How Price-Manipulation Attacks Drain DeFi Lenders

Decentralized finance (DeFi) lenders are designed to make borrowing automatic. A user deposits collateral, the protocol checks its market value through a price oracle, and a smart contract determines how much the user can borrow.

That automation is one of DeFi’s biggest strengths. It can also become a weakness when the price feeding the lending system can be manipulated.

In a price-manipulation attack, an attacker temporarily pushes the market price of an asset sharply higher or lower and takes advantage of a DeFi protocol that treats that distorted price as genuine. When the target is a lending protocol, the attacker may make weak collateral appear far more valuable than it really is, borrow valuable assets against it, and leave the protocol with bad debt once the manipulated price returns to normal.

Flash loans can make these attacks easier by giving attackers access to large amounts of temporary capital, but they are usually an amplifier rather than the underlying vulnerability.

What Is a DeFi Price-Manipulation Attack?

A DeFi price-manipulation attack happens when someone deliberately changes an asset’s observable market price and then uses the distorted price to profit from another protocol.

The distinction between market manipulation and an oracle exploit is important.

An oracle exploit can involve corrupting or compromising the mechanism responsible for delivering data. Market manipulation can happen even when the oracle itself works normally: the attacker manipulates the market the oracle observes, and the oracle faithfully reports that manipulated price. In practice, the two often overlap. A correctly implemented oracle can still be unsafe if it samples a thin pool, uses too short a window, or can be updated inside the same transaction as the exploit. 

Consider a token that normally trades at $10. If an attacker can temporarily push its observable price to $50, a lending protocol relying on that market may begin treating each token as $50 of collateral.

The attacker has not necessarily changed the token’s fundamental value. They have changed the price signal used by the smart contract. 

That difference can be enough to drain a lending market.

How DeFi Lending Normally Works

Most overcollateralized DeFi lending systems follow a relatively straightforward model.

Suppose Alice deposits $10,000 worth of ETH into a lending protocol. If the protocol allows her to borrow up to 70% of that collateral value, she may be able to borrow roughly $7,000.

The protocol must therefore continuously answer an important question:

What is Alice’s collateral worth right now?

Smart contracts cannot independently determine the economic value of ETH, BTC, or another token. They rely on an oracle or another pricing mechanism. If the price falls far enough, Alice’s position may become undercollateralized and eligible for liquidation.

If the price is artificially inflated, however, the opposite can happen: the protocol may allow someone to borrow far more than their collateral is actually worth. That is where price manipulation becomes dangerous.

How a Price-Manipulation Attack Drains a Lender

The exact transaction sequence varies between attacks, but the economic structure often looks like this:

Stage	What Happens	Why It Matters
Find weak liquidity	Attackers target a token or market whose price can be moved relatively cheaply.	Thin markets are easier to manipulate.
Obtain capital	The attacker uses owned funds, borrowed assets or a flash loan.	Large trades become possible without permanently committing equivalent capital.
Manipulate price	Large swaps or trades push the target asset sharply higher or lower.	The protocol’s pricing system may observe the distorted price.
Exploit valuation	Inflated collateral is deposited or an existing position rises in apparent value.	Borrowing power can increase dramatically.
Borrow assets	The attacker withdraws stablecoins, ETH, BTC, or other liquid assets.	Real assets leave the lending pool.
Price normalizes	Manipulative trading stops and the token returns toward its normal market value.	The collateral can become worth far less than the outstanding loan.
Bad debt remains	Liquidating the collateral cannot fully repay the loan.	Depositors or the protocol treasury ultimately absorb the shortfall.

The critical failure is therefore not simply that a token price moved. It is that the lending protocol allowed a short-lived or unreliable price to determine how much real liquidity could leave the system.

Why Low-Liquidity Tokens Are Especially Dangerous

Imagine two assets.

Asset A trades billions of dollars per day across several deep exchanges. Asset B trades mainly in one small decentralized exchange pool containing limited liquidity.

Moving Asset A by 50% could require enormous amounts of capital and attract immediate arbitrage activity. Moving Asset B by the same percentage may be substantially easier.

This becomes dangerous when a lending protocol accepts Asset B as collateral while giving borrowers access to highly liquid assets such as USDC, USDT, ETH, or WBTC.

The protocol effectively creates an economic bridge between weak liquidity on the collateral side and deep liquidity on the borrowing side.

