English
Deutsch
日本語
العربية
한국어
Español
Português
Indonesia
中文
The two strategies the VIP Crypto Arbitrage Bot routes between — directional one-leg arbitrage and market-neutral hedge arbitrage. Mechanics, math, capital sizing, broker requirements, risk profile, and when to use each.
One-leg arbitrage opens a single directional position when one venue’s price spikes away from a reference, then closes when the spread closes. Higher return per trade, market-direction risk during the hold.
Hedge arbitrage opens simultaneous opposite positions on two venues — long on the cheaper, short on the expensive — and closes both when prices converge. Near-zero net market exposure, but capital is committed on both sides and per-trade profit is smaller.
If you are starting: begin with hedge arbitrage. If you have $5k+ and tolerate variance: add one-leg on top.
Risk: Medium · Edge: High
One-leg arbitrage detects a transient price dislocation on a single venue against a reference (the bot’s aggregated mid-price) and opens a single directional position that profits when the dislocation reverts. Unlike hedge arbitrage, only one leg is in the market — so the position carries directional exposure during the hold.
Bot reads tick on Exchange A; compares to aggregated mid across reference venues.
Confirms spread > threshold after fees, slippage, and depth-adjusted execution price.
Long the cheaper side or short the expensive side — whichever is mispriced relative to mid.
Closes when price reverts to within tolerance of reference, or stop-out fires on adverse move.
# trade
size = 0.20 BTC
entry (Bybit ask) = 43,184
exit (when Bybit mid ≥ 43,225) target = 43,225
fees (taker 0.055% × 2) = 9.51 USDT
expected slippage on entry/exit = 3.20 USDT
# P&L if convergence completes
gross = (43,225 − 43,184) × 0.20 = +8.20 USDT
net = 8.20 − 9.51 − 3.20 = −4.51 USDT # too small
# bot rejects: spread (after fees) is below threshold
# bot fires only on dislocations >= 30 USDT (typical for 0.2 BTC size)
The example illustrates why fee + slippage gating is critical. A 48-USDT raw dislocation looks attractive but evaporates after 9.51 USDT in taker fees and 3.20 USDT in expected slippage. The bot’s threshold logic rejects this trade and waits for a wider dislocation.
One-leg arbitrage works best when the dislocation is large relative to typical volatility on the leg, when the reference feed is high-quality (low-latency aggregation across at least 3 venues), and when the holding period is short (typically < 30 seconds). On low-volatility hours and stablecoin-dominant pairs the strategy degrades because dislocations rarely exceed the fee threshold.
Risk: Low · Edge: Medium
Hedge arbitrage opens simultaneous opposite positions on two venues — long on the cheaper side, short on the expensive. The position is delta-neutral; profit is the spread that closes regardless of where the market moves. Capital must be deployed on both sides, and each side incurs its own fees.
Bot scans all configured venue pairs for the same asset; flags spreads above threshold.
Confirms combined fees + slippage on both legs is less than the spread net of expected close-slippage.
Sends paired orders — long on the cheap side, short on the expensive. Tries to fill within a 200–500 ms window.
When the spread reaches close-threshold (typically 5–10% of entry spread), unwinds both legs.
# entry — paired
LONG Bybit @ 3,158.40 notional = 15,792 USDT
SHORT OKX @ 3,162.10 notional = 15,810 USDT
gross spread captured = +18.50 USDT
# waiting for convergence (typical: 5–60 seconds)
… spread narrows to +0.60 USDT
exit Bybit @ 3,160.10 exit OKX @ 3,160.70
# close P&L
Bybit leg = (3,160.10 − 3,158.40) × 5 = +8.50
OKX leg = (3,162.10 − 3,160.70) × 5 = +7.00
gross = +15.50 USDT
fees (4 taker 0.055% × 15.8k) = −13.91 USDT
funding (5 ETH × 5s on OKX perp) = −0.04 USDT
net = +1.55 USDT # thin but positive — bot fires only when expected net >= +3
# bot threshold for ETH at 5-ETH size: net >= 0.06% of notional ~ 9 USDT
# this trade gets rejected → bot waits for wider spread
Hedge arbitrage profits the convergence of the spread, not the direction of either market. Note that 4 trading fees (entry-long, entry-short, exit-long, exit-short) plus optional perp funding form the cost floor. The bot rejects this trade despite being technically positive because the +1.55 USDT net falls below the bot’s edge-margin threshold.
