How to Run a Crypto PnL Challenge

A PnL challenge is a permissionless competition you can create in a few minutes: everyone pays the same USDC entry ticket, the tickets build the prize pool, everyone trades their own account on Hyperliquid or Lighter, and the leaderboard settles it. No custody, no keys, no approval queue.

This is the mechanism end to end — the economics, the knobs that matter, and the ones that quietly decide whether your challenge is any good.

Two modes, and the money in each

The first decision is whether you are playing or bankrolling.

Ticket mode. You pay an entry ticket like everyone else and compete. Your ticket is the first seed of the pot, and paying it is what publishes the challenge. Every ticket, yours included, splits the same way: 95% to the prize pool, 5% platform fee.

Sponsored mode. You fund the pool up front instead of paying a ticket, and then take a share of every ticket sold. You set that share when you create the challenge. The default split is 50% pool, 45% you, 5% platform — and you can move your own share anywhere from 0% up to 70%.

Two structural facts about that table. First, the platform fee is 5% in both modes and is not negotiable — which means raising your own share in sponsored mode only ever comes out of the players' pool, never out of the fee. Second, the ceiling of 70% exists precisely because of that: at least a quarter of every ticket has to reach the pool, or the "sponsored pool" stops being a competition and becomes a toll.

Sponsoring a challenge covers the sponsored side as a revenue product in detail. The rest of this guide applies to both.

Start from a template

Templates prefill the ticket, volume band, capacity, duration and prize split. They exist because the four numbers interact, and getting them coherent by hand takes more thought than most creators want to give a first challenge.

Everything is editable afterwards. The templates are a starting point with the relationships already sane: the ticket scales with the volume band, and the number of paying places scales with the number of seats.

Choosing the volume bracket

The bracket is the single most important design decision, and it is what makes a small-account challenge worth entering at all.

The minimum is an eligibility gate. Trade less than the floor across the challenge and you are not in the prize split, however well you scored. It exists to stop someone entering, making one lucky trade, and taking the pot without ever really competing.

The maximum is a pro-rating ceiling, not a hard stop. You are not blocked from trading above it. What happens is that each fill's realized PnL is scaled by the fraction of that fill still fitting under your remaining cap, and any unrealized PnL on open positions is pro-rated the same way. Once you are at the cap, additional trading books volume but carries no more profit into your score.

This is what "whales can't buy the win" means mechanically. A trader with fifty times the capital cannot simply run fifty times the size and outscore everyone on absolute dollars, because only the bracketed slice counts. Volume brackets explained walks through the arithmetic with worked numbers.

Set the band around the accounts you want in the room. A $10k–$25k band is a small-account contest; a $100k–$500k band is a different sport. A 1v1 duel usually wants no cap at all, because with two players there is nobody to protect.

Everyone starts flat

Scoring does not begin until you are flat on the challenge's markets at or after the start. A position you were already holding when the challenge opened contributes nothing until you close it.

Without this rule, the winning strategy would be to open a large position days early, enter a challenge, and close it inside the window — cashing pre-existing PnL as if it had been earned in the contest. The flat-start gate makes the challenge measure what happened inside it. It also means a trader who is never flat during the window scores nothing at all, which is worth saying plainly on your challenge page.

Duration, capacity, and the knobs that decide turnout

• Window. You set a start and an end. Short windows (24h) are high-variance and easy to promote. Longer windows (a week) reward consistency and give late entrants a reason to join.
• Max participants. Optional, minimum 2. It caps people, not leaderboard lines — a re-buy is an extra line for someone who already holds a seat.
• Minimum participants to start. Optional, minimum 2. If the challenge is below it at the start time, it auto-cancels and entries are refunded. This is the safety net for an ambitious challenge that does not fill.
• Late join. Closed by default: the field locks at the start. Open it, optionally with a deadline, if you would rather keep selling tickets while the challenge runs. Note the trade-off honestly — a late entrant has less clock but a fresh volume bracket.
• Visibility. Public challenges take anyone. Private challenges take join requests that you approve one by one, then each approved entrant pays.
• Re-buys. Off by default. Turned on, a player who is out of contention can buy another line: same ticket price, same split, scored from zero over its own window. Buying one freezes the line they were on. You can set a deadline and a per-player cap.

