Open the depth window on a liquid instrument in the middle of the London session. The best bid shows a price and, next to it, a quantity. That quantity is the answer to a question most traders never ask: how much can actually be traded at the price on the screen. Below it sits the next level, at a slightly worse price, with its own quantity, and so on down the book.
An order smaller than the top quantity fills entirely at the top price. An order larger than it does not. It takes everything available at the top, then everything at the next level, then continues until it is complete. The platform reports one average price, which is worse than the quote that prompted the trade. That gap is the cost of size, and it grows faster than the size does.
Where the boundary sits
There is no fixed threshold. Depth varies by instrument, by venue and above all by time of day. A major FX pair during the London and New York overlap can absorb a large order with almost no visible degradation. The same pair at 22:00 server time, around the daily rollover, has a fraction of that depth, and an order that filled cleanly twelve hours earlier now walks through several levels. The same holds for indices outside cash-market hours and for gold in the Asian session.
Scheduled events compress it further. In the seconds around a central bank decision, providers reduce quoted size before they widen the spread, so the first symptom of a thin book is not a wider price but a smaller one available at that price. A trader watching only the bid and ask sees nothing wrong until the fill report arrives. This is why depth of market is worth having on screen for anyone trading meaningful size, even if it never informs a single entry decision.
What each order type promises
Order types exist to let you choose which certainty you want, and every choice gives one up. A market order guarantees execution and accepts whatever prices the book contains, which is how a large market order ends up with a poor average. A limit order guarantees the price and accepts that part or all of it may never fill. Between them sit the instructions that address partial fills directly: fill-or-kill cancels the whole order unless it can be completed immediately, and all-or-none holds it until the full size is available at the limit.
Platform support varies and the OTC world is less uniform than exchange trading here, so it is worth confirming what your platform actually does with these instructions rather than assuming. The full set is covered in order types explained. What matters is the decision behind them: an arbitrage or hedging trade that only works complete should refuse a partial, and a discretionary directional trade usually should not, because half a position at a good price beats no position.
A partial fill leaves you with a live position and an unfilled remainder. Decide in advance what happens to the remainder, because leaving it working while price runs away is how a planned entry becomes an unplanned average-up.
The arithmetic of walking the book
Consider an order that takes three levels: a third at the quote, a third one tick worse, a third two ticks worse. The average is one tick worse than the screen price. On a scalping strategy targeting five ticks, that single tick is twenty per cent of the gross target, paid before the trade has done anything. On a swing position held for two hundred ticks it is noise. Size cost is not absolute, it is relative to the edge the strategy is trying to capture, which is why the same execution quality can be fine for one trader and fatal for another.
The other half of the arithmetic is the exit. Traders model entry cost and forget that closing a position pays the same tax, often in worse conditions, because exits cluster around moves that thinned the book in the first place. A stop loss on a large position is a market order sent into exactly the moment when depth has disappeared. That interaction, not the stop level itself, produces most of the extreme fills people report, and it connects directly to how slippage works.
Splitting orders and what it costs
The standard institutional response is to break the order into pieces and work them over time, either by hand or through an execution algorithm that targets a benchmark such as a time-weighted or volume-weighted average price. This reduces the immediate impact because each child order is small enough to sit inside available depth.
It replaces impact cost with timing risk. Spread the order over twenty minutes and you get better individual prices, and you also get whatever the market did during those twenty minutes. If the move you were trying to catch happens in the first three, the patient execution captured a worse entry than the aggressive one would have. There is no version of this where you avoid both costs, and choosing which to pay is a function of how time-sensitive the idea is.
For retail-sized accounts the honest conclusion is usually simpler: trade a size the market absorbs comfortably at the time you intend to trade. That is a position sizing constraint sitting on top of the risk-based one, and it binds earlier than most traders expect on exotic pairs and minor indices.
What firms have to build for it
On the operator side, partial fills are a reporting problem as much as an execution one. A position built from four fills at four prices has to be recorded as one position with a weighted average, the commission has to be charged once against the total rather than per fragment, and the trader's statement has to show something they can reconcile. Systems that get this wrong produce support tickets that look like execution complaints but are accounting bugs.
Prop firms carry an extra version of the problem. Drawdown and consistency rules are evaluated against filled positions, so a rule engine that treats each fragment as a separate trade will miscount a trader's activity. Anyone designing those checks should read how a prop firm rules engine works before defining what counts as a trade, because the definition has to survive partial fills, partial closes and re-entries within the same minute.
Leveraged trading carries a high risk of loss, and size is the variable that turns a manageable execution cost into an unmanageable one. The book tells you what it will absorb if you look at it before the order rather than after.
"Traders scale up their size and keep the same expectations about fills. The strategy that worked at one lot has a different cost structure at twenty, and nobody tells you when you cross the line."
— Alex Onta, Executive Director, SINGUARD
Key Takeaways
- A partial fill happens when the size available at your price is smaller than the size you asked for.
- The visible quote applies to a limited amount, so average fill price degrades as order size grows.
- Order type decides the outcome: market orders accept worse prices, fill-or-kill and limit orders trade certainty of price for certainty of execution.
- Splitting a large order over time reduces market impact but adds timing risk, and one of those costs is always being paid.
Frequently Asked Questions
Why did my order fill at several different prices?
Because the size available at the best price was smaller than your order. The remainder walked into the next levels of the book, each at a worse price, and the platform reported the volume-weighted average. This is normal for larger orders and has nothing to do with the broker changing the price.
Can I stop partial fills?
You can prevent the partial part by using a fill-or-kill or all-or-none instruction where the platform supports it, which cancels the order rather than filling some of it. You are choosing no position over an incomplete one, so it suits arbitrage and hedging much better than discretionary trading.
How do I know how much size my market can absorb?
Look at depth of market rather than the top quote, and watch how it changes through the session. The same instrument can hold significant size in London hours and very little at the daily rollover, so size limits are a function of time as well as instrument.
About the Author
Alex Onta is an Executive Director at SINGUARD. He built eTrader, the terminal, the mobile apps, eTrader Broker, Copytrading, Business and Community, along with the worldwide clustered-server infrastructure it all runs on, with his brother Roman Onta helping on the design, and he leads that division today. Together with Roman he builds the Prop Firm CRM, the Broker CRM, Scalegram and CopySignals, and the two of them carry worldwide compliance, payment processing and international business structuring side by side. He lives and works in Dubai for most of the year. Meet the executive duo leading Singuard's five divisions.