Tick Size Arbitrage Across Cryptocurrency Exchanges

Abstract. We describe a class of microstructure arbitrage strategies that exploit heterogeneous tick sizes across cryptocurrency exchanges trading the same perpetual future or spot instrument. When one exchange enforces a coarse minimum price increment (tick) while another permits finer granularity, a systematic edge arises: a trader can post limit sell orders on the granular venue at sub-tick prices above the coarse venue's best ask—prices that simply cannot exist on the coarse venue—and hedge by market-buying on the coarse venue at its lower ask. The difference is pure structural profit. We formalise the setup, derive profitability conditions, present a worked example using APEUSDT across five exchanges, and discuss extensions including queue priority optimisation, funding-rate overlays, and dynamic quoting.

Introduction

Cryptocurrency perpetual futures and spot pairs are listed simultaneously on many centralised exchanges (CEXs) and decentralised venues (DEXs). Although the same underlying asset is traded, market microstructure parameters—in particular the tick size (minimum price increment) and the lot size (minimum order quantity step)—vary widely across venues.

A tick size mismatch creates a structural inefficiency. On a coarse venue the price grid jumps in large steps, leaving sub-tick price levels that a granular venue can address. A market maker who recognises this can:

1. Post passive limit sell (ask) orders on the granular exchange at sub-tick prices above the coarse exchange's best ask—prices that are unreachable on the coarse venue but still attractive to buyers on the granular venue (cheaper than the next coarse tick up).
2. When filled, immediately hedge by sending a market buy on the coarse exchange at its (lower) best ask.
3. Pocket the difference: we sold on the granular venue for more than we bought on the coarse venue, minus fees, slippage, and any funding differentials.

The mirror trade is equally valid: post a limit buy on the granular venue at a sub-tick price below the coarse exchange's best bid, and upon fill, market sell on the coarse venue at its (higher) best bid.

This note formalises the strategy, provides concrete profitability conditions, walks through a live example, and suggests several extensions.

Setup and Notation

Consider an instrument (e.g.\ APEUSDT perp) listed on two exchanges $A$ (coarse) and $B$ (granular).

Definition (Tick size). The tick size $\delta_X$ of exchange $X$ is the minimum permissible price increment. Any resting limit order price must satisfy $p \in \{k \delta_X : k \in \mathbbZ^+\}$.

Let:

• $\delta_A \gg \delta_B$, i.e.\ exchange $A$ has the coarser tick.
• $a_A, b_A$ denote the best ask and best bid on $A$.
• $a_B, b_B$ denote the best ask and best bid on $B$.
• $f_A^t$, $f_A^m$ denote the taker and maker fee rates on $A$.
• $f_B^t$, $f_B^m$ denote the taker and maker fee rates on $B$.
• $q$ denote the trade size in base units.

Because $\delta_A$ is large, the spread on $A$ is at least $\delta_A$:

Core Strategy

Sell-side arbitrage (Granular Ask, Coarse Buy)

The key observation is that on the coarse exchange $A$, the price grid jumps from $a_A$ to $a_A + \delta_A$ with no representable levels in between. On the granular exchange $B$, we can post limit orders at every sub-tick increment in that gap. Participants on $B$ who wish to buy and cannot (or do not) access $A$ will lift our offers at these sub-tick prices.

The trader posts a limit sell on exchange $B$ at price $p^{\text{sell}}_B$ satisfying:

When this limit sell is filled on $B$, the trader simultaneously sends a market buy on exchange $A$ at $a_A$.

The per-unit profit is:

Proposition (Profitability condition — sell-side). The sell-side arb is profitable if and only if

$$p^{\text{sell}}_B > \frac{a_A (1 + f_A^t)}1 - f_B^m.$$

Since $p^{\text{sell}}_B > a_A$ by construction, the edge before fees is $p^{\text{sell}}_B - a_A > 0$; the trade is profitable whenever this edge exceeds the combined fee drag.

