How to read Hyperliquid wallet positions

What on-chain transparency gives you

Hyperliquid is an on-chain perpetual exchange. Positions, orders and trades are tied to wallet addresses, and the venue makes this data available through its public interface. On most centralised exchanges, the same information stays private.

This visibility lets you inspect a position directly: the asset, the side, the size, the entry price and the margin used. You do not need to rely on a second-hand report of what a large trader did.

The transparency has a clear boundary, though. You see activity on one venue. You do not see the same person’s activity on other exchanges, in spot markets or in private agreements. A wallet view is a window into one room of a larger house.

A public wallet shows what happened on one venue, not the whole portfolio of the person behind it.

An address is not an identity

A wallet address is a string of characters. It does not carry a name. One person can control many addresses. One address can also be controlled by a group, a fund or a program. A label next to an address is an interpretation, not a proof.

Some addresses are vaults or program-controlled accounts that pool money from many depositors. The position then reflects a strategy that many people fund, not the decision of one trader. The page may not show how many depositors stand behind the address.

Labels can come from several places, such as public statements, observed transfers or pattern matching. Each source has its own error rate. Treat a label as context, and inspect the underlying activity before you accept it.

The same logic applies to the word whale. The word describes size of position or account, nothing else. It does not describe skill, information or intent. A large account can be wrong as often as a small one.

Size describes an account. It does not describe skill or intent.

Positions can be hedged elsewhere

A trader can open a large short on Hyperliquid and hold the same amount of the asset in spot or on another exchange. The two positions cancel each other in price risk. On Hyperliquid alone, the short looks like a strong directional view.

Basis trades and funding trades follow this logic. The trader holds offsetting positions to collect a payment, such as a funding rate, and does not intend to take price risk. A wallet that runs such a trade can look very large and very one-sided.

Market makers add another case. They hold positions to provide liquidity, and they manage their risk across venues. Their visible position often reflects inventory, not a view.

Because you cannot see the other leg, avoid sentences such as this wallet expects the price to fall. Prefer sentences such as this wallet holds a large short on this venue.

Describe what the wallet holds on this venue; do not claim what the trader believes.

Position size versus account value

A large position says little unless you know the size of the account behind it. Leverage links the two. Leverage equals the notional value of the positions divided by the account equity. A large notional value on a large account can be less risky than a small notional value on a small account.

Check three values together: the notional value of each position, the total account value and the share of the account that backs each position. A position that uses most of the account is fragile. A position that uses a small share of the account has room to absorb a price move.

Concentration matters too. An account that holds one large position in a single asset depends on one price. An account that holds several smaller positions across assets depends on several prices that may move together or apart. Compare the largest position with the total, not only the largest position alone.

Margin mode matters here too. In cross margin, all positions share the same collateral, so a loss in one position uses the margin of the others. In isolated margin, each position has its own collateral. Look at the total exposure across positions rather than at one position at a time.

Unrealised and realised profit and loss

Unrealised profit and loss is the gain or loss on a position that is still open. It is calculated from the current mark price, and it changes every time the price changes. Realised profit and loss is the result of a position that has been closed, plus other settled items such as fees.

A wallet with a large unrealised profit has not earned that profit yet. A price move can reduce it to nothing before the position closes. Do not add unrealised and realised results into one number without saying so.

Funding payments and fees also change the result. A position can show a price gain and still lose money after it pays funding for a long time. Check whether the page includes funding and fees in the figure it shows.

Realised history has limits as well. It depends on the period the page covers, and it may leave out transfers between wallets. A realised figure from a short window is a poor summary of a trader’s record.

Unrealised profit is an open result that the next price move can change.

Estimated liquidation prices

The liquidation price of a position is the price at which the margin falls below the maintenance requirement. A page can estimate it from the wallet data. The estimate depends on the margin balance, the margin mode, the other positions in the account and the venue’s margin rules.

The estimate changes whenever the trader acts. Adding collateral moves the level away from the price. Reducing the position does the same. Opening another position in cross margin can move it closer. A level you read now can be out of date in a short time.

Use the estimate to describe distance, for example how far the mark price is from the estimated level in percent. Do not use it to say that a liquidation will happen or that other participants will target the level. Liquidation data and the liquidation heatmap give separate views of forced closures.

An estimated liquidation price is a snapshot that changes with every margin action.

Timing, snapshots and changing positions

A wallet page is a snapshot taken at a specific time. Between two snapshots, a trader can add to a position, reduce it, close it or reverse it. A single snapshot cannot tell you which of these happened.

Fills can also be partial. A large order can fill in many steps at different prices. The average entry price then differs from the price of any one fill. Read the entry price as an average, not as the price at one moment.

Funds can move between wallets. A trader can split a position across several addresses or move collateral from one to another. A drop in one wallet does not always mean that the trader left the market. Always note the observation time next to any number you quote.

Survivorship bias in wallet lists

Lists of top wallets are built after the results are known. A wallet appears on the list because it earned a large profit. Wallets that lost money, closed down or never grew do not appear. A list of winners therefore says little about the average outcome of all traders.

Luck matters too. If many traders take large risks, some of them will show a long run of gains by chance. Looking at that group afterwards makes the run look like skill. Past profit does not establish future results.

Ask what the list would look like if it had been made earlier. Ask how many wallets were removed or reset. Ask whether the track record covers one market phase or several. These questions do not give answers about the future, but they protect you from reading too much into a list.

A list of winners is selected by its outcome, so it cannot show how common the outcome is.

A reading routine for any wallet

Use the same questions every time you open a wallet. They keep the reading factual and they slow down conclusions that the data does not support.

• What is the observation time of this snapshot?
• How large is each position compared with the account value?
• Is the profit realised or unrealised?
• Could this position be a hedge for something I cannot see?
• How far is the estimated liquidation level, and what can change it?

Describe the snapshot, name its limits and stop before the claim that the data cannot support.

Questions