Most traders pick an order type by habit and discover its price afterwards, in the fill. Every type is the same trade in disguise: you are choosing between certainty of execution and certainty of price, and you cannot have both. What follows is what each choice costs, and where the cost hides.
There are two things you might want from an order. You might want it done, now, whatever the price turns out to be. Or you might want a price, and be willing to wait or to miss the trade entirely. No mechanism gives you both, and every order type in existence is a particular way of splitting the difference between them.
Once you see the menu that way, the names stop mattering. A market order is the extreme of the first choice, a resting limit order the extreme of the second, and everything else is a rule attached to one of the two that says when to give up. The interesting question is never which type is best, it is which of the two certainties you can afford to lose on this particular trade.
Certainty of execution and certainty of price are the two things being traded. Every order type is a ratio between them.
A market order says take whatever is there. It crosses the spread immediately, which means it always pays the fee schedule's taker side, and it consumes resting liquidity starting at the best price and working outwards until the size is filled. On a deep book with a tight spread that second effect is invisible. On a thin book, or at a size that exceeds what sits at the top level, it is the largest cost of the trade and it does not appear on any invoice.
This is the part traders underestimate most consistently. The taker fee is a published number you can look up. The walk up the book is not published anywhere, it depends on the size you sent and the depth that happened to be there in that second, and it can easily exceed the fee by several multiples on a mid-size instrument during a quiet hour.
Market orders are the right tool in exactly one situation: when not being filled costs you more than being filled badly. Closing a position that is moving against you qualifies. Opening a position because a chart looked interesting does not.
A limit order names a price and refuses anything worse. If it rests in the book rather than crossing, it adds depth for someone else to trade against, and fee schedules reward that with the maker side. The reward is real and it compounds across a year of trading, which is why active traders eventually reorganise their entries around it.
The cost is equally real and much harder to measure: the trades you did not get. A limit order that never fills has a cost equal to whatever the move was worth, and unlike a fee that cost never shows up in a statement. Traders who move to limit orders for the fee saving and never look at their fill rate frequently make the saving and lose more than it on the entries they missed.
This is the single most common misunderstanding in fee optimisation. A limit order is not maker by nature. It is maker only if, at the moment it arrives, it does not cross the opposite side of the book. Send a limit buy at or above the best offer and it executes immediately against what is resting there, which makes it a taker in every respect that matters to your fee schedule. The word limit describes the price constraint, not the fee side.
Post-only is a limit order with one added instruction: if this order would execute immediately, cancel it instead of executing. It converts a preference into a guarantee. You will never accidentally pay taker on an order sent post-only, because the order simply ceases to exist rather than crossing.
That guarantee has a failure mode worth understanding before relying on it. In a fast market, the price you meant to post at is frequently the price the book has already moved through by the time your order arrives, so the order is rejected. You then have a position you thought you had opened and did not, discovered at the worst possible moment. Post-only is a tool for patient entries, not for anything that has to happen.
Post-only never costs you a taker fee. It regularly costs you the trade instead. Both are prices, and only one is on the schedule.
A stop is not an order type in the same sense as the two above. It is a trigger with an order attached, and the order it releases is either a market order or a limit order. Everything about its cost comes from that attached order, not from the stop itself, and the stop is invisible to the book until the moment it fires.
It guarantees that you are out, at a price nobody can promise. Stops trigger during exactly the conditions that thin the book, and a stop market fired into a fast move can fill considerably worse than the trigger price. That is not a malfunction, it is the mechanism working as designed under conditions where certainty of execution is expensive.
It guarantees a price and abandons the guarantee that you are out. If the market gaps through your limit, you are still in the position with a resting order that will not fill, which is the specific outcome most stops are placed to prevent. Choosing between the two is choosing which failure you would rather have, and there is no answer that is correct in general.
Every order carries a rule about how long it lives, and traders who never set it explicitly are using their venue's default without knowing what it is. The rules that matter are few.
It rests until it fills or you remove it. The risk is not the order, it is forgetting the order: a resting bid from a setup you abandoned three days ago is a live position waiting for a bad afternoon.
