dYdX Review 2026: Perps Trading, Fees and Real Risks

What dYdX is in 2026

dYdX is a decentralized perpetual-futures exchange. The version most people mean when they say "dYdX" today is v4, which runs on its own Cosmos-based chain rather than on Ethereum. The order book lives on that chain. Matching happens on that chain. Settlement happens on that chain. There is no central server sitting in the middle holding your coins while you trade.

That architecture matters for how you read everything below. On a centralized venue, you deposit collateral into an exchange-controlled account and trade against an internal ledger. On dYdX v4, you keep custody of your funds in a wallet and interact with the protocol through a front-end. The protocol is the exchange. The front-end is a window into it.

That distinction is worth spelling out because it is where a lot of confusion lives. dYdX the protocol is a set of smart contracts and chain logic. dYdX the website is an interface. Anyone can build a front-end that talks to the same protocol. The provider states that the exchange runs on its own Cosmos-based v4 chain with an on-chain order book and more than 200 markets. Those are the numbers to anchor on: 200+ markets, on-chain book, self-custody.

What self-custody actually means here: your collateral sits in a wallet you control, and the protocol holds it as margin against your position. You are not handing your coins to a company that can freeze withdrawals, file for bankruptcy, or lose customer funds in a side venture. You are also not getting a customer-support line that can reverse a bad fill or a help desk that can unfreeze an account. The tradeoff is real in both directions, and I come back to it later.

One thing to be clear about up front: this review is written from the provider's published disclosures, fee schedule and documentation. I have not signed up, funded an account, or placed a trade on dYdX. Everything below is what the provider states, what the protocol's own data shows, and what the mechanics imply for someone holding a position.

Funding rates and the real cost of holding a position

Most perps reviews spend their words on leverage limits. Leverage limits are the least interesting number on the page. The thing that actually drains an account over time is funding, and the thing that actually kills an account in a hurry is a liquidation cascade. So let's start with funding.

Perpetual futures have no expiry date. To keep the contract price tethered to the spot price, the protocol charges a periodic payment between longs and shorts. When the perp trades above spot, longs pay shorts. When it trades below, shorts pay longs. The payment is called funding, and it is usually settled every hour on dYdX v4, per the protocol documentation. That hourly cadence is worth noting: it means funding compounds against you faster than on venues that settle every eight hours, if the rate is against you.

Here is the part that gets skipped. Funding is not a fixed cost. It drifts. On a calm week, BTC perp funding can sit near flat. During a strong directional move, it can spike hard in one direction and stay there for days. If you are long and the market is euphoric, you are paying to hold. If you are short into a selloff, same story in reverse. The annualized cost of a position held for a month is not the rate you saw when you opened it. It is the average of every hourly rate over that month, and that average can be several times the opening rate.

The protocol publishes historical funding data, and that is where I would point anyone doing their own math. Pull the BTC and ETH perp funding history, annualize it, and compare it against a centralized venue's funding on the same pair over the same window. The gap is not always large, but it is rarely zero, and it moves around. A trader who checks funding once at entry and never again is flying blind on their largest recurring cost.

To make the drift concrete without inventing numbers, here is the structure of the calculation. Take the hourly funding rate, multiply by 24 for a daily figure, then by 30 for a monthly figure. That is the cost of holding one unit of notional for 30 days, before fees. If the rate is positive and you are long, that is what you pay. If it is negative and you are long, that is what you receive. The sign flips with your side of the trade.

Funding level (hourly, illustrative)	Annualized cost of holding	30-day cost on a position
Flat	Near zero	Near zero
Mildly positive	Low single-digit annualized	Small drag
Elevated	Double-digit annualized	Noticeable drag
Stressed	Triple-digit annualized	Position cost can exceed the move you are trying to capture

The table is deliberately qualitative. The exact rates change hourly, and quoting a specific figure from a past window would be misleading by the time you read it. The point is the shape: funding is cheap in calm markets and expensive in stressed ones, and stressed markets are exactly when most people are holding a directional position. Check the protocol's published funding history for the current numbers before you size anything.

Compare that against a centralized perps venue. Centralized exchanges also charge funding, and their rates are usually published in the same hourly or eight-hourly format. The difference is not that one is free. The difference is that on dYdX you can read the funding history directly from the chain, and on a centralized venue you are trusting the exchange's published number. For a trader who cares about verifying their own costs, the on-chain record is a genuine advantage. For a trader who just wants the cheapest hold, the comparison is empirical and changes month to month.

My read: funding is the single line item that decides whether a multi-day perp position was worth taking. If you are not going to check it, you are not really trading perps, you are paying an unmeasured fee for the privilege of holding a directional bet.

Liquidation mechanics and cascade risk

Liquidation is where perps stop being abstract. On dYdX v4, each position has a liquidation price calculated from the collateral you posted, the size of the position, and the maintenance margin requirement for that market. When the mark price crosses that threshold, the protocol's liquidation engine takes over. The provider's documentation describes the sequence: the position is closed, the insurance fund absorbs any shortfall, and if the insurance fund cannot cover it, auto-deleveraging kicks in.

