How to swap large amounts of crypto without hitting provider limits

A single $250k BTC to USDT swap will fail on roughly half of the providers in any non-custodial aggregator — not because of compliance, but because their per-order maxSwapUsd for that pair sits at $100k–$150k while others stretch to $500k–$1M. The same order, split into three $80–90k legs across providers with the highest caps, routes cleanly and avoids the internal review tier most providers set between $20k and $50k. The job at size isn’t “find a provider with no limit”. It’s to read the cap table, split sensibly, and let routing do the rest. Below: worked examples, per-coin numbers, and the rule for when an OTC desk becomes the right call.

What counts as “large” in a non-custodial swap

The line is lower than most people assume.

$10k baseline. This is where routing matters. Below it, spread differences between providers are rounding errors. Above it, the difference between the tightest and widest provider spread on the same coin starts to be worth real money.

$50k strict threshold. Where the cap table genuinely starts to bind. Some providers refuse the order, others quote a worse spread on a larger ticket, and one-off reviews start to appear. Verification thresholds vary by provider and amount.

$250k whale zone. At this size you’re explicitly choosing between an aggregator that splits the order and an OTC desk that prices it as a single block.

“Large” isn’t only about dollars — it’s coin × network. $50k of BTC on mainnet is routine. $50k of XMR is close to the per-leg ceiling for half the set, and $50k of a lower-cap altcoin can blow through caps and still hit thin liquidity once routed.

Why one big swap quietly hurts you

Three things go wrong when you push the whole amount through a single order.

Slippage on illiquid pairs. Aggregators ultimately route into someone’s liquidity — an exchange order book, an OTC inventory, or a DEX pool. A $100k market sweep on an illiquid pair walks up the book; you pay the average fill, not the top quote. Headline rate degrades by 0.3–2% the moment your order is the dominant size in the book.

Per-provider maxSwapUsd ceilings. The cap table that matters at size, by aggregator data from May 2026:

For BTC and ETH the constraint is rarely the cap — it’s the spread between providers and the size-impact on the destination book. For XMR the cap is the constraint: a $250k XMR leg simply will not route as a single order on most providers. The aggregator reads this matrix in real time and either routes to the highest-cap provider or splits.

Internal review tiers. Verification thresholds vary by provider and amount. Several providers apply enhanced compliance checks once a single order crosses an internal tier — frequently between $20k and $50k. A single $80k swap is more likely to be paused for review than three $25k swaps routed across different providers within the same hour.

A $250k swap doesn’t fail because of compliance. It fails because the provider’s maxSwapUsd for that coin is $150k, and nobody in the aggregator UI told you before you clicked Continue.

The splitting strategy

The default heuristic is simple. Keep each leg under the smaller of:

• 50% of the receiving provider’s maxSwapUsd for that coin
• $50k
• 1% of the destination market’s daily volume on that pair

Whichever is tightest wins. The first protects you from edge-of-cap orders that get awkward to route. The second keeps you below the review tier on most providers. The third stops you from being the dominant size in the destination book.

Worked example — $250k BTC to USDT-TRC20. BTC caps allow a single $250k order on at least three providers, and TRC20 caps allow $250k–$500k per leg, so the order can go in one shot. But destination liquidity for USDT-TRC20 at $250k can move the rate by 0.3–0.6% on slower providers. Splitting into 3 × $80–85k orders across xgram, stealthex, and simpleswap (the three with highest caps and best spreads at this size by aggregator data) typically lands 0.2–0.4% better fill and routes around any single-provider review tier. On $250k that’s $500–$1,000 of saved cost.

Worked example — $120k something to XMR. This is the one where the cap matters. XMR caps run $100k–$200k across the full set. One $120k order routes only to the two highest-cap providers, which kills price competition. Two $60k legs route to every provider in the set. Net result: better headline rate and less reliance on a single hop continuing to work for the whole order.

Time-spacing. Splitting across providers is the big lever; splitting across time matters less. A reasonable spacing is 5–30 minutes per leg. Avoid splitting across known volatility events (CPI prints, FOMC, ETF approval headlines). Either run before, run after, or use fixed rate for the legs that straddle.

Don’t over-split. Each leg has fixed costs — a network fee on the source side, a flat portion of the provider’s spread. Ten $3k legs cost more than one $30k order. The sweet spot at this size band is usually 2–5 legs.

How an aggregator helps

The widget on the homepage does the cap-table work for you. It pulls live quotes from every supported provider, knows each one’s maxSwapUsd per coin, and only surfaces providers whose ceiling clears your amount. Providers that would silently reject the order are filtered out before you see them — for the mechanics, see how SwapZilla aggregates quotes.

On the floating-rate path, the aggregator picks the best ToAmount from providers that can take your size. That’s why for large amounts floating is often the cleaner choice. Fixed-rate quotes at $50k+ carry a wider buffer, and at $100k+ several providers won’t quote fixed at all. Floating wins more often at size than at retail — except during invoice payments or known volatility events. Longer version in Fixed vs floating rate on a crypto swap.

The aggregator also surfaces real-time rateSpread per provider. Spreads on the same coin range from 0.85% to 1.45% across the integrated providers by aggregator data — at $100k that’s a $600 difference per leg between the tightest and widest spread.

The verification reality at higher tiers

Verification thresholds vary by provider and amount, so this is band-by-band.

Under ~$5k. Standard non-custodial flow. No account on SwapZilla’s side, no identity collected. The provider receives your deposit and payout addresses and nothing else.

