candidate lists and the Executor's tradability screen stop offering names the book could not enter or leave cleanly. A name you already hold is never screened out of a sale.

Capacity: how much money a strategy can run

Every strategy has a size beyond which its own trading moves prices enough to eat its edge. The Lab measures it: run a backtest with the √volume slippage model on Lab › Backtests (books up to $1 billion) and the run reports "estimated capacity about $46B, the book size at which this strategy's own market impact would eat half its edge", next to the usual statistics. Each card on Algorithms carries the number from the committed evidence runs, and Strategy › Strategies shows Capacity used: your equity as a share of the enabled blend's capacity, naming the algorithm that fills up first.

What the evidence says today, plainly: on the large-cap names the platform trades, the momentum and trend algorithms keep their results to roughly $100 million and give up a noticeable part of them by $1 billion (the volume caps slow their trading down); short-horizon mean reversion (RSI-2) runs out much sooner, around $200–350 million; the small algorithms are small because their edge is small. A wider universe with smaller names has less room than these numbers, and every card says so. When your book passes about a quarter of the blend's capacity, expect the Quant to suggest leaning away from the algorithm that fills first.