Auto-deleveraging, usually shortened to ADL, is the part most users have never read about until it happens to them. In simple terms: if the insurance fund is exhausted, the protocol closes profitable positions on the opposite side of the book to cover the bad debt. You can be profitable, correctly positioned, and still have your trade closed early because someone else's liquidation blew through the fund. That is not a bug. It is the backstop that keeps the exchange solvent. But it is a real risk that centralized venues handle differently, often by socializing losses or by drawing on exchange reserves.

The insurance fund is the buffer between a normal liquidation and an ADL event. The protocol publishes its balance, and watching that number is one of the more useful things a dYdX user can do. A healthy fund means liquidations get absorbed quietly. A thin fund during a volatility spike is when cascades get ugly.

Cascade risk is the second-order effect. When prices move fast, liquidations trigger market orders, which push price further, which triggers more liquidations. On a venue with deep books, this is self-limiting because there is enough resting liquidity to absorb the selling. On a thinner book, the same cascade runs further before it stops. dYdX's on-chain order book is deep on BTC and ETH and thinner on the long tail of the 200+ markets the provider lists. That is the honest picture: major pairs handle stress reasonably, smaller markets do not, and the difference is not something the marketing page will tell you.

What happened during past volatility spikes is documented in the protocol's liquidation and insurance fund data. I would not paraphrase specific events from memory, because the details matter and the numbers move. The pattern to understand is this: in fast markets, liquidation prices can be hit faster than you expect, the insurance fund can be drawn down, and ADL can close positions that had nothing to do with the original liquidation. Anyone trading perps with leverage needs to have read the provider's liquidation documentation before, not after, their first margin call.

Plainly: perps trading carries a high risk of loss. Leverage magnifies both directions. Liquidation can wipe a position entirely, and ADL can close a winning trade before you decide to exit. None of that is unique to dYdX, but the on-chain mechanics make the sequence more visible, which is either reassuring or alarming depending on your temperament.

Fee schedule and maker/taker economics

dYdX charges trading fees on a maker/taker basis. The exact rates and tier thresholds are set by the protocol's governance and published in the provider's fee schedule. I am not going to quote a specific number here, because the fee schedule is the authoritative source and it changes with governance votes. Check the current schedule on the provider's site before you size a strategy around it.

What I can describe is the structure. Makers post limit orders that rest in the book and provide liquidity. Takers cross the spread and remove liquidity. Makers generally pay less, sometimes receive a rebate, and takers pay more. Tiers usually key off 30-day trading volume, and sometimes off staked tokens or other protocol-specific criteria. The provider's disclosures spell out the current tiers.

There is a second cost layer that centralized venues do not have in the same form: chain costs. Every action on dYdX v4 is a transaction on the dYdX chain, and transactions cost gas. The provider's documentation covers how gas works on the chain, including the fact that fees are paid in the chain's native token. For an active trader, gas is a real line item, though it is usually small relative to trading fees on any meaningful size. For someone placing many small orders, it is not negligible.

Fee drag scales with trade frequency, and that is the part worth doing arithmetic on. A trader who opens and closes one position a day pays fees twice a day. A trader who scalps twenty times a day pays forty times. At maker rates the drag is modest. At taker rates, on a strategy that crosses the spread every time, the fees can quietly exceed the edge. This is true on every perps venue, but it is worth stating because dYdX's fee tiers reward volume, and a small account trading frequently will sit in the higher-fee tier while paying the same per-trade cost as a large account.

Worked example, using structure rather than invented numbers: suppose an active trader does a round trip per day on a position of some notional size. The annual fee cost is roughly the round-trip fee rate multiplied by the notional, multiplied by the number of trading days. Add gas per transaction. Add funding if the position is held overnight. The funding line is often larger than the fee line for anything held more than a day, which is why I put funding first in this review. The fee schedule gives you the trading component; the protocol's funding history gives you the holding component; the chain documentation gives you the gas component. Add all three before deciding a strategy is viable.

Compared with centralized perps venues, dYdX's fee structure is competitive but not obviously cheaper. The advantage is transparency: the fee schedule is on-chain and governance-controlled, so you can see what you are paying and why. The disadvantage is that you also pay gas, and gas is a cost that centralized users never think about.

Self-custody vs centralized perps: the real tradeoff

This is the section where most reviews pick a side and cheer. I am not going to do that, because the tradeoff is genuinely two-sided and depends on what you are afraid of.

On a centralized perps venue, you deposit funds into an exchange-controlled account. The exchange holds custody. That means you are exposed to exchange counterparty risk: insolvency, hacking, withdrawal freezes, regulatory action. It also means you get customer support, account recovery, fiat on-ramps, and a familiar interface. KYC is usually required, which is a privacy cost and a compliance feature depending on your view.

