Iceberg and Hidden Orders: Why They Exist and What They Tell You

A trader wants to buy 500,000 shares but doesn’t want anyone to know. If they display the full order on the lit book, every market participant on earth sees it. The price will move against them. They get eaten alive.

So they post an iceberg. The order shows 1,000 shares at the touch. When 1,000 shares get filled, another 1,000 appear. The displayed size never changes. From the outside, it looks like a small, routine order. Only the trader posting it knows it’s a 500,000-share position working in the background.

Iceberg orders, also called “hidden orders” or “reserve orders”, are one of the oldest tricks in market microstructure. They’re everywhere. On liquid US equities, somewhere between 5% and 30% of displayed volume is iceberg refreshes, depending on the time of day and the stock. On European exchanges, the percentage is often higher because the displayed liquidity is thinner.

This article is about why icebergs exist, how to read them, and when they leak information. If you’ve ever wondered whether the “1,000 shares at the bid” you saw was really 1,000 or 500,000, the answer is in here.

Harris (2003, Ch. 20) discusses hidden orders as a response to the information problem: large orders signal information, and that information gets priced. Icebergs are the institutional trader’s defense against signaling. They work, until they don’t.

Why Icebergs Exist

The fundamental problem of a large order: if you display it, you move the market against yourself. This is true even if you don’t believe in market impact theory. Empirically, displayed large orders get worse fills than equally-sized orders worked in pieces.

The mechanism is straightforward. Suppose you display a buy order for 500,000 shares at the offer. Other market participants see this and infer that there’s a buyer of size. They buy ahead of you (front-running) or pull their offers (liquidity withdrawal). Either way, you pay more.

Harris (2003) puts it bluntly: “Information is costly to reveal.” An order on the book is a signal. The signal’s content is the trader’s intent. The market prices that intent immediately.

So the trader hides the order. Shows only a fraction of the total. Lets the market see a normal-sized order. Replenishes when filled. The hope is that the iceberg refreshes look like ordinary trading activity, not a single large position.

This works in liquid, fast-moving markets. It works less well in slow markets where any order attracts attention. It works less well when the same trader uses the same icebergs at the same time every day.

How Icebergs Work Mechanically

The exchange’s matching engine handles iceberg orders specifically. The trader submits an order with two parameters:

• Total quantity. The full size, e.g. 500,000 shares.
• Display quantity. The visible size, e.g. 1,000 shares.

When a marketable order hits the iceberg, the displayed quantity is filled. The exchange then immediately refreshes the displayed quantity (up to the remaining total) without re-entering the queue at the back. This is called replenishment from the front. The refreshed portion retains queue priority.

The exact replenishment behavior varies by exchange and by order type:

• Immediate refresh. New displayed quantity appears the instant the old is filled.
• Random refresh. Refresh happens at a randomized interval (used by some dark pools to obscure icebergs).
• Randomized display. Displayed quantity varies randomly within a range.
• Minimum fill size. Some venues let the trader set a minimum fill, so the iceberg only refreshes if a fill above a threshold size hits it.

The front-of-queue replenishment is the iceberg’s most misunderstood feature. A naive trader thinks: “I’m showing 1,000 shares, so when 1,000 get filled, I go to the back of the queue.” That’s wrong. The iceberg stays at the front.

This means a passive iceberg order is very sticky at the front of the queue. It will absorb fills before any other order at that price. Sophisticated traders exploit this by “sniffing” the queue. They send tiny orders to detect the iceberg’s presence.

Detecting Icebergs

You can detect icebergs through several methods:

Refresh pattern. If an order at the touch keeps getting filled and immediately replaced with the same size, that’s an iceberg. If the displayed size drops to zero after a fill and a new order appears seconds later at the back of the queue, that’s a regular order.

Trade-and-quote matching. If the trade tape shows a 1,000-share print at the bid, and the displayed bid is still 1,000 shares, the bid was almost certainly an iceberg that refreshed.

Order book patterns. Some iceberg detection algorithms use machine learning to classify orders based on their refresh behavior and timing. The features include time between fills, size consistency, and correlation with the trade tape.

Queue depth probing. Send a small marketable order to fill the displayed size, then observe what happens. If the same size reappears instantly at the same position, it’s an iceberg.

The probing technique is the most reliable but also the most informative to the iceberg’s owner. When you probe and the order refreshes, the owner sees the fill and the refresh. They might infer they’re being detected and pull the order.

Some venues explicitly forbid probing. Most don’t enforce it. The practice is in a gray area. It is not illegal but considered poor form by old-school traders.

