Modulis 2 · Security and practice

3/9

DeFi Risks

Video · 2 min.

DeFi risks

Luke learns that 'old' does not mean 'risk-free': in April 2026 Aave depositors took losses because of someone else's collateral (rsETH) whose backing broke on a bridge, not because of Aave's own code. DeFi is open water: smart-contract bugs, falling prices, exchange failures, no deposit insurance, and scammers.

Narrated, with light music

The full lesson with numbers and rules is below.

Watch on YouTube

This is an important lesson in the whole course. DeFi offers higher returns than banks — and when you see a bigger number, it's worth asking yourself: "what risk am I being paid for here?" In this lesson you'll get to know every risk type well enough to understand how it works and how to protect yourself from it.

Smart Contract Risks

A smart contract — like an automat that holds money

Picture a vending machine: you put a coin in, it hands you the item, and no shopkeeper is involved — the rules live inside the machine. A smart contract is that same kind of machine, except it manages your money. If it works correctly, everything's fine. But if it has a bug, someone can exploit it and "drain" all the funds.

A smart contract is a program. Programs are written by people. And people make mistakes. Here are the main types of risk — and how to spot each one:

1. Bug (a code error)

Bug — the simplest and most common risk

A bug is an error in the program's code. Like a typo in a dictation exercise, except here a mistake can cost millions.

Example: in 2023, Euler Finance lost $197M to a single bug in one function. One line of code — and the whole protocol was drained.

How to protect yourself: only use protocols that have been checked by independent experts (auditors) and have run for a long time without problems.

2. Reentrancy attack

Reentrancy — an exploit based on repeating an action

Imagine an ATM that hands out cash first and only checks the account balance afterward. If you can repeat the withdrawal between those two steps — you can drain the ATM.

Reentrancy works in a similar way — the program gets called again and again before it manages to update its own records.

Example: in 2016, "The DAO" lost $60M in ETH. It was the first major DeFi incident.

How to protect yourself: modern protocols have safeguards against this. But always check whether a protocol has been audited.

3. Oracle manipulation (faking the price feed)

Oracle — when someone fakes the price

DeFi programs don't know on their own what SOL or ETH costs. They rely on oracles — external price feeds. Think of it as a price board. If someone alters the price on the board, the program makes the wrong decisions.

Example: in 2022, someone faked the price on Mango Markets (Solana) and borrowed $114M — more than they were entitled to.

How to protect yourself: reliable oracles (Chainlink, Pyth) pull from many different price sources. They're hard to fool.

4. Flash loan attack (an instant loan)

Flash loan — an attack with no starting capital

Imagine you could borrow millions of dollars with no collateral at all — on the condition that you pay it back within a few seconds. That's a flash loan — a legitimate DeFi feature.

But hackers use it as a tool: borrow a huge sum → manipulate a price → pocket the profit → repay the loan. All of it happens within seconds.

Examples: Cream Finance ($130M), Pancake Bunny ($45M), Harvest Finance ($34M) — all fell victim to flash loan attacks.

How to protect yourself: good protocols have safeguards. But it remains a hard problem to solve completely.

5. Upgrade risk (when the program can change)

Upgrade risk — when the rules can change

Most DeFi programs can be upgraded (their code can be changed). That's useful for fixing bugs. But if someone took control, they could rewrite the program and steal all the funds.

How to protect yourself: check whether the protocol has a timelock (a delay before any change takes effect) and multi-sig (approval required from multiple people). That means one person alone can't change the program.

Rating Protocol Security

Not all protocols are equally secure. Here's a visual map of our strategy's protocols — by security (audits + age) and liquidity (TVL + ease of exit):

Protocol risk matrix

Colors (chain)

solana

ethereum

CEX (centralized)

Bridge

Bubble size

= TVL size

Above — each protocol as a bubble. Size = TVL (total value locked). Color = chain. The ideal quadrant is top-right (secure + liquid) — that's where Aave and Uniswap sit. The danger zone is bottom-right (liquid, but not secure) — that's usually where new, hyped-up protocols end up.

Our strategy's protocols

Assessments verified 2026-09. Protocol TVL and risk ratings shift over time — check current figures on defillama.com.

Protocol	Chain	Audited?	Running since	Risk tags
Aave	ETH	Many audits	Since 2020	AUDITED
Uniswap V3	ETH	Many audits	Since 2021	AUDITED
Orca	SOL	A few audits	Since 2021	AUDITED
Kamino	SOL	Many audits, solid TVL	Since 2022	AUDITED
Deribit	CEX	Centralized exchange	Since 2016	CEX
Meteora	SOL	Audited, but newer	Since 2022	NEW SC RISK
Uniswap V4	ETH	Audited, live since January 2025 — shorter history than V3	Since 2025	NEW SC RISK

What do the risk tags mean?

