What is lot size? Quantity step, minimums and contract units

Lot size is the set of rules a venue applies to order quantity for a symbol — a minimum, a maximum, and a step size that every quantity must be an exact multiple of — so a size your code computes has to be moved onto that grid before the order can be accepted.

Almost everyone meets it the same way: a position-sizing formula returns a number with too many decimals and the order is refused. The fix everyone learns is “round down”. This page is about what comes after that: which filter applies to which order type, how minimum quantity and minimum notional interact, what the quantity actually counts on different venues, and how to round in a way that also survives partial exits and closing orders.

The general case for rounding down, and for skipping a trade rather than rounding up to a minimum, is covered in how to calculate position size and symbol filters and minimum order sizes. This page assumes both.

Why this matters now

Three things make lot size harder than it looks:

• There is more than one quantity rule per symbol. On Binance spot, MARKET orders have a quantity filter of their own.
• Quantity is not always counted in coins. Some venues’ formulas are written in contracts, and the grid applies to whatever the order field counts.
• A quantity check and a value check both apply. A size can be a perfect multiple of the step and still fail because it is worth too little.

How it actually works

1. The venue publishes the quantity rules per symbol. On Binance spot these are the LOT_SIZE and MARKET_LOT_SIZE filters in GET /api/v3/exchangeInfo. Binance lists them among the per-symbol values that change without notice, so treat them as data, not constants.
2. LOT_SIZE is the symbol’s quantity rule; MARKET_LOT_SIZE is the same kind of rule for MARKET orders. Both reject a quantity that is too high, too low, or off the step.
3. Value checks run separately. MIN_NOTIONAL rejects an order whose price * quantity is too low; NOTIONAL rejects one outside a range with both a minimum and a maximum.
4. The rejection names the filter — on spot, as -1013 with a msg such as “Filter failure: LOT_SIZE”.
5. On USDⓈ-M futures, the same violations have their own codes. An off-step quantity is -4023 QTY_NOT_INCREASED_BY_STEP_SIZE, “Qty not increased by step size.” A too-small order is -4164 MIN_NOTIONAL: “Order’s notional must be no smaller than %s (unless you choose reduce only)”. The threshold is in the futures exchangeInfo filters.
6. Hyperliquid publishes size precision per market as szDecimals in its meta info response, with the rules on its tick-and-lot-size page, and rejects small orders with “Order must have minimum value of …” — the threshold is embedded in the string, so read it from the error-responses page rather than from any article, including this one.

LOT_SIZE vs MARKET_LOT_SIZE

The split causes a specific bug. A client places a limit order sized against LOT_SIZE, the order does not fill, and a fallback path converts it to a market order of the same quantity. That quantity was never checked against MARKET_LOT_SIZE, and the fallback — usually the urgent path, such as exiting — is the one that gets rejected with “Filter failure: MARKET_LOT_SIZE”.

Validate against the filter for the order type you are about to send, at the moment you decide the type, not when you first computed the size.

Minimum quantity vs minimum notional

Rule	Unit	Depends on price?	What fails it
Minimum quantity	Quantity units	No	A small size in any market
Step size	Quantity units	No	Too many decimals, or a non-multiple
Minimum notional	Money	Yes	A small size at a low price
Maximum notional (NOTIONAL)	Money	Yes	A large size at a high price

The notional rules are the ones that move. The same quantity can pass at one price and fail at another, so they must be checked at runtime against the order’s actual price, not validated once in a config file. For how notional is computed on different contract types, see what is notional value.

One exception is worth knowing. The futures -4164 message itself says reduce-only orders are exempt from the minimum notional. That matters when closing a small leftover position: sent as a reduce-only order, it is not blocked by the value floor. It still has to be on the step grid.

Contracts vs coins

The step applies to whatever the order’s quantity field counts, and that is not the same thing everywhere. What the venues’ own formulas show:

Venue	Published formula	What that tells you
Binance spot	Notional = price * quantity	Quantity is in units of the asset
Bybit USDT perpetual	Order value = qty × executed price	Quantity multiplies price directly
OKX USDT-margined swap	Fee = rate × (contracts × multiplier × contract size × fill price)	Position value is built from a number of contracts
Hyperliquid perps	Initial margin = position_size * mark_price / leverage	Size is in units of the asset, with per-market decimals

If a venue sizes orders in contracts, the conversion comes before the rounding: turn your coin quantity into a contract count using the instrument’s contract size and multiplier, then round the count down to the step. Rounding the coin quantity and converting afterwards can land between contracts. Inverse contracts add a further twist, covered in inverse vs linear contracts.

