Gas and slippage are both real costs — but most traders obsess over one and ignore the other. The optimal trade size sits at their exact crossover. The approximation formula most tools use overstates that size by up to 2×. Here's the exact math.
Read →A health factor of 1.3 sounds safe until you run the actual numbers. With WBTC as collateral (LT = 75%), HF 1.3 means a 23% price drop triggers liquidation. This is how to think about it correctly — and what buffer you actually need.
Read →Opening a 10× long on GMX feels like a $10 decision on a $1,000 position. After 48 hours, borrow fees alone have taken $1.92 — before the position moves a dollar. The math on why short holds on leveraged perps are better than you think.
Read →Looping amplifies your BTC exposure cheaply — but the fee drag and liquidation buffer compress simultaneously. With WBTC at max LTV, four loops give 2.27× exposure but only a 10.7% BTC drop to liquidation. Here's the math behind the risk.
Read →A ±10% range gives a 10× capital efficiency multiplier — but on ETH, it gets breached 28% of weeks. The fee APR your pool explorer shows assumes you're always in range. The actual math on range selection and out-of-range risk.
Read →Both protocols let you borrow USDC against WBTC — but the risk model, liquidation mechanics, and yield distribution are fundamentally different. Morpho's LLTV can reach 86% vs Aave's 75% LT, but the liquidation works differently and there's no supply-side safety buffer. A practical comparison.
Read →GMX's borrow fee runs 24/7 — even when your position is making money. A 10× ETH long held for 7 days pays ~0.84% in borrow fees alone, before any market move. Here's how to account for it in your strategy and when to avoid it entirely.
Read →Three stablecoins, three completely different risk models — and three different reasons to hold them. USDC has regulatory clarity. USDe has the highest yield. USD0 bridges both. A structured comparison across depeg risk, counterparty, liquidity, and protocol safety.
Read →Same WBTC, same borrow target — two very different results. Morpho's LLTV reaches 86% vs Aave's 73% LTV on Arbitrum, lower liquidation penalties, and cheaper borrow rates. When Aave wins and when Morpho does — with the full parameter table for June 2026.
Read →Your health factor is not static — borrow interest grows your debt daily even when prices don't move. At 6.5% APY, a $40,000 debt becomes $42,600 in one year, raising your liquidation price by $3,300 without you noticing. The full formula, the zones, and the thresholds to set.
Read →Liquidation is not bad luck — it's a predictable outcome of ignoring specific thresholds. On Aave V3, the liquidation penalty is 10% for WBTC: on a $50,000 position that's $5,000 lost in seconds. The full playbook to avoid it, from opening rules to monitoring habits.
Read →1inch, Camelot, Paraswap, CoW Swap — the right choice depends entirely on your trade size. A DeFi user doing two $10,000 swaps per week at 0.5% slippage loses $5,200/year vs $1,040 with optimized routing. The breakdown by size, MEV protection and hidden costs.
Read →Your Uniswap trades are costing you more than you think
Gas and slippage work against you from opposite directions. Most traders treat gas as fixed overhead and ignore how it interacts with trade size — that's leaving money on the table every single trade.
The U-curve problem
Every AMM trade has a total cost that forms a U-shape as a function of size. Trade too small: gas dominates, and you're paying a large fixed overhead relative to any gains. Trade too large: slippage dominates, and you're moving the pool price against yourself. The bottom of the U — the optimal size — is where gas cost and slippage cost are equal.
This isn't intuition. It's a consequence of the constant-product formula (x·y=k). For a pool of depth R (single-sided TVL), a trade of size A produces slippage of exactly A/(R+A). Gas is a fixed cost G in dollars. The total cost as a fraction of trade size is:
Minimise this with respect to A and you get the optimal size:
The 2× error in common tools
Most online calculators and rule-of-thumb formulas use √(2·G·TVL) instead. The extra factor of 2 comes from using TVL directly instead of single-sided depth R = TVL/2. The result: they tell you the optimal trade is twice as large as it actually is.
In a $1M TVL pool with $3 gas: the correct optimal is √(3 × 500,000) = $1,225. The approximation gives √(2 × 3 × 1,000,000) = $2,449. Trade at $2,449 and you're paying roughly twice the optimal slippage.
Practical numbers
On Ethereum mainnet with 8 gwei gas (~$4 per swap) in a $2M WBTC/USDC pool: A* ≈ $1,414. On Arbitrum with 0.01 gwei gas (~$0.004 per swap) in the same pool: A* ≈ $45. The depth of the pool matters as much as the gas — on Arbitrum, tiny gas means even a small trade is "gas-efficient," so the optimal barely moves from the minimum gas-limited size.
