Trading strategies·other

Rebate capture

≈commoditised
Reviewed 4 June 2026. As of 2026: widely known and implemented; the edge is in execution, not the idea.

Trade to collect the maker rebate rather than to predict the price. Only works while the rebate exceeds the adverse-selection cost: a fee-schedule game, alive on inverted venues and specific tiers.

The idea

Rebate capture annotated diagramfigure
Trade to collect the maker rebate rather than to predict the price. Only works while the rebate exceeds the adverse-selection cost: a fee-schedule game, alive on inverted venues and specific tiers.

Reference figure. This concept is explained in prose and diagram; the interactive widgets live on the flagship pages it links to under Where this fits.

What is rebate capture?

Rebate capture is a strategy whose primary P&L source is the liquidity rebate an exchange pays for adding (posting) liquidity, not the price change on the position. The trader posts passive orders to collect the maker rebate, aiming to do so at scale and to exit without paying it back as a taker fee. The price move is incidental; the rebate is the trade.

Intuition first. On a maker-taker venue, the exchange pays you a rebate for resting an order that someone else trades against, and charges a fee to whoever crosses the spread. If you can buy and sell while being the maker on both legs, you collect two rebates and the price barely needs to move. Your profit is the venue's payment for providing liquidity.

Why is this its own strategy and not just market making? A market maker earns the spread and the rebate. Rebate capture is the limiting case where the rebate is the dominant or sole intended profit, and spread capture is secondary or absent. In its purest form the trader is nearly indifferent to direction over the holding period: they just want to be the maker, fill, and flatten without ever paying the taker fee.

The maker-taker economics that create it

Maker-taker pricing is the venue's incentive scheme: pay makers a rebate, charge takers a fee, keep the difference. Rebate capture exists because of this asymmetry. The rebate is a real, scheduled payment, and a strategy that reliably ends up on the maker side of trades harvests it.

The mechanics, precisely. A venue posts a fee schedule: rebate RR per share for adding, fee FF per share for removing, with the venue's take of about F−RF - R. Under US Reg NMS Rule 610 the access-fee cap historically bounded taker fees at $0.0030/share, and the SEC's 2024 access-fee reforms (phasing in through 2025–2026) lower that cap and tie fees to tick bands, directly shrinking the rebate pool. Inverted (taker-maker) venues flip the sign, paying takers and charging makers to attract aggressive flow, which creates the cross-venue rebate games below.

Your fee-side P&L is what the venues paid you minus what you paid them. A pure rebate-capture trade wants this strictly positive before any price P&L.
Πfee=∑Rmaker−∑Ftaker−∑feesexch/clearing\Pi_{\text{fee}} = \sum R_{\text{maker}} - \sum F_{\text{taker}} - \sum \text{fees}_{\text{exch/clearing}}

Since RR is a fraction of a cent, this must be earned on enormous volume with a very high maker-fill ratio. The economics are the thin-margin case in its starkest form (the economics of an HFT desk): a large fixed-cost base, a sub-cent prize per trade, and survival only through volume. The full treatment of aggressive-versus-passive trading and the fee schedules lives on maker-taker.

Queue and rebate games

Because the rebate is fixed and small, rebate capture becomes a queue game (be early enough in the FIFO queue to actually fill and earn the rebate) and a venue game (route to whichever venue's fee/rebate combination pays best for this order). The "alpha" is logistics (fee-aware routing and queue priority), not price prediction.

The queue game. A rebate is only earned if your passive order fills, and fill probability is dominated by queue position. So rebate capture is intensely sensitive to being early in the queue, which is why it co-evolved with low latency. Joining a deep queue for a rebate exposes you to the same adverse-selection tax as spread scalping: the fills you get are the ones you would rather not have.

The venue game (rebate arbitrage / routing). With maker-taker and inverted venues side by side, the net price of the same trade differs by venue once rebates and fees are netted. Fee-aware smart order routing posts where the rebate is richest and takes where the fee is cheapest (or where takers are paid). The cheapest displayed price is not the cheapest net price (market fragmentation). One footnote for recognition, not how-to: maker-taker rebates create well-documented routing conflicts of interest (a broker routing for its own rebate rather than the client's best execution), a recurring SEC and MiFID II concern.

▸ Show the derivation optional

Per posted order, expected P&L is the rebate weighted by fill probability, less the adverse move on toxic fills, less the chance you must cross the spread to flatten and pay the taker fee.