Hedge arbitrage works best on liquid pairs (BTC, ETH, top-15 alts) across high-volume venues, during periods of moderate-to-high cross-venue dispersion (post-news, during liquidations, during regional opens), and with stablecoin-quoted pairs to avoid FX exposure. On thin pairs or quiet sessions the spread fails to exceed the 4-fee floor and the bot remains idle.
| Dimension | One-leg | Hedge |
|---|---|---|
| Market exposure during hold | Directional (long or short) | Delta-neutral (paired) |
| Capital required for $1 of position | $1 (single side, plus margin) | $2 (both sides — long capital + short margin) |
| Number of fees per round-trip | 2 (entry + exit) | 4 (entry ×2, exit ×2) |
| Break-even spread (typical, 0.055% taker) | ~0.11% of notional | ~0.22% of notional |
| Per-trade return (typical) | 0.05–0.40% on captured side | 0.03–0.15% on combined notional |
| Variance / drawdown profile | Higher (directional risk) | Lower (market-neutral) |
| Latency sensitivity | High (single fast leg) | Medium (paired execution) |
| Funding-rate exposure | None (or one leg if perp) | One leg if perp-perp; both if both perps |
| Number of accounts required | 1 (plus reference feeds) | 2 minimum |
| Suitable starting capital | $2,000+ (variance manageable) | $1,000+ (lower variance) |
| Best for | Active traders, experienced operators | First-time arbitrageurs, scaling capital |
Three correlated pairs on a single venue (e.g., BTC/USDT, ETH/USDT, ETH/BTC) where the implied cross-rate diverges from the quoted rate. The bot can run this as a single-venue, three-leg sequence; it is technically a one-leg variant because all legs are on the same exchange and capital flows through the cycle. Best on Binance, KuCoin, Bitfinex where pair coverage is dense.
Long spot, short perpetual on the same underlying — e.g., long BTC spot, short BTC USDT perpetual. The strategy is delta-neutral and harvests the basis (perp price − spot price) plus funding. Variant of hedge arbitrage where both legs are on the same venue. Suitable for capital $5k+ that can be locked for 1–7 days at a time.
Where the perpetual funding rate is sustainably negative (or positive) on one venue, long-spot / short-perp captures the funding flow as a yield. Lower turnover than tick-arbitrage; closer to a yield trade than a true arbitrage but logically a hedge variant.
Long perp on Venue A, short perp on Venue B for the same underlying when funding rates diverge across venues. Lower capital efficiency (margin on both sides) but funding spread can be persistent during stressed markets.
Many users configure threshold logic on the bot’s default fee assumption (0.10% taker) and discover later that their actual venue tier is 0.075% or 0.18%. Always pull the live fee schedule from each connected venue before deploying.
A 0.01% per-8h funding rate is small per cycle but can dominate P&L on multi-hour holds. Calculate funding cost per expected hold time and add to the threshold equation.
If your strategy occasionally runs the long side empty on one venue (because the long was unwound but the asset hasn’t transferred back), withdrawals from venue to venue can take 10–30 minutes during congestion. Plan for capital lock-up; do not assume same-block settlement.
Some venues actively widen spreads or inject latency on accounts that consistently take liquidity at top-of-book. The bot’s slippage-monitoring layer flags this; if slippage on a venue rises > 1.8× baseline over a 24h window, switch to maker-only orders or reduce participation on that venue. See the anti-arbitrage plugins guide for the full taxonomy.
API outages on a major venue (Binance, Bybit) propagate across the wider market — spreads explode but no one can trade. The bot’s circuit-breaker logic auto-pauses on stale-quote detection, but check that exposure is flat before downtime, not during.