What to actually do
Load your target pool in the Trade Size Optimizer. It fetches pool depth and fee live from GeckoTerminal, applies the exact constant-product formula, and shows you the bottom of the U. If you're trading a round number far from that bottom, you're either overpaying gas (too small) or overpaying slippage (too large).
Open Trade Size Optimizer →Health factor 1.3 — how much margin is that, really?
A health factor of 1.3 is routinely called "safe" in DeFi communities. Whether it actually is depends entirely on what's in your collateral — and the math is less forgiving than the label suggests.
What health factor measures
Aave's health factor is a weighted ratio: the sum of collateral values adjusted by liquidation thresholds, divided by total debt. The formula read from the contract is:
When HF drops to 1.0, a liquidator can seize up to 50% of your collateral in a single call, receiving a bonus (typically 5–15%) on top of the debt they repay. Below 1.0 is not a place you want to be.
From HF to price drop
The question traders actually need answered is: "how far can my collateral fall before I'm liquidated?" If you have a single collateral asset with liquidation threshold LT and total debt D, the liquidation price is:
The margin until liquidation — expressed as a percentage drop in collateral price — is:
Wait, simpler: if HF = 1.3, you need HF to drop by 0.3 to reach 1.0. Since HF scales linearly with collateral price, you need collateral to drop by 0.3/1.3 = 23.1%.
The LT factor people miss
The margin calculation above assumes uniform collateral. In practice, each asset has its own LT. WBTC on Aave V3 Arbitrum: LT = 75%. ETH: LT = 82.5%. USDC: LT = 85%. A mixed portfolio doesn't liquidate all at once — the weighted average LT determines the actual threshold. This is why reading your HF directly from the contract is more accurate than any manual estimate.
Safe buffer rules of thumb
For volatile collateral (BTC, ETH): maintain HF > 1.5, which gives roughly 33% downside buffer. For stablecoin or LST collateral: HF > 1.2 is more defensible. Set a mental alert at HF 1.25 for volatile assets — by the time you notice a sharp drop, you may already be below 1.2.
Check your position live →GMX perpetuals: the borrow fee is eating your PnL
The opening fee on GMX is visible and easy to account for. The borrow fee is continuous, compounding, and invisible to most traders until they finally close. Here's the actual math — with numbers.
The fee structure, clearly
GMX V2 charges two main fees on leveraged positions. First, the opening fee: 0.07% of position size, charged on open and again on close — a round-trip costs 0.14%. Second, the borrow fee: approximately 0.01%/hour on the borrowed portion of the position, continuously accruing.
The borrowed portion is position size minus collateral. For a $10,000 position with $1,000 collateral (10× leverage), borrowed = $9,000. The borrow fee accrues at roughly:
Running the numbers over time
Opening a 10× BTC long for $10,000 position size with $1,000 collateral:
Opening fee: $10,000 × 0.07% = $7. After 1 hour: +$0.90 borrow fee, total overhead = $7.90. After 24 hours: +$21.60, total = $28.60 — nearly 3% of your collateral. After 48 hours: +$43.20, total = $50.20 — over 5% of $1,000 collateral.
The liquidation price drift
Borrow fees reduce your effective collateral continuously. This means your liquidation price drifts toward your entry the longer you hold. The liquidation formula:
After 24 hours on the 10× long above, fees = $28.60, so effective collateral drops from $1,000 to $971.40. Liquidation price moves ~0.14% closer to entry. Negligible on short holds — significant on multi-day positions.
What this means for strategy
GMX perpetuals are not designed for multi-day holds at high leverage. They're most cost-effective for short-term directional trades (hours, not days) where the fee overhead is a small fraction of the expected move. If you're planning to hold a leveraged position for days, the borrow fee drag makes lower leverage or a lending loop (Aave) more capital-efficient.
Model your GMX fees →Leverage loops on Aave: the real cost of recursive borrowing
Looping — depositing, borrowing, redepositing — is the standard way to build leveraged exposure inside a lending protocol. The math is clean, but the fee drag and liquidation risk accumulate faster than most users model.
How a leverage loop works
Start with $1,000 of WBTC as collateral at 67% max LTV on Aave V3 Arbitrum. You borrow $670 of USDC, swap it back to WBTC, deposit again. Now your collateral is $1,670. Borrow 67% again: +$1,119. Each round multiplies by LTV. After N loops, total exposure is:
With LTV = 0.67 and 4 loops: exposure ≈ $1,000 × 2.27 = $2,270. Your $1,000 controls $2,270 of BTC — 2.27× leverage without touching a perpetual.
