The quote is a summary, not the market
Spot gold has no single central exchange. Prices come from a set of institutions quoting bid and offer, and a provider builds the number you see by taking those streams, filtering them and presenting one top of book. Different providers subscribe to different sources, filter differently and update at different intervals, so two platforms can show prices that disagree slightly without either being wrong.
That has a consequence worth internalising early. The price on your chart is an indication of where business could recently have been done in a given size. It is not a promise, and it says nothing about depth, which is how much can actually trade at that level. A quote good for a small order may not exist for a large one. In quiet conditions the distinction rarely matters, because depth is plentiful relative to retail size. In fast conditions it is the whole story, because depth disappears before the displayed price does, and your order then meets the market rather than the picture of it.
Principal and agency: who takes the other side
There are two basic relationships a provider can have with your order. Acting as principal means the provider itself becomes the counterparty: you buy from it and sell to it, and it carries the resulting position. Acting as agent means the provider passes your order to an external pool of liquidity and earns a fee for doing so, holding no position of its own.
Most retail gold trading sits somewhere on that spectrum rather than at a pure extreme. A provider may take the other side of small orders itself while passing larger or more directional flow out to external counterparties, an arrangement usually described as splitting flow between internal and external books. The decision is typically automatic and based on the characteristics of the account and the order rather than made by a person. Published execution policies describe the general approach, since providers are usually required to document how they seek the best available result for clients. The figure sets out the routes an order can take once it leaves your platform.
Internalising flow is not automatically a conflict
Taking the other side of a client order sounds like a conflict, and in a narrow sense it is: if the provider holds your losing position, your loss is its gain. The reality is more boring. A provider that internalises flow holds a book of many clients whose positions partly offset each other, hedges the residual exposure externally, and earns its money on volume rather than on individual outcomes. Deliberately warehousing risk against a client is a regulatory problem rather than a business model.
Where the model does show up legitimately is in behaviour under stress and in the detail of execution. An internalising provider can often offer tighter pricing and fewer rejections in normal conditions, because it is quoting its own book rather than relaying a third party quote. When volatility spikes, its appetite for risk changes, and that is where you may see wider spreads or restrictions. An agency route passes through whatever the external pool is doing, which can mean better pricing in size and more rejections when the pool pulls back. Neither model is strictly superior. They fail differently, and knowing how yours fails is the useful part.
Aggregation, last look and rejections
When an order goes out to external liquidity, the provider has usually built its quote by aggregating several streams and showing you the best bid and the best offer available across them. Your order is then routed to whichever source is showing that price. The source may have a contractual right to a brief final check before accepting, commonly called last look, which exists because it is quoting continuously into an uncertain market and wants protection against being picked off on a stale price.
From your side, last look appears in one of three ways. The order is accepted at the price you asked for. It is rejected, and you are told so. Or you are offered a new price to accept or decline, which older platforms call a requote. The reason this matters is that rejection is not randomly distributed. It clusters precisely when price is moving fast in one direction, which is when you most wanted the fill. An approach that depends on reliably being filled at a displayed price during rapid movement is depending on the part of the system that is least reliable, and no amount of platform configuration changes that.
Two wrappers for the same cost
Cost reaches you in one of two wrappers. Either the provider widens the spread it quotes and charges nothing separately, or it quotes a raw aggregated spread and charges an explicit commission per trade. These are the same economic transaction described differently, and comparing the spread alone between the two is how people reach wrong conclusions about which is cheaper.
To compare properly, convert everything into one figure per round trip in the units you think in, usually dollars per ounce or cash for your standard position. Add the spread you cross, plus commission on both sides, plus financing if the position is held overnight, plus the slippage you actually experience rather than the slippage you hope for. Then compare that total against the distance your trades typically aim for. A raw spread with commission can easily be more expensive than a wider all in spread at one size and cheaper at another, because the two components scale differently. The arithmetic is dull and it is the only comparison that means anything, which is the argument made at length in the real cost of the spread.
Order types decide what you meet
Your order type decides which parts of the machinery you are exposed to.
- A market order accepts whatever price is available. It fills in almost all conditions and it gives away the price.
- A limit order fixes the worst price you will accept. It protects the price and risks not filling at all.
- A stop order is an instruction that becomes a market order once a level trades. It protects you from staying in a losing position, not from the price you get on the way out.
- A stop limit sets a boundary on the exit price, which means in a fast move it can fail to execute and leave the position open.
Some platforms also expose a maximum deviation or slippage tolerance, which converts a market order into something closer to a limit with a margin of acceptance. That reduces bad fills and increases missed fills. There is no setting that gives you both, and choosing between them is a risk decision rather than a technical one. It matters most where cost is large relative to the target, which is the situation described in short horizon gold trading.
Slippage symmetry is the measurable part
You cannot audit a provider's routing decision from the outside. You can measure its results, and the most informative measurement is the distribution of your own slippage. For each order, record the price you intended and the price you received, with a timestamp. Then look at the shape of the collection rather than at individual cases.
If the provider passes on improvements as well as deteriorations, you should see fills on both sides of your intended price, roughly balanced in quiet conditions. If every deviation is against you and none is ever in your favour, that is asymmetric slippage, and it is a cost you were never quoted. Separate the data by condition, because mixing a quiet afternoon with a data release destroys the signal entirely. Everyone receives worse fills in a fast market, and that is the market rather than the provider. What you are testing is whether normal conditions are symmetric and whether fast conditions are merely bad or genuinely catastrophic.
What you can verify, and what you cannot
So what is actually checkable? The execution policy document, which providers typically publish and which describes order handling in general terms. The instrument specification, which gives you contract size, minimum stop distances and financing. Your own trade records, which are the only evidence that exists about your own fills. And the behaviour of your platform during known volatile moments, which you can observe deliberately rather than discovering by accident with a large position open.
What is not checkable is the internal decision about where any individual order went. No amount of chart analysis reveals that, and claims in either direction, from traders or from providers, are not verifiable from your seat. The practical response is to stop arguing about intent and start logging outcomes, which is the habit described in keeping a journal that is actually useful, and to pay close attention to how your fills behave when a release lands, as set out in spread and slippage around news. Watching a release arrive on the live gold chart while noting what your own orders did is more instructive than any comparison table.
FAQ
Why do two platforms show different gold prices at the same moment?
Because spot gold has no single central exchange. Each provider builds its quote from its own set of institutional streams, filters them and updates on its own cycle. Small disagreements are normal and structural. Large persistent disagreements are worth investigating, but a few cents of difference is simply how the market works.
Is a market maker model bad for me?
Not inherently. A provider that takes the other side of small orders often quotes tighter and rejects less in normal conditions, because it prices its own book. It behaves differently under stress. Agency routing passes through an external pool, which can price better in size and reject more often.
What is a requote?
It is the provider returning a different price instead of filling at the one you clicked, usually because the original quote is no longer available. You accept or decline. It appears most often when price is moving quickly, which is exactly when the fill mattered. Rejection and slippage are the other two outcomes.
How do I tell whether my slippage is fair?
Log the intended and actual price for every order with a timestamp, then look at the distribution separately for quiet and fast conditions. Deviations on both sides in quiet conditions suggest symmetric handling. Deviations only ever against you suggest an unquoted cost. A single bad fill tells you nothing at all.
Is commission plus raw spread cheaper than a wider spread with no commission?
Sometimes, and it depends on your size and frequency. Convert both into a total cost per round trip for the position size you actually trade, then compare that against your typical target distance. The wrapper does not matter, the total does, and the answer changes as size changes.
ⓘ See these ideas on real price: open the free XAUUSD live chart.