An attacker may manipulate the weaker side and extract value from the stronger side.

This is why collateral selection is a security decision, not simply a product-growth decision.

How Flash Loans Amplify Price Manipulation

Flash loans are uncollateralized DeFi loans that must generally be borrowed and repaid within the same blockchain transaction. If the borrower cannot repay the loan plus required fees by the end of the transaction, the entire transaction reverts.

Legitimate traders use flash loans for arbitrage, liquidations and other capital-efficient strategies. But the same mechanism can provide attackers with temporary access to large amounts of capital.

Research into early DeFi attacks found that flash loans gave attackers access to large amounts of capital without upfront collateral, allowing them to amplify exploits and chain actions across multiple protocols within a single transaction.

An attacker can theoretically borrow millions of dollars, manipulate a liquidity pool, interact with a vulnerable lender, extract a profit, reverse part of the trade and repay the flash loan—all as part of an atomic transaction.

The attacker therefore does not necessarily need millions of dollars sitting in a wallet before beginning. But calling every such incident a “flash-loan attack” can be misleading.

The flash loan supplies capital. The vulnerable pricing or risk model usually creates the exploit opportunity.

Removing flash loans alone would not solve the underlying problem if someone with sufficient capital could reproduce the manipulation.

Case Study: Inverse Finance’s $15.6 Million Price Manipulation

One of the clearest examples occurred at Inverse Finance in April 2022.

According to Inverse Finance, the Keep3r time-weighted average price oracle used for its INV token was manipulated through trading involving the INV/WETH market on SushiSwap. Inverse’s post-mortem also described heavy trading in the relatively thin INV/DOLA SushiSwap pool and an error in the TWAP sampling method, including a window-size issue in the INV-DOLA averaged oracle. 

The manipulation sharply increased the reported value of INV (Inverse said the oracle price briefly printed around $20,926). That inflated INV could then support much larger loans through Anchor, the lending market now associated with Inverse’s Frontier product.

The resulting borrowing involved approximately $15.6 million worth of DOLA, ETH, WBTC, and YFI, according to Inverse Finance — specifically about 1,588 ETH, 94 WBTC, 4 million DOLA and 39 YFI. Inverse stated that the incident was not a flash-loan attack and was not caused by a bug in its lending-contract or front-end code. Independent analyses put the economic loss in a similar range, around $14.5 million to $15.6 million, depending on how token prices are marked.

The case demonstrates an important point: using a TWAP instead of a single instantaneous DEX price improves manipulation resistance, but a TWAP is not automatically immune to attack.

If liquidity is sufficiently weak and an attacker has enough capital, the attacker may sustain the manipulation long enough to influence an averaged price. A short or poorly sampled averaging window can collapse that protection even further. 

Indeed, research continues to examine manipulation-resistant oracle designs because TWAP-based systems still involve trade-offs between responsiveness and manipulation resistance.

Also Read: Inverse Finance Faces $240K Loss in DOLA Manipulation Alert

Case Study: Mango Markets and Inflated Borrowing Power

The October 11, 2022 Mango Markets incident provides another striking example of manipulated prices affecting borrowing capacity.

Mango was not simply a traditional DeFi lender; it combined spot markets, perpetual futures, margin and borrowing. But the exploit illustrated the same economic vulnerability.

According to U.S. prosecutors, large purchases of MNGO helped push the relevant price dramatically higher. The price of MNGO perpetuals rose roughly 1,300% within about 20 minutes. The SEC separately alleged that MNGO’s spot price on the venue rose more than 2,200% during the episode.

The resulting increase in the apparent value of the attacker’s position dramatically increased the account’s borrowing capacity. Roughly $110 million in crypto assets was subsequently withdrawn from Mango Markets. Other contemporaneous tallies put the drain at about $114 million to $117 million; the SEC used a figure of approximately $116 million. Avraham Eisenberg later returned about $67 million under a DAO settlement and kept about $47 million.

Security researchers similarly highlighted low MNGO liquidity as a major factor that made manipulating its market economically feasible.

The incident later became an important U.S. legal test involving DeFi market manipulation. A jury convicted trader Avraham Eisenberg in April 2024 of commodities fraud, commodities manipulation and wire fraud. On May 23, 2025, U.S. District Judge Arun Subramanian vacated the commodities-fraud and manipulation convictions, finding that the government had not established venue in the Southern District of New York, and entered a judgment of acquittal on the wire-fraud count, concluding there was insufficient evidence that Eisenberg made a material false representation to a permissionless, automatic protocol. U.S. prosecutors appealed; as of 2026 that appeal remained pending. Civil cases from the SEC and CFTC also followed. Mango Markets later wound down operations in January 2025 after an SEC settlement.