Entry cost per loop
Each loop requires a USDC→WBTC swap. On a 0.05% fee pool, four loops cost $670 × 0.05% + $449 × 0.05% + $301 × 0.05% + $201 × 0.05% = $0.81 in swap fees. At 0.1% slippage per swap, add $1.62. Gas on Arbitrum: ~$0.50 per transaction × 8 transactions = $4. Total entry cost: ~$6.43 on $1,000 initial capital, or 0.64%. Cheap — but not free, and it must be paid again on exit.
Net APY: the uncomfortable math
A loop earns supply APY on the full collateral stack and pays borrow APY on all borrowed capital. With WBTC supply APY 0.2%, USDC borrow APY 4.5%, and 2.27× leverage:
This is a negative-carry position. You're paying 5.27% per year to hold 2.27× BTC exposure. The bet is directional: it only pays off if BTC appreciates faster than the carry cost. That's historically true in bull markets — and false in flat or down periods.
The liquidation buffer that doesn't change with loops
The key insight most guides miss: the liquidation buffer is determined by the LTV/LT gap, not by the number of loops. With WBTC max LTV = 67% and liquidation threshold LT = 75%, liquidation triggers when collateral value drops enough that:
Whether you run 2 loops or 8 loops, a 10.7% BTC price drop liquidates you. More loops = more exposure, identical buffer. Running fewer loops does not increase safety — it just reduces your BTC upside.
Uniswap V3 ranges: why width matters more than entry
Concentrated liquidity makes V3 up to 20× more capital-efficient than V2 — but only while price stays in range. The trade-off is quantifiable, and most LPs pick widths by feel rather than by the numbers.
The capital efficiency multiplier
In V2, your liquidity is spread from price 0 to ∞. In V3, you concentrate it between Pa and Pb. The efficiency multiplier — fees earned per dollar vs. an equivalent V2 position — is approximately:
For a ±10% range (Pa = 0.9×P, Pb = 1.1×P): multiplier ≈ 10.5×. For ±20%: ≈ 5.4×. For ±5%: ≈ 21×. Narrow ranges are dramatically more capital-efficient — while in range.
Why displayed APRs are misleading
Pool explorers (DefiLlama, GeckoTerminal, Uniswap UI) show fee APR based on 24h volume divided by total in-range liquidity. This assumes your position is always in range. It isn't. The correct APR is:
For ETH/USDC with a ±10% range, the breach probability over 7 days (based on ETH's ~4%/day log-volatility) is roughly 28%. Your position is out of range ~28% of the time — earning nothing. Real APR is 72% of displayed. A "45% APR" pool actually yields ~32%.
Breach probability by width
For ETH (σ ≈ 4%/day), approximate weekly breach probability by range width:
±5% range: ~62% breach probability per week. ±10%: ~28%. ±20%: ~7%. ±30%: ~2.5%. For BTC (σ ≈ 3%/day), multiply breach times by ~0.6. For a stablecoin pair (σ ≈ 0.1%/day), even ±0.5% ranges are safe for weeks.
Practical range selection
For volatile pairs (ETH, BTC) with active management: ±20% gives multiplier ~5× and ~7% weekly breach — manageable. For set-and-forget positions: ±30–50% is more realistic. For stablecoin pairs: ±0.5–1% is standard practice, as breach risk is near zero and the multiplier can reach 100×+.
The V3 Range Calculator computes impermanent loss, fee income estimate, and the capital efficiency multiplier for any price range. It also feeds into the IL Calculator to model the full position P&L including fees vs. holding.
Calculate your V3 range →Morpho Blue vs Aave V3: which one for your WBTC?
Same collateral, different risk models. Understanding the tradeoffs before you borrow.
The setup looks similar — the mechanics aren't
Both Aave V3 and Morpho Blue let you deposit WBTC and borrow USDC. The surface-level similarity ends there. Aave is a pooled protocol with governance-managed risk parameters; Morpho Blue is an immutable, isolated-market protocol where every parameter is fixed at market creation. The difference matters when markets move fast.
Aave V3: LTV 70%, LT 75%, buffer built in
On Aave V3 Arbitrum, WBTC has a max LTV of 70% (how much you can borrow) and a liquidation threshold of 75% (when liquidation triggers). The 5% buffer is your safety margin — the protocol can begin liquidating before you're technically insolvent, giving liquidators time to act. The supply-side also benefits from Aave's Safety Module (staked AAVE) as a last-resort backstop for bad debt.
Liquidation is partial by default (up to 50% of debt per call) with a 10% liquidation bonus for WBTC. You can exit a position, add collateral, or repay during a liquidation — the system is designed to recover iteratively, not in one shot.
