Wash-Trading Resistance in Volume Campaigns

A prize pool is a bounty on your scoring rules. If volume is the score, someone will manufacture volume, and the only interesting question is what it costs them relative to what they can win.

"We monitor for it" is not an answer. These are the mechanisms that change that ratio, and what each one does not do.

The economics first

This does most of the work, and it is structural rather than a rule you have to switch on. A prize pool is fixed. Under a pro-rata split, each trader gets a share of that constant pot proportional to their counted volume. A wallet trading against itself is therefore paying real fees, spread and slippage on every round trip to win back a proportional slice of a pot it is also diluting. The larger the honest field, the smaller that slice.

This does not eliminate wash trading. It prices it. Where fees are near zero and the field is thin, self-trading can still be profitable, which is why the rules below exist on top.

Fixed-rank tiers behave worse: first place pays the same whether the honest field is four wallets or four hundred, so a farmer's return does not decay with turnout. If abuse resistance is your priority, pro-rata is the stronger default.

Per-day counted-volume caps

The most direct defense. A campaign can carry a daily volume cap: for each trader, on each campaign day, only volume up to the cap counts toward score. The arithmetic is deliberately simple — counted volume for the day is min(actual volume, cap), and points come from the counted figure, not the raw one. A derived daily points ceiling backs it up, so the cap holds regardless of which multiplier path a campaign uses.

What this changes:

• It bounds the return on any single day's manufactured volume. Past the cap, every additional round trip costs fees and earns nothing.
• It flattens the whole field, not just the abusers. A large honest trader hits the cap too. That is a real tradeoff — a cap moves the campaign from rewarding size toward rewarding consistency.
• It makes "trade every day" the optimal strategy. Which is usually what a venue wanted anyway.

Days are anchored to the campaign's own start time rather than midnight UTC: a campaign beginning at 14:00 rolls its day at 14:00, on every venue. If the cap, the leaderboard and the settlement job disagree about where a day begins, they will disagree about who won.

Minimum qualifying volume

The complementary gate. A campaign can require a minimum counted volume before a participant is eligible for main-pool rewards at all: below the bar, a wallet appears on the leaderboard but is excluded from the split.

Its purpose is not to stop large-scale wash trading — a determined farmer clears any threshold — but to make Sybil splitting unprofitable. Without it, the optimal attack on a pro-rata pool is to spread the same volume across a hundred fresh wallets, since a hundred small shares equal one large share and the extra wallets cost nothing. A minimum turns wallet creation from free into linear in fees. It stacks with a campaign's other eligibility rules — a points floor, a minimum trade count, a verified social connection — each raising that cost in a different currency.

Two caveats. The minimum is judged on counted volume, so with a cap you have defined a band, not a floor. And it is an operator-configured knob rather than a step in the self-serve wizard; permissionless competitions get the window and the pro-rata economics by default.

Flat-start requirements

Specific to PnL-scored events, and closes an attack that has nothing to do with volume. A trader who enters while already holding a large position is not competing on what they did during the campaign — they are banking a move that started before it, and in a multi-entry format they could do it repeatedly with hindsight.

The defense: a line does not start accruing until the account is flat on the markets the event scores, at or after the moment of entry. The scoring engine enforces it after the fact, and the entry path checks it before taking the money — reading live positions from the venue rather than cached rollups, because a rollup refreshed every five minutes is exactly the window in which someone closes a position before clicking.

The detail worth copying: when the venue cannot be reached, the check returns unknown, not flat. Failing open on an anti-abuse check is how a defense becomes decorative.

Windows and per-day eligibility

Volume before the start does not count. The sync refuses to read fills before the campaign's start time, even if the campaign row is already active. Without that guard, a campaign retroactively rewards flow that was happening anyway — not abuse exactly, but paying for volume you already had.

Random draws are scoped to the day. A given day's draw includes only entrants who traded within that day's own 24-hour window, above a minimum daily volume where one is set. The alternative — anyone who ever qualified stays in every draw — turns one qualifying trade into a permanent free lottery ticket. Draws are solvency-checked, so a reserve cannot promise more than it holds.

Raw volume next to counted volume

A transparency mechanism rather than a barrier, and underrated. Leaderboards display two numbers: counted volume, what points and rank are computed from after caps, and raw traded volume, everything the wallet actually did. The API exposes both — totalVolumeUsd for the capped figure, uncappedTotalVolumeUsd for the raw one.

Showing both does three things at once:

• It makes the rules legible. A trader who hit the cap can see they hit it, rather than concluding the leaderboard is broken.
• It makes farming visible. A wallet whose raw volume is many multiples of its counted volume is a public signal, on a public page.
• It stops a dispute before it starts. "My volume wasn't counted" is answerable with a number rather than an assurance.

Collapsing the two into one column removes all three properties.

Detection, after the fact

Rules bound the payoff; they do not identify who tried. That needs data the scoring pipeline throws away — a sync reduces each fill to a timestamp and a notional, which cannot answer "did these two accounts trade with each other?". Where a venue exposes full fill records, four signatures generalise:

• Matched pairs. Two participants sharing a trade identifier, or one account's counter-order being the other's order. Direct evidence, not a heuristic.
• Churn ratio. Gross volume over net inventory. A directional trader runs single digits; a volume farmer never builds inventory, so the ratio explodes.
• Coordination timing. Opposite-side fills from different accounts within a couple of seconds, repeatedly.
• Share of market volume. Credited campaign volume for a day against everything that traded on that market. Each trade credits at most twice — once per side — so a ratio above 1 forces that excess to be participant-versus-participant, and above 2 is arithmetically impossible, meaning the pipeline is over-crediting. This one needs only public candles and yields a lower bound rather than a suspicion.

On the ingestion side, syncs must deduplicate: paginated endpoints re-serve rows at window boundaries, and some venues split one economic event across two record types. Double-counting a fill is indistinguishable from manufactured volume, except you did it to yourself. Some integrations also exclude self-trades at the source — crediting both maker and taker, but dropping trades where both sides are the same participant.

What none of this achieves

Be blunt about the ceiling:

• You cannot make wash trading impossible. Every defense here raises its cost. On a low-fee venue against a thin field, some manufactured volume is rational and will happen.
• Wallets are cheap. On-chain attribution identifies a wallet, not a person. Sybil resistance comes from per-wallet cost, not identity.
• Caps have a real price. They dampen the headline number and annoy large honest traders. That should be a decision, not a default nobody examined.
• Rules must be frozen once published. Tightening one mid-campaign damages credibility more than the abuse it was meant to stop.

Checklist before you fund anything

• Pro-rata rather than fixed tiers, if abuse resistance outranks spectacle.
• A per-day counted-volume cap, sized so a normal large trader occasionally hits it.
• A minimum qualifying volume that makes one-wallet-per-hundred-dollars uneconomic.
• Day boundaries anchored consistently, and identical scoring in the leaderboard, the projection and settlement.
• Raw and counted volume both displayed.
• Per-day, per-trader rows retained, so a dispute has an answer.
• Published rules, frozen at launch.

Voltrade for exchanges is where this machinery lives, trading campaign infrastructure for perp DEXs is the full build-versus-buy inventory, and user acquisition for perp DEXs covers the strategy these rules sit inside. Counted and raw volume sit side by side on any live competition.