TradeAlmanac
Sign in

Iceberg order: the visible tip, the hidden reserve and the place in the queue

· 5 min · beginner

Automated material · TradeAlmanac editorial deskDraft prepared by a language model from our stored data; not reviewed by an editor.

89 views
Iceberg order: the visible tip, the hidden reserve and the place in the queue — Market infrastructure

An iceberg order is a limit order that shows only a small visible portion of its size in the order book, while most of the volume sits in a hidden reserve and is released in fresh portions as each previous one is filled. To a counterparty it looks like an ordinary limit order of modest size that, for some reason, does not disappear after a trade. An iceberg has the same price, type and priority as a limit order; only its size is hidden.

What the order book shows and what it does not

The order book is the set of limit orders aggregated by price level. A large order placed in full changes the picture at its level and becomes a signal: market participants see that a buyer or seller with substantial size is standing at that price and adjust their own quotes before the order has a chance to be filled. An iceberg breaks the link between volume and its display: the visible part sits at the level, while the rest is held in the exchange's or the broker's system and is not published in the order book.

The hidden reserve is not left hanging in the air — it is registered and executable. The difference from spoofing is fundamental: an iceberg intends to trade its entire size and simply does not show it; spoofing shows size that its owner never intends to trade. The first is a standard order-book tool; the second is market manipulation.

How execution works

The mechanics repeat in a cycle. The visible portion waits in the queue at its level and is filled according to the general priority rule: price first, then time. Once it has been taken out, the system replenishes it with the next portion from the reserve — and that new portion receives a new timestamp, meaning it joins the back of the queue at the same level. This is the key cost of an iceberg: an order placed in full keeps its place in the queue for its entire size, whereas an iceberg goes back behind everyone who arrived earlier each time it reloads. On a liquid stock such as SBER, losing priority costs little because the queue at a level turns over quickly; on a thinly traded instrument it can mean the order never gets filled at all.

The second limitation is that the hidden reserve does not improve the price and does not chase the market. If the price moves away from the level, the iceberg remains a limit order behind the market, with everything that implies — unlike a market order, which does get filled, but at whatever price is available. This trade-off is examined in our piece on market versus limit orders.

Why hide size at all

The reason is not secrecy for its own sake but market impact. Large size that becomes public moves the price against whoever revealed it: sellers pull their offers ahead of a visible buyer, and buyers pull their bids ahead of a visible seller. An iceberg reduces this component of trading costs, keeping the order in the book and preserving the chance of being filled at the limit rather than at the market price.

The price paid is what has already been described: losing queue position with every reload and taking longer to build a position. Individual trading venues treat hidden liquidity differently in their fee schedules — the platform's database holds no specific terms for Russian markets, and there is nothing to be gained by inventing them here; this is a question for the specification of the trading mode and for the broker's fee schedule.

How an iceberg differs from time slicing

An iceberg and algorithmic order slicing solve the same problem in different ways. An iceberg is a property of the order itself: it stands at a single price level and hides its size. An algorithm such as TWAP hides nothing — it breaks the instruction into a sequence of ordinary orders and spreads them over time, accepting whatever price emerges in each interval. An iceberg protects against revealing size but not against price movement; time slicing, by contrast, averages the price, but each of its slices is visible.

That is why in practice the two are often combined: the algorithm decides when and how much to send, and each child order goes out as an iceberg.

How icebergs are detected

Size cannot be hidden completely: trades are printed to the tape. The telltale sign is a stream of trades at a single level whose total exceeds everything that was ever visible in the order book at that level, while the level itself does not disappear. A second sign is a level that "refills" immediately after every trade. This is precisely why some venues randomise the size of the visible portion: a constant size gives an iceberg away faster than a variable one.

For a reader with a trading intention, this cuts both ways. A persistent buy-side iceberg below the price signals genuine demand that was not exhausted by the first trade — and breaking through such a level costs more than the order book suggests. But the conclusion that a "big player" is present is easy to draw wrongly: several independent participants with limit orders produce exactly the same pattern.

What an iceberg does not do

It does not guarantee that the full size will be filled — there is neither a completeness condition such as an all-or-none order nor protection against the price moving away. It does not replace a stop order and does not manage risk: it is an execution tool, not an exit tool. Nor does it make a participant invisible — it merely postpones the moment of disclosure, buying time at the cost of queue position.

The availability of icebergs differs across markets and trading modes, and venues use different names for the term itself. The glossary definition is in the entry "Iceberg order"; the other order types are collected in the glossary, and the list of instruments themselves is in the section on Russian market stocks.

Related instruments

Share
Was this useful?
How this material was prepared

Draft prepared by a language model from our stored data; not reviewed by an editor.

Model: claude-opus-5

How we use language models

Similar articles