What Iceberg Refreshes Tell You

When you see an iceberg refresh, what does it mean?

Buyer/seller of size is present. The basic signal. Someone is working a large order at this price.

They believe the price is right. An iceberg at the bid means the buyer thinks the bid is fair or undervalued. If they wanted a better price, they’d post deeper.

They’re patient. Icebergs don’t panic. They sit and wait.

They’ll likely continue. If the iceberg has been refreshing for 30 minutes without showing signs of slowing, the total size is likely large.

They might be informed. This is the cynical view. Large institutional icebergs are often driven by fundamental information. The trader knows something the market doesn’t. They’re hiding the order to avoid front-running while they build a position.

The informed-trader interpretation is overused. Most icebergs are simply large institutional orders, not informed flow. But if you see an iceberg in a name that doesn’t usually have hidden flow, and there’s been a recent news event, the informed interpretation becomes more plausible.

When Icebergs Leak Information

Icebergs are supposed to hide information. They do, mostly. But they leak in specific situations:

Persistent refreshes. If an iceberg refreshes 50 times at the same price, anyone watching the book knows there’s a large order there. The total size is roughly 50 × displayed size. Sophisticated traders can estimate the total size from refresh counts.

Refresh timing. If the iceberg refreshes at perfectly regular intervals, the owner is running an algo with predictable parameters. Other algos can exploit this.

Single venue. If the iceberg is only on NYSE and nowhere else, the owner has chosen NYSE for a reason, maybe queue priority, maybe fee structure. But the choice is observable, and a sophisticated counterparty can infer the owner’s intent.

Consistent size. If the displayed size is exactly 1,000 every time, the owner is using a fixed-parameter iceberg. A randomized display size would be harder to detect.

Same time every day. A trader who works the same iceberg at the same time every day is leaking information about their schedule. Other participants can predict when they’ll be active.

The fix for these leaks is randomization. Modern execution algos randomize display size, refresh timing, and venue. The randomization is bounded. Too much and the order becomes inefficient, but it’s there. If you’re a trader posting icebergs, vary your parameters. If you’re a trader trying to detect icebergs, look for lack of randomization as a signal.

Icebergs Across Asset Classes

Icebergs behave differently in different markets.

Equities. Standard on most exchanges. NYSE, Nasdaq, CBOE all support iceberg orders. Display quantity can be as low as 100 shares for some stocks. Dark pools offer similar functionality but without the queue priority.

Futures. Most futures exchanges support iceberg orders with similar mechanics. CME, ICE, Eurex all have reserve order types. The display sizes are typically larger than equities because futures contracts are larger notional.

FX. Icebergs are less common in FX because the market is decentralized. Electronic FX platforms (EBS, Reuters, Hotspot) support hidden orders but the convention is less established.

Options. Hidden orders exist but the structure is different. Market makers on options exchanges often post “synthetic” icebergs by combining displayed quotes with hidden reserve at the same price.

Crypto. Decentralized exchanges don’t have order books in the traditional sense. Some centralized exchanges offer iceberg orders; the rules vary by venue.

The principles transfer across asset classes but the specifics matter. If you’re trading equities icebergs, the rules from NYSE and Nasdaq apply. If you’re trading CME futures, different rules.

The Display Size Tradeoff

How much should you display? This is a real optimization problem.

If you display too much, you leak information. Other traders see the large order and trade ahead.

If you display too little, you don’t get filled. A 100-share display in a stock trading 1 million shares per minute will sit forever.

A rough heuristic: display 1-2% of average daily volume per fill. For a stock with ADV of 1 million shares, that’s 10,000-20,000 shares per refresh. For a less liquid stock with ADV of 100,000, that’s 1,000-2,000.

The optimal display size also depends on the spread. In a wide-spread stock, you want to display more (because each fill is meaningful). In a tight-spread stock, less display is fine.

The trade-off: more display = more fills but more leakage. Less display = less leakage but fewer fills. Most execution algos default to a display size that fills the order within 1.5-2x the scheduled horizon.

Icebergs and Queue Priority

The most counter-intuitive feature of iceberg orders is the front-of-queue replenishment. When the displayed portion fills, the new displayed portion takes the same queue position. This is enormously valuable for institutional traders.

Consider: an institutional trader wants to buy 500,000 shares. They post an iceberg of 1,000 displayed. They get filled for 1,000. The next 1,000 takes their old queue position, which is at the front. They keep getting filled before anyone else at that price.

This is why iceberg orders can fill at a rate much higher than their displayed size. A trader displaying 1,000 shares might actually fill 100,000 in a day because they retain queue priority on every refresh.