Buy-side arbitrage (Granular Bid, Coarse Ask)

Symmetrically, the trader posts a limit buy on $B$ at a price below the coarse exchange's best bid:

Proposition (Profitability condition — buy-side). The buy-side arb is profitable if and only if

$$p^{\text{buy}}_B < \frac{b_A (1 - f_A^t)}1 + f_B^m.$$

Optimal quoting price

The trader faces a trade-off: quoting closer to $a_A$ (more aggressive) increases fill probability but shrinks the per-trade edge, while quoting closer to $a_A + \delta_A$ (less aggressive) widens the edge but reduces fills. A simple heuristic is to quote at the midpoint of the exploitable range, adjusted for fees:

More sophisticated approaches model the fill probability as a function of queue position and price aggressiveness and maximise expected profit per unit time.

Worked Example: APEUSDT

We illustrate with a snapshot of APEUSDT perpetual futures across five exchanges (see Figure~\ref{fig:orderbook}).

Observation. Binance and Hyperliquid use a tick of 0.0001, while OKX, Gate.io, and Bybit use 0.00001—a $10\times$ finer grid. On Binance, the price grid around the current level is $\{\ldots, 0.00904, 0.00905, 0.00906, \ldots\}$. The nine sub-tick levels $0.009051$ through $0.009059$ simply do not exist on Binance, but they are perfectly valid on Bybit, OKX, and Gate.io.

Trade. Suppose the Binance best ask is $a_A = 0.00905$ (the next tick up is $0.00906$). On Bybit (granular), we post limit sells at every sub-tick level from $0.009051$ up to $0.009059$. These prices are above the Binance ask: we are offering to sell on Bybit for more than we will pay to buy on Binance.

Buyers on Bybit who cannot or do not access Binance see our asks at $0.009051$–$0.009059$. These are still cheaper than $0.00906$ (the next Binance-compatible level), so they are attractive. When a buyer lifts our offer at, say, $p_B^{\text{sell}} = 0.009055$:

• We sell on Bybit: receive $0.009055 \times (1 - 0.0001) = 0.0090541$ per unit (maker fee).
• We buy on Binance at the ask: pay $0.00905 \times (1 + 0.0002) = 0.0090518$ per unit (taker fee).
• Net P&L: $0.0090541 - 0.0090518 = +0.0000023$ per unit $\approx +\mathbf{0.25$ bps}.

The edge is small per trade but strictly positive. Using maker rebates (some granular venues offer $f_B^m < 0$), quoting at higher sub-ticks, or trading larger size all amplify the total profit.

Scaling the edge. The further above $a_A$ we quote, the larger the per-unit profit but the lower the fill rate. Quoting at $0.009059$ (the highest sub-tick before the next coarse level) yields the maximum per-unit edge:

Wider-spread scenario. When Binance widens to a 2-tick spread (bid 0.00904, ask 0.00906), the buy-side mirror trade also opens: we post limit buys on Bybit at $0.009041$–$0.009049$ (below the Binance bid of $0.00904$? No—below $b_A = 0.00904$ would require quoting at $0.009031$–$0.009039$, which is the buy-side exploitable range). The combined two-sided book on Bybit captures flow on both sides of the Binance spread.

When does the strategy NOT work?

• When fees are so high that the sub-tick edge cannot cover $f_B^m + f_A^t$.
• When the granular venue lacks enough organic flow (nobody lifts our asks).
• When latency to the coarse venue is too high, causing the hedge price to slip.

Extensions and Variations

Multi-leg sub-tick market making

Rather than a single arb leg, the trader can run a full sub-tick market-making book on the granular exchange, quoting both bid and ask at prices inside the coarse exchange's spread. The coarse exchange serves as the hedging venue. This is the continuous-time analogue of the one-shot arb and generates a stream of small profits from the bid–ask spread on the granular venue, hedged by the coarse venue's liquidity.

Queue priority and passive flow capture

On the granular exchange, sub-tick price levels have zero or very thin queue depth because few participants quote there. This provides:

• Nearly guaranteed queue priority (you are first at that price level).
• Higher fill rates per unit of adverse selection risk compared to quoting at a crowded integer-tick level.

Funding rate overlay

Perpetual futures accrue funding payments every 1–8 hours depending on the exchange. The funding rate $r_X$ on exchange $X$ is the periodic payment from longs to shorts (when $r_X > 0$) or shorts to longs (when $r_X < 0$). Crucially, funding rates differ across exchanges because each venue computes them from its own order book and mark price.

