Liquidity, spreads and weekend gaps — what a crypto “price” conceals

The largest number on a crypto app is almost always the least informative. It is the last price — sometimes a trade, sometimes an index, sometimes a mark inherited from a derivatives book — set in type large enough to suggest certainty. Around that number sit the facts that decide whether a reader in Singapore can actually transact: who is bidding, who is offering, how much size is there, and what happens if you walk further than the first lot.

This briefing is a vocabulary lesson. It is not a method for beating a spread. For the clock that changes these numbers, read crypto markets in Singapore time. For the business model that profits from some of this friction, read how crypto exchanges earn money. Both belong, with this page, in the markets hub.

Bid, ask, and the last print — three different objects

A bid is an advertised willingness to buy, at a price, for a size. An ask (or offer) is an advertised willingness to sell. The spread is the distance between the best bid and the best ask. The last is the most recent trade that the venue chose to publish.

Those sentences are older than bitcoin. They still fail in retail interfaces that show a single pulsing figure.

A last print can be stale. On a quiet book it may be minutes old. It can also be aggressive: someone crossed the spread, paid the ask, and left a souvenir that now looks like “the price”.

A spread can be tight and meaningless. One bitcoin advertised at a one-dollar gap does not mean a hundred bitcoin are available at a one-dollar gap.

A mid — the average of bid and ask — is a convenience for charts. It is not a price anyone is entitled to receive. People who mark their holdings to mid every night are choosing an optimistic accountant.

In Singapore-dollar terms the confusion multiplies. Many books are quoted in US dollars. The app then translates at an FX rate that has its own bid and ask. A reader who thinks they are looking at one spread is often looking at two.

Depth is the rest of the iceberg

Depth is the size available at each price level away from the touch. A book with a polite spread and no depth is a shop window with empty shelves.

Market-data screens that professionals use show a ladder: price, size, sometimes the number of orders. Retail apps often hide the ladder or truncate it. Social media hides it by design. A screenshot of a green candle contains no depth at all.

Why depth matters, in ordinary language:

• Small retail clips may clear at the touch on a major pair during a weekday afternoon, SGT.
• Larger clips walk the ladder. The average price you receive is worse than the last, worse than the mid, and sometimes worse than the first ask you saw.
• Advertised “blocks” on chat desks are a different market. They are negotiated. Their reference price may be an index that the block itself would have moved if it had gone across the screen.

Depth is also where games are played. Spoofing — advertising size and pulling it — is illegal in regulated markets and not unknown in unregulated ones. Wash prints can decorate a last price. A licensed platform in Singapore is supposed to police some of this. An offshore book may treat it as atmosphere.

The educational habit is to ask, every time someone shows you a price: how much could I have done, then, without moving it? If the answer is “I don’t know”, you have a headline, not a market.

Slippage is what the screenshot omits

Slippage is the difference between the price you thought you would get and the price you did get. It has several honest sources and several dishonest ones.

Honest sources include a moving market between the moment you tap and the moment the matching engine hears you; a book that was thinner than the interface implied; and a fee that was not in the large number. Some platforms charge a taker fee that is larger than the displayed spread. The economic spread is then fee-plus-gap, not the gap alone.

Dishonest-adjacent sources include hidden latency, last-look behaviour on OTC chat, and interfaces that refresh the mid while your order is in flight. None of those require a conspiracy. They require a reader to remember that the other side is not a charity.

A worked example, with invented round numbers for teaching rather than as a quote. Suppose the screen shows 100,000 Singapore dollars per bitcoin, bid 99,950 / ask 100,050, last 100,020. You send a market buy for a size that exceeds the first ask. The engine sells you the resting size at 100,050 and then the next level at 100,400. Your average might be 100,220 before fees. The candle that Twitter displays may still close near 100,020. You did not get “the price”. You got the book.

Weekend versions of this example are uglier because the second level may be much further away.

A last price is a souvenir of a trade that someone else did, in a size you do not know, on a venue you may not use. Treating it as your price is how people confuse a photograph with a quote.