Rounding a computed size down, safely

from decimal import Decimal, ROUND_FLOOR

def order_qty(ideal: str, step: str, min_qty: str, max_qty: str,
              min_notional: str, price: str) -> str | None:
    s = Decimal(step)
    q = (Decimal(ideal) / s).to_integral_value(rounding=ROUND_FLOOR) * s
    cap = (Decimal(max_qty) / s).to_integral_value(rounding=ROUND_FLOOR) * s
    q = min(q, cap)
    if q < Decimal(min_qty) or q * Decimal(price) < Decimal(min_notional):
        return None  # does not fit the risk budget: skip, do not round up
    return format(q.normalize(), "f")

• Inputs are strings. Building decimals from floats imports the float’s error; the same reasoning is laid out for prices in what is tick size.
• Division by the step, then floor. This works for any step, not only powers of ten, and always moves toward less risk.
• The maximum is also floored. Capping at a maximum that is itself off the grid would reintroduce the error.
• None is an answer. A caller that receives no quantity should not place a trade; the position size calculator gives the ideal size, and this function decides whether it is expressible.

Partial exits and closing orders

Rounding down is safe for entries. For exits it needs one more rule: compute the last piece as a remainder, never independently.

Suppose you close half a position, then the rest. If each half is computed as “position ÷ 2, rounded down”, the two pieces can sum to less than the position, leaving a residue. If that residue is smaller than the step, it cannot be expressed as a valid quantity at all. Instead, round the first piece down and send the second as position − first piece. If the position was on the grid, the remainder is too, and the two orders close it exactly. The trade-offs of scaling out are in partial exits and expectancy.

The same reasoning applies to closing a position you did not size yourself: read the exact position quantity the venue reports and send that, rather than recomputing it from your own records.

Where the boundary should sit

• Sizing produces an ideal; the lot filter decides what is sendable. Keep these as two functions, and let the second one refuse.
• Check the filter that matches the order type, including on fallback paths.
• Check notional at the order’s price, at send time.
• Convert units before rounding on venues that count contracts.
• Size exits as remainders, and close positions from the venue’s reported quantity.
• Never let a model’s number reach the order field directly. A model asked for a size returns a plausible decimal with no knowledge of the step; it is an input to this routine, not its output.

FAQ

What is lot size in crypto trading?

Lot size is the set of per-symbol rules for order quantity: a minimum, a maximum and a step size the quantity must be a multiple of. On Binance spot they are published as the LOT_SIZE filter, with a separate MARKET_LOT_SIZE filter for market orders. A quantity that breaks them is rejected before it reaches the matching engine.

What is the difference between LOT_SIZE and MARKET_LOT_SIZE?

Both constrain quantity on Binance spot. LOT_SIZE is the symbol’s general quantity rule, and MARKET_LOT_SIZE is a separate rule for MARKET orders. A quantity checked only against LOT_SIZE is not guaranteed to pass MARKET_LOT_SIZE, so a client that falls back from a limit order to a market order must re-check the quantity against the market filter.

What is the difference between minimum quantity and minimum notional?

Minimum quantity is measured in units of the asset or contract and does not depend on price. Minimum notional is measured in money — price times quantity — so the same quantity can pass at one price and fail at another. Check both at send time, using the price the order will actually carry.

What does “Qty not increased by step size” mean on Binance futures?

It is error -4023 on USDⓈ-M futures: the order quantity is not an exact multiple of the symbol’s step size. Round the quantity down to the step using decimal arithmetic, and read the current step from the futures exchange information, since it is set per symbol and can change.

Related

• Symbol filters and minimum order sizes
• How to calculate position size
• What is tick size?
• Binance futures error -4164
• Hyperliquid order minimum value error
• Position size calculator