Tag	What it means	How risky
AUDITED	Many checks, long track record with no problems	Low risk
NEW	Newer protocol — shorter track record, less tested over time	Medium risk
SC RISK	Code is more complex than average	Medium-to-high risk
CEX	Centralized exchange — not DeFi. You depend on the company	Medium risk
BRIDGE	A bridge between networks. Historically the most exploited component	High risk

The Biggest DeFi Hacks in History

This table shows how large the real-world losses have been:

Year	Project	Loss	Cause	Type
2022	Ronin Bridge (Axie)	$625M	Validators taken over	Bridge
2021	Poly Network	$611M	Cross-chain code bug	Bridge
2022	Wormhole Bridge	$326M	Signature verification bug	Bridge
2022	Nomad Bridge	$190M	Initialization bug	Bridge
2023	Euler Finance	$197M	Bug in a single function	Lending
2023	Mixin Network	$200M	Database hack	Infrastructure
2022	Mango Markets	$114M	Price manipulation	DEX/Lending
2021	Cream Finance	$130M	Flash loan + oracle attack	Lending
2025	Cetus (Sui)	$223M	Concentrated liquidity math bug	DEX/LP
2025	Balancer V2	$128M	Rounding error in calculations	DEX
2026	Drift (Solana)	~$285M	Admin keys taken over after ~6 months of social engineering	Lending
2026	KelpDAO rsETH bridge	~$292M	Single-verifier (DVN) configuration on the bridge	Bridge

Notice two patterns

First: bridge protocols (bridges between networks) dominate the list of biggest losses. Bridges connect two different networks — that's technically very complex, and complexity is where most of the exploits live.

Our strategy doesn't use bridges directly. SOL positions stay on the Solana network (Kamino, Orca). ETH positions stay on the Ethereum/Arbitrum network (Aave). That reduces this risk.

Second: the most recent large losses were not code bugs. In the Drift case in 2026 the code was never broken — attackers spent roughly six months on social engineering and took over the human-held admin keys. So "audited and running for years" is no longer full proof of safety: you also have to look at who holds the keys.

A sixth kind of risk: admin key takeover and social engineering

When it's the people, not the code, that breaks

Almost every protocol has a group of people (a multi-sig or a "security council") who can upgrade the code or pause the protocol. If an attacker tricks enough of those key holders, no code bug is needed at all.

Example: in April 2026 Drift (Solana) lost around $285M exactly this way — after a long, patient social engineering campaign against the key holders.

What to check: how many signatures a change requires (2-of-3 is weak, 5-of-9 is stronger), whether there is a timelock (a delay that gives you time to exit), and whether the key holders all sit inside one team.

A lending protocol's risk is more than its own code

When you deposit into a lending protocol you take on the risk of every collateral asset that protocol accepts. In April 2026 Aave depositors took real losses — not from a bug in Aave's code, but from a third-party collateral asset (rsETH) whose backing was broken on a bridge, leaving roughly $177M of bad debt.

So before depositing, don't only ask "is this protocol audited" — ask "what collateral does it accept, and what happens if one of those assets goes to zero".

Bridge Risk — Connections Between Networks

A bridge — like a ferry across a river

Imagine two islands (two networks). Each has its own currency. To move money from one island to the other, you use a ferry (a bridge). But if the ferry breaks down or someone hijacks it — your money disappears into the river.

A bridge is a protocol that moves assets between different networks — for example, from Ethereum to Solana.

Why is a bridge dangerous?

Complexity: it has to work correctly on two different networks at the same time
Lock-and-mint: the original tokens are locked on one side and new ones are minted on the other. If the originals are stolen, the new ones become worthless
Few guardians: many bridges are controlled by a small group of validators (e.g. Ronin had only 9, of which 5 were compromised)

Our strategy's bridge rule

Never move a large amount through a bridge in one transaction. If you need to move $10,000:

Move $100 first and confirm it arrived
Then move the rest in parts
Best of all: keep SOL positions on the Solana network, ETH positions on the Ethereum network, and avoid bridges altogether

Rugpull — When the Team Steals the Money

Rugpull — like a fraudulent pop-up shop

Imagine a new shop opens. It offers unbelievable discounts. People rush in and pay upfront. The next day, the shop is gone with all the money. That's a rugpull.

A rugpull is when a project's team deliberately builds a protocol just to collect investors' money and disappear.