Fill whatever you can right now at my price or better, cancel the rest. This is the correct instruction for taking a specific slice of visible depth without leaving a footprint in the book afterwards, and it is what most execution algorithms are built from.
All of it at my price, right now, or none of it. It exists for cases where a partial fill is worse than no fill, which is rarer than it sounds and mostly concerns hedged or paired trades where one leg alone is a new risk rather than a smaller version of the intended one.
On a derivatives venue, an order to sell is not necessarily an order to close. If your position is smaller than the order, the excess opens a position in the opposite direction, and you can end a session short when you meant to be flat. Reduce-only prevents this by capping the order at the size of the existing position.
It matters most in the situations where attention is scarcest: a stop that was sized for a position you have since partly closed, a manual exit typed quickly during a move, an automated system whose position tracking has drifted from reality. In each case reduce-only turns a possible reversal into a simple exit. There is almost no reason to place a closing order without it.
A large resting order is information. Everybody watching the book can see the size, and a certain amount of trading exists specifically to act on it. Iceberg orders answer this by showing only a slice at a time and replenishing as each slice fills, so the full size never appears.
The concealment is not free. A refreshed slice generally loses its place in the queue and joins the back of the price level, which means the same total size fills more slowly than it would have if displayed. Fully hidden orders go further and frequently pay the taker side regardless of whether they crossed, on the reasoning that invisible depth is not depth anyone could have traded against. The trade is straightforward: you pay in queue position or in fees for the right to be invisible, and you should know which currency your venue charges in before you rely on it.
Strip out the vocabulary and a small number of rules do most of the work. Entries that can wait belong on post-only limits, because that is where the fee difference accumulates across a year and the missed fills are affordable. Exits that must happen belong on market or stop-market orders, because at that moment the cost of not being filled dominates every fee consideration. Size that exceeds the visible top of the book belongs in slices, whatever type you use, because the walk up the book is the real cost and it grows faster than linearly.
The one habit that pays for itself immediately is measuring rather than assuming. Compare the price you intended with the price you received, per trade, for a month. That difference is your true execution cost, it includes the fee and everything the fee does not cover, and it is almost always larger than traders expect. Once it is written down, the choice of order type stops being a preference and becomes an arithmetic problem with a visible answer.
The fee is the part of your execution cost that somebody publishes. The rest is the part you have to measure yourself, and it is usually bigger.
No. A limit order is cheaper only when it rests in the book and earns the maker side. A limit order that crosses on arrival pays taker exactly like a market order, and a limit order that never fills has a cost that no fee schedule shows: the trade you did not take. The saving is real but it is conditional on the order actually resting.
Intent versus guarantee. A limit order will execute immediately if the price allows it, which quietly makes it a taker. A post-only order is cancelled instead of executing in that situation, so it can never take. Use post-only when the maker side is the point of the order and missing the trade is acceptable.
Because a stop market order guarantees execution, not price. The trigger releases a market order into whatever book exists at that instant, and stops trigger during exactly the moments when the book is thinnest. The gap between trigger and fill is slippage, and it is the price of the certainty that you are out.
Only if you would rather stay in the position than exit at a bad price. A stop limit protects the price and abandons the exit: if the market moves through your limit, the order rests unfilled and you are still exposed. Neither choice is safer in general, they fail in opposite directions and you are choosing which failure you can live with.
It fills as much as it can against what is available right now at your price or better, and cancels whatever is left instead of resting in the book. It is the standard instruction for taking a defined slice of visible depth without announcing your remaining interest to everyone watching.
Careful is not a mechanism. Reduce-only caps an order at the size of the position it is closing, which means a stale stop or a mistyped size cannot flip you into a new position in the opposite direction. It costs nothing and removes an entire class of expensive accident, so there is little reason to trade a derivatives account without it on every exit.
Usually not. Concealment matters when your order is large enough relative to the book that showing it changes how others behave. At sizes that sit comfortably inside the top of book, nobody is reacting to you, and the queue position you give up each time a slice refreshes is a real cost paid for a benefit you are not receiving.
Record the mid price at the moment you sent each order and compare it with your average fill, then add the fee. The sum is your total cost per trade. Traders who do this for a month usually find that the part outside the fee schedule is the larger half, and that it varies enormously by time of day and by instrument.