$5k–$50k. Still standard for almost every supported provider. Some apply automated risk scoring (deposit-address taint analysis, exchange-of-origin checks) — invisible unless the address is flagged.

$50k–$250k. The band where reviews start to appear. Several providers may pause an order for an internal compliance check, especially if the deposit comes from a flagged source, the destination jurisdiction is restricted, or the payout is a privacy coin. This is not the same as opening an exchange account — it’s a one-off check on the specific order. Mitigations: split into legs across different providers, use a clean deposit address, and set a refund address.

$250k+. Every leg may get attention. This is where you weigh aggregator-with-splitting against an OTC desk.

At $50k–$250k a properly split aggregator route avoids the worst of the friction without forcing you into an exchange account. Verification thresholds vary by provider and amount — splitting across providers spreads that risk too.

What this guide won’t claim: that any non-custodial flow is “free of compliance”. Providers run real compliance programs. The claim is more precise: in this band, a well-split route is materially less friction than a custodial alternative would be.

Refund addresses become load-bearing at size

Refund addresses are optional in most provider UIs. That’s a UX choice that makes sense at retail and stops making sense at size.

At $200, the worst case for a missed refund address is a slow support ticket. At $200k, the worst case is a slow support ticket for $200k of stuck funds. When a leg expires (TIME_EXPIRED), gets paused for review and rejected, or hits a network reorg, the provider returns funds to the refund address automatically if you set one, and to support-ticket purgatory if you didn’t.

For every leg, set the refund address to a wallet you control on the source network. Not the same wallet as your deposit address, but on the same network so the return is a normal transaction. Belt-and-braces rule for $50k+ legs: refund address set, screenshot the order ID, don’t close the tab until the leg finishes.

If your large swap involves USDT, network choice changes the fee math at size — see USDT TRC-20 vs ERC-20 vs Solana before picking the destination network. The general FAQ covers what happens if a refund actually fires.

When an OTC desk is the right answer

Above roughly $250k–$500k, the aggregator-with-splitting approach has a ceiling. Splits get awkward, per-leg spreads add up. This is where OTC desks become competitive.

OTC economics. For $100k–$5M of spot BTC and ETH on liquid pairs, OTC desks typically quote 5–25 basis points (0.05%–0.25%) over a reference mid-price, by 2026 industry roundups. The desk holds inventory and absorbs the market-impact cost you’d otherwise pay walking up a public book.

Aggregator total cost at the same size. Typical 10–40 bps end-to-end on liquid pairs, more on illiquid (XMR, smaller altcoins).

The honest tradeoff:

• OTC desk — better pricing at $500k+ on BTC, ETH, and major stablecoins. But: requires an account, source-of-funds documentation, and a call or chat per block.
• Aggregator with splitting — better pricing at under $250k on BTC and ETH, materially better pricing at any size for XMR and privacy-routed swaps (most OTC desks won’t quote XMR at all), no account required. The widget does the splitting for you.

Stay in aggregator land for under $250k on any coin, and for any XMR-leg swap at any size. Consider OTC for $500k+ on BTC, ETH, or USDT major. Between $250k and $500k it’s a judgment call — get a quote from both.

Common mistakes at size

The mistakes that actually cost money on large swaps.

• Sending the whole amount as one order without checking the cap table. The order quietly routes to the one provider that can take it — possibly not the one with the best rate. Spread cost at $100k: $200 to $1,000 vs. an optimally routed split.
• Forgetting the refund address on a large leg. Worst-case impact scales linearly with leg size. The highest expected value of any field in the form at $50k+.
• Splitting too thin. Ten $5k legs of a $50k swap cost more than two $25k legs — per-leg network fee dominates.
• Picking fixed-rate for a $100k+ swap during a quiet market. The buffer at this size can run 1–2% — $1k–$2k of insurance against a 30-minute move that on a normal day is well under that. See Fixed vs floating rate on a crypto swap.
• Splitting across volatility events. A 90-minute split across the FOMC door means leg 3 prices materially worse than leg 1.
• Using a fresh, untested destination wallet for the first big leg. Send a $50 test deposit first. Ten minutes of waiting is cheap insurance against a typo in a long XMR or Solana address.
• Trusting old pricing screenshots. Live rate at click ≠ rate at deposit confirmation for floating. For fixed, the lock starts at click. Either way, 20-minute-old screenshots aren’t load-bearing.
• Treating “no account” as “no checks ever”. It’s not. Plan for one in five large legs to need an extra step at $50k+ and you’ll be pleasantly surprised more than half the time.

Quick decision checklist for large swaps

A pocket-sized rule for the next time you’re staring at a large balance and a swap widget:

• Under $10k — single swap, normal flow.
• $10k–$50k — single swap, floating rate, refund address set, FAQ tab open in case anything stalls.
• $50k–$250k — check the cap table on the homepage for your coin pair, split into 2–4 legs across providers if the headline rates differ by more than 0.3%, floating unless it’s an invoice or event day.
• $250k–$500k — lean aggregator-with-splitting for XMR or any privacy-routed swap. Get an OTC quote for BTC, ETH, or USDT major if you have an OTC relationship. Compare and choose.
• $500k+ on BTC, ETH, or major stablecoin — OTC is usually the cleaner answer. The aggregator still works for XMR-routed legs of a larger workflow.
• Always — refund address per leg, source-network test send, no splitting across known volatility events.