How to spot a dangerous project

Anonymous team
If the team members aren't publicly known, the risk is much higher. Legitimate projects have public founders with LinkedIn profiles.
Anonymity doesn't automatically mean fraud (Bitcoin was created by an anonymous person). But for small projects, it's a big warning sign.
An unaudited smart contract
If the code hasn't been reviewed by independent experts (auditors), nobody knows whether it actually does what it claims. It might have a hidden backdoor — a secret function that lets the team take all the funds.
Unlocked liquidity
If the team can withdraw all the funds from the protocol at any time, they can simply walk away. Legitimate projects LOCK their liquidity so the team itself can't touch it.
Unrealistic promises
1000% APY, 'guaranteed returns', 'risk-free' — these are classic signs of fraud. Nothing in DeFi is guaranteed.
If it sounds too good to be true, it probably is.
Aggressive marketing
If a project invests more in influencers than in security and code, that's a very bad sign.
Unpublished code
If the program's code isn't publicly visible on Etherscan or Solscan, you can't see what it actually does. Legitimate projects always publish their code.

Liquidation Risk — When Your Collateral Gets Sold

When you borrow in DeFi, your collateral can be automatically sold if its value drops too far.

Liquidation — like a mortgage

Imagine you took out a bank loan and put up your house as collateral. If the house's value fell sharply, the bank could force you to sell it to cover the debt. In DeFi, this happens automatically and within seconds.

How does liquidation work? An example:

You deposit 126 SOL (example price $88) = $11,088 of collateral
You borrow 4,704 USDC (that's 42% of the collateral's value — called LTV)
The SOL price drops to $55 → your collateral is now worth only $6,930
LTV rises to 68% — Kamino's liquidation threshold for SOL collateral is 75%, so only 7 percentage points of buffer are left. That's no longer a "keep watching" zone, it's an "act now" zone
Liquidation kicks in once LTV goes above 75% — for this position that's roughly a $50 SOL price (4,704 USDC / (126 SOL × 0.75)). Only then does the program automatically sell part of your SOL

The threshold depends on the collateral, not on the protocol

There is no single "Kamino threshold". Every collateral type has its own:

SOL collateral: Max LTV 74% (the most you can borrow), liquidation threshold 75% (from here you can be liquidated)
USDC collateral: liquidation threshold 90%

The gap between 74% and 75% is deliberately small: borrow to the maximum and you are already standing in the liquidation zone. So a safe working level is not "whatever is allowed" but well below it.

Kamino Lend runs as several dozen separate markets whose parameters are set by different curators. Before you open a position, check the threshold in the specific market you are opening it in. These figures were verified on 2026-09-05 via Kamino's public API and do change over time.

The planned LTV color system on the platform

On the DeFi Risk OS platform (in development), the LTV indicator will be shown with colors:

Green (under 40%): safe, liquidation is far away
Yellow (40–55%): keep watching, but still fine
Orange (55–65%): danger zone, action needed
Red (over 65%): critical level, urgent action needed

Our strategy keeps LTV around ~42% — that leaves roughly a 44% price-drop buffer before the 75% threshold.

Why the platform will warn before $50

Above we used the plain LTV: debt / collateral value. That is essentially how Kamino itself calculates it.

The DeFi Risk OS platform will deliberately use a more conservative formula with two safety factors: LTV = (debt × 1.05) / (collateral × price × 0.90). The 1.05 leaves room for accrued interest, the 0.90 for oracle drift and a sudden price gap. Under that formula the same position would turn red at around $58 instead of $50.

That's not a contradiction — it's the size of the margin. Better to react a few dollars early than to find out about a liquidation after the fact.

Impermanent Loss — in LP Positions

When you provide liquidity (Orca, Meteora), you can experience impermanent loss — a reduction in your gains compared with simply holding the assets.

Impermanent loss — like a market-stall trade

Imagine you have 10 apples and 10 pears. You hand them over to a market trader's stall. If the price of apples suddenly doubles, you would have earned more just by holding onto your apples. The gap between "what you'd have if you'd done nothing" and "what you have now" is impermanent loss.

But the trader's stall earns fees! If the fees cover the loss, you're still ahead.

Why is concentrated liquidity's IL bigger?

In our strategy, liquidity is provided in a narrower price range. That means:

More fees (you earn more from a smaller range)
Bigger impermanent loss (if the price moves out of your range)

Fees vs. loss — the balance

In our strategy, the LP position earns ~68% per year in fees (a live, changing snapshot — not a forecast). For the strategy to break even, it needs at least ~46% per year. Right now the LP is earning 48% more than it needs (a coverage ratio of 1.48x) — that's a safety buffer.

Options Risk (Deribit)

Deribit is a centralized exchange — not DeFi. That means it carries a different kind of risk.