Weekend books: the same software, fewer people

Public blockchains do not take Sunday off. Matching engines often do not either. The humans who provide depth do.

What thins first is usually the professional inventory. Firms that hedge options, make two-sided markets, or arbitrage one venue against another reduce risk into the weekend. Some automated quotes remain. They are not always as wide or as deep as the weekday versions. Some platforms also restrict leverage or withdrawals for maintenance, which changes who is willing to sit in the book.

The result is a familiar pattern that Singapore mornings then have to interpret.

A move begins on a Friday night, US time — Saturday morning here. Size that would have been absorbed on Wednesday walks an empty ladder. The print looks historic. On Sunday evening, as Asia returns, the book thickens. On Monday the wick is explained in group chats as manipulation, as a “glitch”, or as a buying opportunity. The dull explanation is usually enough: the market was open and the people were not.

Gaps, in the charting sense, appear when one session’s last and the next session’s first do not meet. In a 24-hour market the “session” is unofficial, so the gap is often a thin-print canyon rather than a true halt. It still has the same educational meaning. Price discovery happened with fewer voices.

Two Singapore-specific notes.

First, Sunday night is when many local readers look at markets before a working week. That is exactly when overseas weekend noise is still being digested. Starting a factual argument from a Sunday 11pm screenshot is how families acquire bad data.

Second, local public holidays do not close global books, but they do remove a slice of Asian participation. A quiet Hari Raya or Lunar New Year week can look like a weekend even on a Wednesday.

Why screenshots of “the price” mislead — a short charge sheet

What the picture drops

Screenshots travel well. They are also structured to drop context. A charge sheet, not a complete one:

Venue omitted. A perpetual-futures mark on an offshore site is not a cash pair on a MAS-licensed platform. The numbers can diverge by more than a casual reader expects, especially around events.

Size omitted. A print for a few hundred dollars is displayed in the same type as a print for several million.

Time zone omitted. A candle labelled “18:00” without a zone is a Rorschach test. In SGT that might be London’s morning or New York’s open, depending on the platform’s default.

Index construction omitted. Some popular prices are composites. A composite can look stable while every constituent book is sliding, or volatile while your particular venue is unchanged.

Fees omitted. The number you would have paid includes the spread you walked and the fee schedule you clicked through six months ago.

Halt and degraded-mode omitted. Venues pause, throttle or switch to last-look. The candle may be interpolated. The withdrawal queue is not in the picture.

Currency omitted. USD last, SGD translation, and USDT last are three stories. Tether itself has a spread to dollars.

None of this means prices are fake. It means a picture is a claim, and claims need a caption. Journalists are supposed to know this. So are readers who forward pictures to a finance director.

How exchanges sit relative to the spread

A matching engine that sits between bids and asks can earn the spread, earn a fee, or both. Some platforms act as agent and charge a commission. Some take the other side. Some do both at different times of day. The companion piece on how exchanges earn money is the place for that business model. The market-structure point for this page is narrower.

If the platform is the other side, your “last price” may be the platform’s inventory decision. If the platform is an agent, your last price is someone else’s inventory decision, plus a fee. If the platform’s displayed price is an index pulled from elsewhere, you may be looking at a market you cannot actually access.

Licensing does not abolish these designs. It constrains how they may be described and how client assets are held. An honest interface still owes the user a book, a fee schedule and a timestamp. Many interfaces prefer a big number. Readers can prefer the book.

A working check-list, then stop

Before treating a number as a fact worth repeating, a short list:

• Which venue, which pair, which currency.
• Bid, ask and last — all three, not one.
• Depth at the touch and one or two levels beyond.
• Time in SGT, and who else was likely awake.
• Whether the print is cash, derivative or index.
• Whether the platform was in a normal or a degraded state.

If you cannot fill those in, you are not looking at a price. You are looking at marketing.

The networks run on weekends. The people do not. The screenshot never mentions that, which is why it remains such an efficient way to be wrong.