The tick-size arb naturally holds a cross-venue position: short on the granular venue $B$ and long on the coarse venue $A$ (sell-side), or vice versa (buy-side). This position accrues a net funding flow every period:

\paragraph{Directional bias from funding.} The trader should bias quoting toward the side that aligns with a favourable funding differential:

• If $r_A > r_B$ (longs pay more on $A$ than on $B$): prefer the buy-side arb, which leaves us short on $A$ (collecting high funding) and long on $B$ (paying low funding).
• If $r_A < r_B$: prefer the sell-side arb, which leaves us long on $A$ (paying low funding) and short on $B$ (collecting high funding).

\paragraph{Quantifying the overlay.} Let $T$ be the funding interval (e.g.\ 8 hours) and $N$ the expected number of arb round-trips per interval. The total PnL per funding period is:

\paragraph{Example from the APEUSDT snapshot.} From the exchange headers in the snapshot:

• Binance funding: $-0.0146$ (shorts pay longs).
• OKX funding: $-0.0017$.
• Bybit funding: $0.01$ (longs pay shorts).
• Hyperliquid funding: $0.0013$.

Cross-venue latency considerations

The strategy's profitability depends on hedge execution speed. If the granular-venue fill triggers a hedge on the coarse venue with latency $\tau$, the price may move by $\sigma\sqrt{\tau}$ (where $\sigma$ is per-second volatility). The expected slippage cost is:

Dynamic quoting and regime detection

A more refined system detects regime shifts in the coarse venue's spread. When the spread widens (e.g.\ during volatility spikes or thin liquidity periods), the exploitable range grows and the strategy scales up. When the spread compresses to one tick, the strategy pauses or reduces size.

Key signals include:

• Coarse exchange spread in ticks (integer).
• Order book imbalance on both venues.
• Recent trade flow direction (net buy/sell pressure).
• Volatility regime (realised vol over the last $N$ seconds).

Lot size mismatches as an additional edge

The attached snapshot shows that lot step sizes also vary (e.g.\ step size $1$ on Binance vs.\ $0.1$ on Bybit/Hyperliquid). This creates a secondary inefficiency: on the coarse-lot venue, a trader who wants to fill 0.3 units must round to 1 and eat more risk. On the fine-lot venue, trades can be sized precisely, enabling tighter risk management and more surgical position building.

Triangular and multi-hop tick arb

When three or more exchanges have mutually different tick sizes (e.g.\ $\delta_1 = 0.0001$, $\delta_2 = 0.00005$, $\delta_3 = 0.00001$), a triangle of sub-tick arb opportunities opens. The medium-tick venue can serve as either the coarse leg versus the finest venue, or the granular leg versus the coarsest venue, depending on where the spread is widest.

Risk Management

1. Inventory risk. If the hedge leg fails to fill (e.g.\ the coarse venue's quote disappears), the trader is left with a naked directional position. A hard inventory limit and circuit breaker should be enforced.
2. Adverse selection. Fills on the granular venue may be adversely selected (filled right before a large move). The trader should monitor fill toxicity (e.g.\ post-fill mark-out at 1s, 5s, 30s) and widen quotes if toxicity rises.
3. Withdrawal/transfer risk. Sustained one-directional flow leads to inventory accumulation on one venue. Periodic rebalancing via on-chain transfers or internal transfers is needed, incurring gas/withdrawal fees and delay.
4. Exchange risk. Counterparty risk is real in crypto. Distribute capital across venues according to a risk budget.
5. Regulatory risk. Cross-venue arbitrage may face evolving regulatory scrutiny. Ensure compliance in all operating jurisdictions.

Conclusion

Tick size heterogeneity across cryptocurrency exchanges is a persistent structural feature of fragmented markets. It creates a well-defined microstructure edge for traders who can:

1. Monitor order books across venues in real time.
2. Post sub-tick limit orders on granular venues.
3. Hedge rapidly on coarse venues with market orders.
4. Manage cross-venue inventory and latency.

The raw per-trade edge is small (sub-basis-point to a few basis points), but with high-frequency execution and proper risk controls, the strategy compounds into a consistent, low-Sharpe-variance revenue stream. The edge is most pronounced for mid- and small-cap perpetual contracts (such as APEUSDT) where:

• Tick sizes are often set conservatively and not harmonised across venues.
• Spreads are wider in tick terms.
• Competition from other HFT participants is lower.

As exchanges periodically revise tick size schedules, these opportunities are transient at the individual instrument level but perpetually recurring across the broader crypto universe.