Exchange risks

The exchange can collapse — as happened with FTX in 2022
It can be banned in certain countries
It can be illiquid — hard to close a position when there are no buyers
Counterparty risk — if Deribit ad fails to pay out, there's nothing you can do about it

Options position risks

Position	Biggest risk	Worst-case scenario
Long Put (bought protection)	You lose the entire premium if the price doesn't fall enough	-$1,300 (the full amount paid)
Short Put (sold protection)	SOL's price falls sharply — you must buy at a high price	A large position around $60–65
Short Call (sold upside)	SOL's price rises sharply — you must sell cheap	You sell SOL at $160 (even though it's worth more)

How Much Money to Put Into DeFi?

The most important rule: never invest more than you can afford to lose.

The tables below are a general educational example, not personalized advice. If you're starting out, the right anchor is the bottom of the conservative profile, not the middle of it.

Conservative profile (cautious)

Your assets	DeFi share	Maximum amount
Under $10,000	0–5%	$0–500
$10,000–$50,000	5–10%	$500–5,000
$50,000–$200,000	10–15%	$5,000–30,000
Over $200,000	10–20%	$20,000–40,000

Moderate profile

Your assets	DeFi share	Maximum amount
Under $10,000	5–10%	$500–1,000
$10,000–$50,000	10–20%	$1,000–10,000
$50,000–$200,000	15–25%	$7,500–50,000
Over $200,000	15–30%	$30,000–60,000

Aggressive profile (bold)

Your assets	DeFi share	Maximum amount
Under $10,000	10–20%	$1,000–2,000
$10,000–$50,000	20–40%	$2,000–20,000
$50,000–$200,000	25–40%	$12,500–80,000
Over $200,000	20–40%	$40,000–80,000

The golden rule — how you decide

Only invest money you don't need for everyday life. DeFi isn't a bank with deposit insurance, so pick an amount such that even the worst-case scenario wouldn't change your day-to-day life. It's one sentence that protects you: I only invest what I can calmly walk away from.

Diversification — Don't Put All Your Eggs in One Basket

Even within DeFi, it's important to spread out risk:

Network diversification: use both Solana and Ethereum — not just one
Protocol diversification: use several protocols — Kamino + Orca + Deribit
Strategy diversification: use several strategies — A + B
Time diversification: don't invest everything at once — spread it over several weeks
Options diversification: hold protection (Put) alongside profit positions (Short Call)

Protocol Risk Map

Here's roughly how various places that hold money position themselves — by security and liquidity. Ideal: the top-right zone (secure + liquid).

Protocol risk matrix

Colors (chain)

solana

ethereum

CEX (centralized)

Bridge

Bubble size

= TVL size

This is an illustrative positioning for teaching purposes (like a "magic quadrant"), not exact data. "Bridge" and new, unaudited protocols sit lower (riskier); old, audited protocols with high liquidity sit in the top right.

How DeFi Risk OS Will Help You Manage These Risks

Our platform is being built to help you see and manage these risks in real time. Planned features:

LTV indicator — will show how close you are to liquidation (color-coded)
Delta indicator — will show how much your portfolio depends on price movement
Protection percentage — will show how much of your portfolio is protected by options
Alerts — will automatically flag when something's off
Scenario analysis — will show what would happen if the price fell -30% or rose +50%
Breakeven calculator — will show the minimum yield your LP needs to generate

Summary

Smart contract risks: code bugs, reentrancy, price manipulation, flash loans, upgrade risk
Bridge protocols are the most dangerous — avoid them where you can
Check a protocol's audit history, age, and team
Rugpull warning signs: anonymous, unaudited, unrealistic promises
Liquidation risk is managed by keeping LTV low (~42% in our strategy)
Impermanent loss is managed through higher fees
Only invest what you can afford to lose
On the DeFi Risk OS platform, all risk indicators will be visible in real time

Platform in development

Risk management on the DeFi Risk OS platform

Understand DeFi risks? Once the platform launches, you will be able to monitor and manage them in real time.

For now, all of this is done by hand, as shown in the lesson.

Quick check

Which protocols have historically suffered the BIGGEST losses from hacks?

Practice task

0 / 4

Size up protocol risk with your own eyes

Theory about risk fades — research you did with your own hands doesn't. Takes about 15 minutes, no funds need to go anywhere.

This is a research task — don't connect a wallet or invest anything anywhere. All the numbers and audits are public and need no login.

This is learning, not investing — use only small amounts you treat as tuition.

You just worked through the whole DeFi risk map — from smart contract bugs to liquidation — and you know how to manage each one. Most investors never learn this.

Spotting crypto scams Token approvals and revoking

Mark as read