E[P&L/order]=Pr⁡(fill)⋅R−Pr⁡(fill∣adverse)⋅E[ΔSadverse]−Pr⁡(must take to exit)⋅FE[\text{P\&L/order}] = \Pr(\text{fill}) \cdot R - \Pr(\text{fill} \mid \text{adverse}) \cdot E[\Delta S_{\text{adverse}}] - \Pr(\text{must take to exit}) \cdot F

The tension mirrors spread scalping exactly: raising fill probability raises the rebate term and the adverse term and the chance of a forced taker exit. Because RR is capped and small while FF and the adverse move are not, the two negative terms dominate quickly.

R≤Rcap,∂ E[P&L]∂ aggressiveness  <  0  once the adverse and taker terms biteR \le R_{\text{cap}}, \qquad \frac{\partial\,E[\text{P\&L}]}{\partial\,\text{aggressiveness}} \;\lt\; 0 \;\text{once the adverse and taker terms bite}

Why it is razor-thin and venue-dependent

The prize is a fraction of a cent per share, set entirely by a venue's fee schedule that the venue and the regulator can change at will. After adverse selection and the times you must pay a taker fee to exit, the net is tiny and fragile. It works only for very-high-volume, very-low-latency, front-of-queue operators, and the 2026 regulatory direction is shrinking it.

Razor-thin by construction: RR is bounded by the access-fee cap, and the SEC's phased reductions lower it. A strategy whose entire edge is a capped, shrinking, sub-cent payment has no margin for error. One adverse fill, or one forced taker exit, erases dozens of rebates. Entirely venue-dependent: change one venue's fee schedule and the economics flip. This is not an edge you own; it is a subsidy you qualify for, and the venue controls the tap. The same posture is profitable on Venue X and loss-making on Venue Y purely because of the fee print.

Stated plainly for 2026: a utility for a handful of high-volume firms with the latency, queue priority, and routing infrastructure to harvest sub-cent rebates at scale; a trap for a small operator who sees "the exchange pays me to trade" and ignores the adverse-selection and exit costs. It is structurally under pressure from access-fee reform and the long-running debate over whether maker-taker should exist at all. What AI changes is little directly: the rebate is a fixed payment, not a prediction. ML helps at the edges (predicting toxicity to avoid adverse fills, optimising fee-aware routing) but cannot raise a capped rebate (what AI changes for HFT).

Worked example

A rebate-capture round trip on a US maker-taker venue, illustrative and dated to 2026 (synthetic; figures are order-of-magnitude under the post-reform access-fee regime). The venue posts a maker rebate R≈$0.0020R \approx \$0.0020/share and a taker fee F≈$0.0025F \approx \$0.0025/share.

You post a buy passively and earn the rebate when it fills: +$0.0020/share. You then post a sell passively for another rebate: ideally +$0.0020/share, for a gross of +$0.0040/share on the round trip with the price ideally flat. But suppose 40% of the time your inventory is unbalanced and you must take to flatten, paying F=$0.0025F = \$0.0025/share; and on the passive fills, adverse selection costs about $0.0015/share as the price drifts against the fills you got.

Net per round-trip share: two rebates, less the forced taker exits, less adverse selection, less clearing. Positive, but barely, and only because we assumed a high passive-fill ratio and modest adverse selection.
0.0040−0.40×0.0025−0.0015−0.0003≈+0.0012 $/share0.0040 - 0.40 \times 0.0025 - 0.0015 - 0.0003 \approx +0.0012\ \text{\$/share}

Now break it. Raise the forced-taker share to 60%, or the adverse cost to $0.0025/share (a slightly more toxic regime), and the net goes negative. The whole strategy lives inside a few tenths of a cent, which is exactly why it requires front-of-queue priority, near-zero latency, and immense volume, and why a small operator who only sees the "+$0.0020 rebate" loses money.

The venue-dependence punchline: route the identical behaviour to a venue with R=$0.0010R = \$0.0010 (or to an inverted venue that charges makers), and the same trades are net-negative. Nothing about your skill changed; the fee print did. All figures are synthetic; verify any live rebate or fee against the venue's current schedule and the prevailing access-fee cap before relying on it.

Where this fits

↑ Up · building block of Trading strategies ↔ Across · composes with Maker vs taker ↔ Across · composes with Spread scalping ↔ Across · composes with Transparent costs → Apply · makes money in Equities & futures ⤓ Build / Buy · tool needed Datasets & tools