Re-buys have real guardrails, and they are worth understanding before you enable them. A player must be flat to buy one, their current line must have actually started scoring and traded some volume, and any ticket already reserved and unpaid must be settled first. Those rules stop a player stacking tickets on lines that could never score. In sponsored mode the creator cannot re-buy into their own challenge at all, because they earn a share of every ticket including their own.

Choosing the prize split

Two distribution modes, plus a third shape specific to challenges.

Leaderboard tiers. Fixed percentages by finishing position. Winner-takes-all for duels; 50/30/20 for a small field; 40/25/15/12/8 when you want five paying places. Concentrated and high-variance.

Pro rata. Everyone eligible takes a share proportional to their score. Smooth and inclusive, but in a PnL contest it pays a lot of small cheques and blunts the reason to fight for the top.

Percentage-of-field buckets. Name slices instead of ranks: the top 10% share 60% of the pool, the rest of the top half share 40%. The winner count resolves against the final eligible field, so one config works whether twelve people or five hundred turn up. Inside a bucket, payouts decay down the ranks on a poker-tournament curve — top-heavy at the head, with a long paying tail so the last paid seat still gets a visible cheque. A smoothing pass guarantees no lower rank ever out-pays a higher one.

Buckets are the right default for a challenge whose turnout you cannot predict. Fixed tiers are the right choice when you know the field — a duel, or an invite-only private challenge. The shares in a bucket config must total exactly 100%, and the validator rejects anything else: a bucket ladder resolves its winner count from the final field, so a shortfall would strand that slice of the pool with no position to attach it to. Fixed tiers are no more forgiving — an unfilled rank is simply unpaid — but at least that gap is predictable when you set them.

Publishing

Creating a challenge is free. It exists as a draft until money arrives.

• Public ticket mode: pay your own entry ticket. On confirmation the draft goes live and you are participant one.
• Sponsored mode: pay the sponsorship. On confirmation the pool is seeded and the challenge publishes.
• Private: approve your entrants, everyone pays, then publish explicitly. A private challenge will not publish while an approved entrant still owes their ticket — otherwise it would start with an empty seat somebody had been promised.

Payments are USDC on Base. The transaction hash is recorded before it is verified, so a closed browser or a slow confirmation does not lose the payment: a background job reconciles it and credits the pool when it settles. A hash that no healthy node has seen for an hour is cleared so the entrant can retry with a new transaction.

Running and settling

While it runs, the leaderboard ranks realized plus unrealized PnL, refreshed every few minutes from venue data. Participants also stack VXP for the volume they trade while competing.

At the end, standings freeze and settlement runs once. It ranks every entrant who paid, was not refunded or rejected, actually started scoring, and cleared the volume floor — then splits the pool by your chosen distribution. Any lottery draws still owed are run first, and the leaderboard splits what the lottery did not take, measured against what was actually drawn rather than what was projected.

Payout is a manual USDC transfer on Base against the settled reward records. That is deliberate: a human looks at the split before money moves, rather than after.

A note on lotteries

Challenges can carry a daily draw, with two constraints worth knowing before you plan around it. It is sponsored mode only, because the draws are paid out of the sponsorship — money banked before the challenge publishes rather than a guess about how many tickets will sell. And the challenge must run at least 24 hours, or no draw could ever fire. The total reserve cannot exceed the sponsorship.

The constraint exists because the alternative was insolvency by arithmetic: fifty seats at $50 looks like $2,375 of available prize money, a creator reserves $2,100 of it for draws, two people actually join, and $95 of real money is carrying $2,100 of promises.

Start here

Create a challenge, or read the challenges overview for the shorter version. If you want to bankroll rather than play, sponsoring a challenge covers the economics from the other side. If turnover rather than profit is what you want to reward, volume competitions run different machinery entirely — the complete guide compares them.

Everything above is also available programmatically. The agent docs cover the wallet-signature flow that lets a bot create, fund and run a challenge end to end.