On dYdX, you keep custody. Your wallet holds the funds, and the protocol holds them as margin. There is no exchange that can freeze your withdrawals, because there is no exchange holding your coins. There is also no one to call if you send funds to the wrong address, lose your keys, or get liquidated in a way you did not understand. KYC is not required at the protocol level, though front-end operators may impose their own rules, and the provider's disclosures are the place to check what applies to the interface you use.

Withdrawal friction is different in kind. On a centralized venue, withdrawal is a request the exchange processes. On dYdX, withdrawal is a transaction you sign, and it settles when the chain settles. That is faster in normal conditions and slower when the chain is congested. It is also irreversible in a way that centralized withdrawals usually are not.

Dimension	dYdX (self-custody)	Centralized perps venue
Custody risk	You hold keys; no exchange custody	Exchange holds funds; counterparty exposure
Smart contract / protocol risk	Present; you depend on chain and contract correctness	Lower in this specific sense; you depend on exchange operations
KYC	Not at protocol level; front-end rules vary	Typically required
Withdrawal friction	Signed transaction; irreversible	Exchange-processed; sometimes reversible, sometimes frozen
Support	Community and docs	Customer support desk
Market, funding, liquidation risk	Present	Present

The line I want to be explicit about: self-custody removes exchange custody risk. It does not remove market risk, funding risk, or contract risk. A position can still be liquidated. Funding can still bleed you. A bug in a contract or a chain halt can still cost you money. Anyone who tells you self-custody makes perps safe is selling something.

My own view: self-custody is the right default for someone who already manages their own keys and understands irreversibility. It is the wrong default for someone who wants a support line and a password reset. Neither preference is irrational. They are different answers to the question of which failure mode you can tolerate.

Order types, liquidity, and execution quality

dYdX supports the order types you would expect on a perps venue: market, limit, stop-market, stop-limit, take-profit variants, and conditional orders. The provider's documentation lists the full set, including time-in-force options like good-till-cancelled and immediate-or-cancel. Advanced order types are there, which matters for anyone running anything more structured than a directional bet.

Liquidity is where the on-chain model shows its seams. The order book is on-chain, which means every resting order is a chain state update. On BTC and ETH perps, the book is deep enough that moderate size executes with slippage in line with centralized venues. On the long tail of the 200+ markets the provider lists, depth thins out fast. A market order on a smaller pair can walk the book several levels and fill at a materially worse average price than the top of book suggests.

That is the practical warning: market orders on thin books punish you. If you are trading anything outside the top few pairs by volume, use limit orders and accept that you may not get filled. The spread you see is not the price you get on size. This is true everywhere, but it is more acute on a venue where the book is on-chain and market makers have to factor chain costs into their quoting.

Execution quality also depends on which front-end you use. The protocol is the same, but different interfaces may route orders differently or show different depth views. The provider's official site is the reference point for what the canonical interface does. If you use a third-party front-end, you are trusting that operator's implementation on top of the protocol's correctness.

I would not pick dYdX for a strategy that depends on tight execution in illiquid markets. I would consider it for BTC and ETH perps where the book is deep and the on-chain transparency is worth something to you. That is a narrow but honest recommendation.

Verdict and scoring

Scoring here follows the site's methodology, which weights fees, transparency, liquidity, and risk controls. Full rubric is at the methodology page.

• Fees: Maker/taker structure with volume tiers and gas on top. Competitive but not clearly cheaper than centralized venues once gas is counted. The fee schedule is on-chain and verifiable, which counts for something.
• Transparency: Strong. Funding history, insurance fund balance, and liquidation data are readable from the chain. This is the clearest advantage over centralized venues.
• Liquidity: Deep on BTC and ETH, thin on the long tail. Execution quality varies sharply by market.
• Risk controls: Insurance fund and ADL are documented and visible. The mechanics are sound but the user bears the consequences of both, and ADL can close profitable positions.

dYdX is a credible decentralized perps venue with real transparency advantages and real execution tradeoffs. It is not a place to learn perps, and it is not a place to trade illiquid markets with market orders. It is a place where a user who already understands funding, liquidation, and self-custody can verify their own costs on-chain rather than trusting a published number.

Perps trading carries a high risk of loss. Leverage magnifies losses as well as gains, liquidation can wipe a position, funding can erode a hold, and ADL can close a winning trade. Nothing in this review changes that.

How this review was researched

This review is based on the provider's official site at dydx.trade, the dYdX v4 documentation covering funding, liquidation, the insurance fund, ADL, order types and chain gas, and the protocol's published fee schedule and on-chain data. No account was opened, no funds were deposited, and no trades were placed. Fee figures are not quoted here because the schedule is governance-controlled and changes; check the provider's current schedule for the numbers that apply to you. Market data referenced in the text is qualitative and time-sensitive; treat any specific rate you see elsewhere as a snapshot, not a constant. See the disclosure page for how this publication handles affiliate relationships, and the author page for background. For comparison, see our Kraken review, Binance review, and Coinbase vs Binance comparison. More reviews at the reviews index.