Morpho Blue: LLTV 86%, no buffer, different rules
The primary WBTC/USDC market on Morpho Blue uses an LLTV of 86% — significantly higher than Aave's LT. However, Morpho uses a single parameter for both "max borrow" and "liquidation trigger": once your debt reaches 86% of your WBTC collateral value, liquidation begins immediately.
Morpho liquidation is also different in mechanics: liquidators can close 100% of the position in one transaction, not just 50%. There's no Safety Module — bad debt, if any occurs, is absorbed by the lending vault (MetaMorpho depositors) rather than a protocol backstop. Each Morpho market is also fully isolated: a crash in one market can't contaminate another.
Which gives more borrowing capacity?
On $100,000 WBTC (at current prices):
- Aave V3: Max borrow = $70,000 USDC. Liquidation triggers when WBTC drops to $80,000 (−20%).
- Morpho Blue (86% LLTV): Max borrow = $86,000 USDC. Liquidation triggers when WBTC drops such that debt ≥ 86% of collateral — with $86K debt, that's the moment collateral = $100K = immediately. You need to borrow below 86% for any buffer at all.
In practice, most Morpho borrowers target 70–80% utilization (of the LLTV), giving them a real buffer. But at 80% utilization vs 86% LLTV, a 7.2% WBTC drop liquidates the position. On Aave at 70% LTV vs 75% LT, you have a 6.67% buffer — similar in absolute terms but with explicit governance protections around the parameters.
Yield for lenders
Morpho Blue's supply rate for the lending vault depends on the curator's strategy. Since each market is isolated, utilization in the WBTC/USDC market directly sets the rate curve. MetaMorpho vaults (e.g., Steakhouse USDC) aggregate across multiple markets, smoothing volatility in rates.
Aave's supply rate for USDC on Arbitrum is typically 3–7% APY in normal conditions, rising sharply above the optimal utilization threshold. Morpho rates can be higher or lower depending on curator allocation and demand in specific markets.
The honest answer: it depends on your risk tolerance
Use Aave V3 if you want: governance-protected parameters, partial liquidation mechanics, and a protocol that has operated at $10B+ TVL for 3+ years. The Safety Module is a genuine backstop. E-Mode gives you high LTV on ETH-correlated pairs specifically.
Use Morpho Blue if you want: higher LLTV (more borrowing power per unit of collateral), immutable market parameters (no governance can change the LLTV on you), and you're comfortable managing tighter liquidation margins with more active monitoring.
Open Morpho Calculator → Open Aave Calculator →GMX funding rate: the hidden cost destroying your perps
It runs 24/7. It compounds. And most traders forget to model it before opening a position.
What the funding rate actually is on GMX
GMX V2 doesn't use the traditional perpetual funding rate model (longs pay shorts or vice versa, based on premium). Instead, it uses a borrow fee charged to both longs and shorts, proportional to their position size. The rate depends on utilization of the pool for that market.
Typical borrow fee on GMX V2 ETH/USD: ~0.012% per hour for the dominant side when utilization is 80%+. That's 0.288% per day, or ~105% per year on the borrowed portion. Most traders mentally model this as "small" — it's not small if you hold for more than a day or two.
The math on a real position
Let's take a 10× ETH long with $1,000 collateral → $10,000 position size. Borrow fee of 0.012%/hour applies to the position size ($10,000), not the collateral:
- Hourly cost: $10,000 × 0.012% = $1.20/hour
- Daily cost: $28.80
- Weekly cost: $201.60 — 20.2% of your collateral in 7 days
- Monthly cost: $864 — your entire collateral is consumed in ~35 days
ETH needs to move up 2.8% in the first 24 hours just to break even on borrow fees alone — before the open/close fee (0.05% × 2 = 0.1% round trip = $10 more on a $10K position).
Funding vs traditional perp exchange
On a CEX (Binance, Bybit), the funding rate is charged every 8 hours and can be negative (shorts pay longs). At 0.01% per 8h (typical bullish market), longs pay 3× per day = 0.03%/day = 11% annualized. This is lower than GMX's effective rate — but CEX funding can also spike to 0.1–0.3% per 8h during crypto mania, far exceeding GMX's flat borrow.
The key difference: GMX borrow fee is always a cost (both sides pay). CEX funding can be income (if you're on the paid side). For longs during bullish markets, GMX's borrow fee is predictable — CEX funding is not.
