The first time I watched a size order walk the book, I remember thinking the fill price could not possibly be right. It was a mid-cap alt, decent depth on paper, and the intent was to buy a chunk that looked small next to the daily volume. The problem is that daily volume is spread across a whole day, and the resting liquidity at any single moment is a fraction of it. So the order ate through the offers, each one worse than the last, and the average fill came in well above where the screen said the market was when I started. Nobody stole anything. I just paid to consume liquidity that was not there in the size I wanted.
That gap is the whole reason OTC desks exist, and it is the thing most people are actually asking about when they ask whether they should be calling a desk. So let me lay out how I think about the two, and roughly where the line sits.
Spread you pay once versus impact you pay the whole way in
On an exchange, working a large order means you are paying market impact. Every unit you buy lifts the price a little for the next unit, and the total cost is the area under that curve, not the quote you saw at the top. Impact scales with how much of the visible liquidity you consume, and it gets worse fast once you go past the top few levels of the book. You can soften it by slicing the order over time, using limit orders instead of hitting the market, or leaning on an execution algo that feeds pieces in slowly. But slicing over time trades one cost for another, because now the market can move against you while you wait, and a slow order gives everyone watching more time to notice.
An OTC desk quotes you a single price for the whole clip. You tell them roughly what you want, they give you a number, and if you take it, you are filled at that number for the entire size. The spread they quote is wider than the exchange top-of-book, sometimes noticeably. That is the fee for certainty. You pay a known spread once, instead of an unknown impact cost that you only find out after the order is done. The desk then goes and hedges or unwinds the position on its own book and its own timeline, and that risk is theirs, not yours.
The mental swap is this. On-screen, the visible price looks better but the true cost is hidden and grows with size. On a desk, the quoted price looks worse but it is the actual cost, fixed, for the full amount. For small orders the exchange wins easily because impact is basically nothing. For large orders the desk can win because a wide fixed spread beats an ugly impact curve you cannot see the bottom of.
Information leakage is the part people underrate
Here is what took me the longest to appreciate. When you work a big order on an open book, you are broadcasting. A large resting bid, or a series of aggressive buys stepping up the price, is a signal, and there are plenty of participants whose entire job is to read that signal and trade ahead of you. You are not just paying to cross the spread. You are paying because your own footprint moved the price before you finished.
A desk keeps the order private until it is done. The counterparty knows, and that is it. This matters most for names that are thin, or for a direction that would be obvious and painful if the market saw it coming. A quiet accumulation you do not want front-run is exactly the case where the desk earns its spread, even if that spread looks expensive on a spreadsheet.
The flip side, and it is a real one, is settlement risk. On an exchange with a custodian or a matching engine sitting between you and the other side, you are not trusting a specific counterparty to deliver. With an OTC trade you are. You send, they send, and if the plumbing is not set up right, one side is exposed for the window in between. Reputable desks solve this with escrow, atomic settlement, or delivery-versus-payment style arrangements, and you should never do a large OTC trade without knowing exactly how settlement works and who is holding what, when. If a desk cannot explain their settlement mechanics clearly, that is your answer.
Where the line actually sits
There is no universal number, because it depends entirely on the liquidity of the specific asset, not on the dollar size in the abstract. The same clip that is trivial in a top-cap coin can be a disaster in a thin one. So the threshold is relative. The rule of thumb I use is to compare my order against the real resting depth, not the daily volume headline.
Here is the rough workflow I run before deciding whether to click or call:
- Pull the order book for the venue I would actually trade on and add up the resting liquidity within a price band I am willing to pay, say the first stretch of levels before the price gets ugly. That is my real available depth right now, not the 24-hour number.
- Compare my order to that depth. If I am a small slice of it, the exchange is fine and a desk is just overhead. If my order is a meaningful fraction of it, or larger, impact is going to hurt and I start taking the desk seriously.
- Ask how patient I can be. If I have hours or days, an execution algo slicing into the book can get me close to the desk on cost without paying the spread, as long as the market cooperates. If I need it done now, patience is not an option and the desk's certainty is worth more.
- Ask how much the direction matters if it leaks. Quiet accumulation in a thin name pushes me toward the desk. A routine rebalance in a deep name where nobody cares stays on-screen.
- Get a desk quote anyway, even when I lean toward the exchange. The quote costs nothing and it tells me what the market really thinks my size is worth. If the quote is close to my estimated all-in exchange cost, I take the certainty. If it is way wider, the desk is telling me they think the position is risky to hold, which is information too.
The failure mode I see most often is people anchoring on the top-of-book spread and concluding the exchange is always cheaper, because the desk quote looks fat next to it. They are comparing the desk's honest all-in number against the exchange's advertised price that they will never actually get for their full size. The right comparison is desk quote versus your realistic all-in exchange cost including impact and slippage, and once you frame it that way the desk stops looking so expensive above a certain size.
A rough default
If your order is a small fraction of the resting depth in a liquid name, click. Work it with limits or an algo and do not overthink it. If your order is a large fraction of that depth, or the name is thin, or the direction would be obvious and painful if it leaked, get a quote and seriously consider calling. And in the middle, which is where most interesting orders live, get the desk quote and let the number decide, because the quote itself is the cheapest read you will get on what your size is really worth.
Most of the work here is just seeing the true cost on both sides before you commit, which means having the order book depth and your own footprint in front of you rather than guessing. That is a lot of what we built the execution and depth views in Blockcircle to surface, so you can size against real liquidity instead of the headline volume number. Whatever you use to look at it, the habit is the same. Measure your order against the depth that is actually there, price the leakage, and only then decide whether you are clicking or calling.