For DeFi risk managers, however, the technical lesson is independent of that legal dispute: collateral value should not be allowed to expand borrowing power faster than the underlying market can safely support it.

Also Read: Mango Markets Announces Shutdown After $117M Hack

Case Study: bZx and the Early Flash-Loan Warning

Price-manipulation risk was visible much earlier.

In February 2020, attacks involving bZx became some of DeFi’s first major demonstrations of how composability, flash loans, automated markets and lending infrastructure could interact unexpectedly.

On February 15, an attacker used a large dYdX flash loan, borrowed WBTC on Compound, and pushed the Uniswap WBTC/ETH price while opening a leveraged position on bZx/Fulcrum. Profit estimates for that first incident were on the order of $300,000 to $350,000. A second attack on February 18 manipulated sUSD pricing through Kyber and Uniswap, then used the inflated sUSD as collateral to borrow extra ETH from bZx. Net proceeds there were commonly estimated around $600,000 to $645,000.

Academic researchers studying the incidents showed how flash loans allowed complicated strategies to be financed atomically, while later security research described an attack in which an AMM price was manipulated and overvalued collateral was used against a lending system.

The lesson has remained relevant: a lending protocol cannot evaluate its security in isolation.

Its safety may depend on the liquidity of a DEX, the behavior of an oracle, another protocol’s flash-loan liquidity and the ability of arbitrageurs to restore market prices.

How DeFi Lenders Can Reduce Price-Manipulation Risk

There is no single defense capable of eliminating every manipulation attack. The strongest systems use several layers of protection.

First, protocols can use price feeds that aggregate information across multiple high-quality data sources rather than trusting one thin market. Source diversity makes it harder for a trade on one venue to determine the protocol’s entire collateral valuation. Where a TWAP or VWAP is used, the window, sampling cadence and reference pool should be treated as security parameters, not implementation details.

Collateral parameters also matter. A volatile or thinly traded asset should generally have more conservative loan-to-value ratios, lower borrowing limits and tighter supply caps than a deeply liquid asset.

Debt ceilings can limit the amount of liquidity that can be extracted against one collateral type even if its price is manipulated.

Protocols can also introduce circuit breakers that pause or restrict borrowing when prices move beyond predefined thresholds, when oracle updates become stale or when multiple pricing sources diverge sharply. Research into lending-protocol defenses has similarly explored rejecting or constraining oracle prices when movements exceed acceptable ranges.

Finally, risk teams need to examine market depth, not just token market capitalization.

A token may have a billion-dollar quoted valuation while having very little executable liquidity around its current market price.

For lending security, the relevant question is not only “What is this asset worth?” It is also: “How much money would it cost someone to make our protocol believe it is worth something else?”

Can Users Identify High-Risk DeFi Lenders?

Individual users cannot audit every smart contract or reconstruct every oracle configuration, but they can still evaluate obvious risk signals.

A lender that accepts highly illiquid tokens as collateral while allowing large loans against them deserves more scrutiny. The same applies when a protocol depends heavily on one trading venue, allows very aggressive loan-to-value ratios or provides little public information about collateral caps and oracle design.

Security audits are useful, but price manipulation is partly an economic-design problem, not only a coding problem. A contract can operate exactly as written and still produce catastrophic results if its economic assumptions are wrong.

Why Price-Manipulation Attacks Matter for DeFi

Price manipulation exposes one of the fundamental challenges of decentralized lending.

Blockchains can verify exactly how many tokens an address owns. They cannot independently verify what those tokens are economically worth. DeFi therefore depends on mechanisms that translate changing external markets into numbers that smart contracts can trust.

When that pricing layer is poorly designed, or when protocols allow more borrowing than underlying liquidity can safely support, attackers can transform temporary price distortion into permanent protocol losses.

The most resilient lenders therefore treat oracle architecture, collateral liquidity, borrowing limits and market manipulation as interconnected parts of the same security model.

In DeFi lending, protecting the code is only half the job. Protocols must also protect the price assumptions that tell the code what assets are worth.

Frequently Asked Questions