How to factor it into your trade plan
Before opening any GMX position that you plan to hold more than 12 hours:
- Estimate hours in trade (be honest — "I'll close at the right time" usually means you won't)
- Compute total borrow cost:
position_size × hourly_rate × hours - Add to open + close fees (0.1% round trip on position size)
- Your break-even price move =
total_fees / position_size × 100%
For a 10× position held 48 hours: break-even = 0.1% (fees) + 0.576% (borrow) = 0.676% price move needed just to cover costs. This isn't a trading edge problem — it's a cost accounting problem.
When GMX perps make sense
Short-term directional plays (<24h) where the expected price move far exceeds the fee structure. Hedging a spot position where the borrow cost is acceptable insurance premium. Positions where you expect a fast, clean move — not "I'll ride the trend." The GMX Calculator models all these costs before you open a position.
Model your GMX position →Stablecoin risk ranking 2026: USDC vs USDe vs USD0
Three stablecoins with completely different risk models — and different roles in a DeFi portfolio.
Why stablecoin risk isn't uniform
A stablecoin that holds $1 in a bank is not the same risk as a stablecoin that holds ETH + short perpetuals to synthetically construct $1. Both trade near $1 on most days. The difference only becomes visible during stress. LUNA/UST demonstrated this in May 2022 — an algorithmic stablecoin that was $18B in supply one week and zero the next. Understanding the underlying mechanism is not optional.
USDC: the known-risk leader
Mechanism: Each USDC is backed 1:1 by dollars and short-duration US Treasuries held at US-regulated financial institutions. Circle (the issuer) publishes monthly attestations from Deloitte.
Depeg risk: Very low in normal conditions. March 2023 saw a 10-hour depeg to $0.87 when $3.3B of USDC reserves were stuck at Silicon Valley Bank during its collapse. The depeg resolved when Circle confirmed federal intervention. This was a counterparty failure — not a mechanism failure.
Main risk: Regulatory action against Circle, or a custodian failure. A US government T-bill default would also impact USDC — but that's a systemic collapse scenario affecting everything.
OrnyTools score: 88/100 — highest counterparty score (82), strong liquidity (90), lowest protocol risk (95).
USDe (Ethena): synthetic yield leader
Mechanism: Delta-neutral synthetic dollar. Ethena holds stETH/ETH as collateral and runs equal-size ETH short perpetual positions on major CEXes (Binance, OKX, Bybit). Net ETH exposure = 0. Yield comes from: (1) stETH staking rewards (~3.5%) + (2) ETH funding rates paid by longs to shorts (historically 5–15% APY on shorts).
Depeg risk: Higher than USDC. Two scenarios that break the peg: sustained negative funding rates (shorts pay longs, eroding backing) and a major CEX collapse during settlement. In March 2024, USDe briefly hit $0.97 during a sharp ETH sell-off — it recovered within hours as funding normalized.
Why people hold it: sUSDe (staked USDe) earns 8–20% APY during bull markets — no other top-20 stablecoin comes close. The yield is real (funding rates, staking) rather than inflated token emissions.
OrnyTools score: 76/100 — high protocol score (80, extensively audited), but lower counterparty (55) and depeg base (65) due to CEX and funding rate exposure.
USD0 (Usual): T-bill yield with DeFi composability
Mechanism: USD0 is backed by US T-bills, similar to USDC — but issued by Usual Protocol, not Circle. USD0++ (staked USD0) earns the T-bill yield as a liquid, yield-bearing token. wUSD0 is the ERC-4626 wrapper allowing DeFi composability while earning.
The key difference from USDC: The yield goes to USD0++ holders, not to the issuer. Usual distributes ~90% of T-bill revenue to stakers, keeping ~10% as protocol fee. Circle keeps all the T-bill yield internally.
Depeg risk: Moderate. The mechanism is simpler than USDe (T-bill backing, not synthetic), but Usual Protocol is newer with less track record. The USUAL token incentives also introduce risk — if USUAL collapses, APY inflated by incentives disappears, potentially triggering outflows.
Why people hold it: 4–6% base T-bill yield + USUAL token incentives. Strong Morpho ecosystem (MEV Capital and Steakhouse vaults use USD0/wUSD0). For USDC-equivalent safety with higher yield, USD0 is the current leading candidate.
OrnyTools score: 74/100 — good mechanism (RWA-backed), but newer protocol (protocol score 60) and moderate counterparty (55) vs USDC's institutional structure.
How to allocate across the three
A practical DeFi portfolio typically holds all three for different purposes: USDC for collateral where LTV matters (deepest Aave/Morpho markets), USDe/sUSDe in yield-optimized vaults, USD0/wUSD0 in Morpho Blue vaults for T-bill yield with DeFi composability. The OrnyTools Stablecoin Analyzer provides live prices, risk scores, and DeFiLlama yields for all three.
Analyze all stablecoins →