A regular (non-iceberg) order would have gone to the back of the queue after filling. The iceberg stays at the front. That’s a substantial execution advantage, and it’s why icebergs are so popular among institutional traders.

Some sophisticated traders exploit this. They detect an iceberg at the front of the queue, send a large marketable order to fill the displayed size, and immediately send another marketable order to fill the refreshed size before anyone else. This is called queue-jumping and is the dark side of iceberg priority.

Practical Trading Strategies Around Icebergs

If you see an iceberg you think is significant:

Don’t trade ahead of it. Trading ahead of an iceberg (front-running) is illegal in many jurisdictions and unethical everywhere. The iceberg will detect the front-running through adverse price moves and pull the order. You’ve reduced market quality and possibly violated rules.

Trade with it. If you have a similar interest (e.g., you also want to buy), you can post your order behind the iceberg and let it absorb fills first. This is passive co-trading.

Provide liquidity to it. If you’re a market maker, you can quote at the touch and get filled when the iceberg refreshes. You’re providing immediacy; the iceberg is taking it. Both sides benefit.

Avoid being adversely selected by it. If you’re providing liquidity and the iceberg keeps refreshing at the bid, the price is likely to drift down after the iceberg completes. Reduce your size or your quote price accordingly.

The general principle: icebergs are signals of latent demand or supply. Trade with them, not against them, when possible. The market quality is better when participants respect hidden liquidity.

The Dark Pool Variant

Dark pools use a different iceberg mechanism. Instead of queue priority, dark pool matching is usually based on size, time, and price priority, but with randomization to obscure the order.

When you post an iceberg in a dark pool:

• Your order isn’t displayed
• It rests in the dark pool’s internal book
• When a contra-side marketable order hits the pool, your order matches against it
• If your order is partially filled, the remaining portion rests in the pool

The advantage of dark pool icebergs: no information leakage from queue position. Other participants can’t probe or queue-jump because there’s no public queue.

The disadvantage: dark pool fills are often at inferior prices. The pool’s matching engine tries to give you mid-point pricing, but if your order is large enough to move the mid, you’ll get filled at progressively worse prices.

This is why most execution algos use both lit exchange icebergs and dark pool icebergs. They diversify the leakage risk. If the lit iceberg gets detected, the dark iceberg still has its full size. If the dark pool is toxic, the lit iceberg is unaffected.

The Information Content of Hidden Liquidity

There’s a strand of academic literature on the information content of hidden orders. The summary:

• Hidden orders are slightly more informed than displayed orders. This is intuitive: informed traders prefer to hide.
• Iceberg refreshes predict short-term price moves in the direction of the iceberg. A persistent bid-side iceberg predicts a price increase over the next 30-60 minutes.
• The information content is small (a few bps) but statistically significant.

For traders, this means: if you detect a large, persistent iceberg on one side of the book, the price is likely to move in that direction. You can trade with the iceberg (post on the same side) or fade it (bet on mean reversion after the iceberg completes). Both strategies have been studied; neither is consistently profitable after costs.

The honest read: icebergs contain information, but extracting it profitably is hard. Most academic strategies that work in backtests don’t survive transaction costs in production.

The Bottom Line

Iceberg orders exist because information leakage is expensive. They work by hiding the size of a large order while retaining queue priority on the displayed portion. They leak information when they refresh too consistently, when they’re posted in predictable patterns, or when they’re confined to one venue.

If you’re a trader posting icebergs: randomize your display size, refresh timing, and venue choice. Don’t post the same iceberg at the same time every day. Use a mix of lit and dark venues to diversify leakage risk.

If you’re a trader observing icebergs: respect them. They signal latent interest. The market is healthier when hidden liquidity is respected and not front-run. Exploiting icebergs through queue-jumping or front-running is short-term profitable but long-term corrosive to market quality.

Harris (2003, Ch. 20) is right that “the order book is a story.” Icebergs are the most intriguing chapters. They reveal only what the author wants you to see, but the gaps in the story are themselves informative.

References

• Harris, L. (2003). Trading and Exchanges: Market Microstructure for Practitioners. Oxford University Press. Ch. 20, 22. ISBN: 978-0-19-514470-3.
• Almgren, R., & Chriss, N. (2000). “Optimal Execution of Portfolio Transactions.” Journal of Risk, 3(2), 5-39.
• Bertsimas, D., & Lo, A. W. (1998). “Optimal Control of Execution Costs.” Journal of Financial Markets, 1(1), 1-50.