Detecting Hidden Liquidity

An iceberg order shows a small "tip", say 100 shares, while a much larger reserve, invisible to the book, refills the tip automatically each time it's filled. A trader watching only the displayed size sees a level that never seems to run out, and that's the tell.

A worked example

The offer at 50.02 displays 100 shares. A trader sends a 100-share market buy, it fills, and the level goes to zero. One second later, the displayed size at 50.02 is back to 100. The trader buys another 100; it fills again, and again the level refills to 100 almost instantly.

A genuine, non-iceberg level of 100 shares would show zero after the first fill and stay at zero until some new trader chose to add fresh size, an event with no particular reason to happen within a second, repeatedly, at exactly 100 shares each time. An iceberg reveals itself through this pattern: constant displayed size, refilling immediately after every execution, usually in round tip-sized increments. After five such refills the trader has bought 500 shares at 50.02 without the displayed size ever suggesting more than 100 was available, the true reserve is at least 500, and probably more.

Where it's used. Liquidity-seeking execution algorithms probe suspected iceberg levels deliberately, sending small IOC orders to "ping" a price and see whether it refills, before committing a larger slice there. Market makers watch for icebergs on the far side of the book to judge true resistance, since a level that keeps absorbing size without moving is a much stronger signal than displayed depth alone suggests. And Order Book Features For Machine Learning pipelines sometimes engineer a "refill rate" feature per price level for exactly this reason.