whitepaper
Skill, separated from capital.
every figure on this page is imported from the same constants file the game engine executes — the document cannot disagree with the product.
# abstract
HOODFUND applies the funded-account model — long established in foreign-exchange prop trading — to memecoins and tokenized stocks on Robinhood Chain. A trader pays a fixed entry fee, receives a simulated 10 ETH stack, and trades live market prices through three consecutive challenges of rising difficulty (+50%, +100%, +200%). Clearing all three pays a fixed cash prize of 15x the entry fee, in USDG, straight to the trader's wallet.
The design goal is a market where skill is separable from capital and from luck-of-the-launch, and where the cost of losing is bounded and known before you start.
# the problem
Memecoin trading punishes the wrong things. Position size dominates skill, so the trader with the most capital wins the most even when their timing is worse. Losses are unbounded, so one bad night removes a participant permanently. And there is no durable record of performance: a screenshot of a winning trade says nothing about the losing ones.
Prop firms solved the structurally identical problem in FX by separating the evaluation from the capital: demonstrate skill on a simulated balance under enforced risk limits, and the firm supplies real money to whoever passes. HOODFUND keeps that evaluation and points it at the most volatile market on earth.
# mechanism
Every phase restarts the trader at the same balance against a higher target, so the leaderboard compares decisions rather than bankrolls. The full parameter set:
The exposure cap is measured at entry: a position may not be built beyond 33% of equity, but a winner is never forced down when it grows past that share on its own. Risk is constrained where it is taken — in the sizing decision — not by punishing a correct call.
The drawdown floor is checked on every price tick against net liquidation value — cash plus every position marked at the live quote with the exit fee already deducted. There is no equity number on screen that could not actually be realized.
# the fill engine
Trades are simulated against live market prices. No order is routed to a venue and no on-chain swap occurs, so a fill never moves the real market and never competes with the trader's own wallet.
The fee is not revenue — it exists so that overtrading carries the same drag it carries in the real market. A simulator with zero cost per trade rewards churn, which is the opposite of the skill being measured. The 33% exposure cap and the 10-fill minimum close the two remaining exploits: the all-in coin-flip and the one-lucky-entry pass.
# listing floors
A coin is buyable only if it clears every floor at the moment of the fill:
The attack these prevent is straightforward: since fills settle at the quoted price regardless of size, the engine is only fair if the underlying market is expensive to move. A trader who can push a thin pool for a few hundred dollars could multiply a simulated position against a price they created. The floors make that manipulation cost more than the prize is worth.
Robinhood Chain has no single launchpad: coins arrive through hood.fun, Pons, Clanker, Bankr and others, and every one of them lands on a Uniswap-style pool that the indexers see within minutes. So provenance is the pool, not the launchpad. Two classes of token are excluded by name: the quote assets themselves (WETH, USDG) and Robinhood's tokenized stocks and ETFs, which are real securities, not the game's universe.
why the floors exclude tokens that look eligible
# stocks & leverage
Robinhood Chain carries Robinhood's tokenized stocks and ETFs — NVDA, TSLA, SPY, QQQ, GME, GOOGL, AMD and others — on the same pools the memecoins trade on. The desk treats them as a second asset class: their own lane, a 0.1% fill fee (a stock pool's real take), and the only floor a stock needs, pool depth of at least $5K. A token is classed by its on-chain name, read once through multicall and remembered — the pool indexers only carry the ticker.
A position is margin plus a bet of qty tokens at an entry price, long or short. Its value at any mark is margin + pnl − exit fee, floored at zero: at 10x a 10% adverse move consumes the margin and the engine closes the bet where it stands, on the same tick that checks the drawdown floor. Memecoins remain spot-only and long-only — a thin launchpad pool is not a market to lever. Both classes count toward the same challenge: a stock fill is a fill, a stock pnl is pnl.
# when a token falls under the floor
Eligibility is evaluated continuously, not once. A token you already hold can drop below the market-cap floor while you own it. Exactly three things happen, and nothing else:
The first rule is absolute: no market-cap check, liquidity check or eligibility check is applied on the sell path, in the engine or in the interface. A collapsing token can cost you money, but it can never trap you in a position.
The third rule is what keeps the evaluation honest: a held token whose market cap falls through the $30K floor is force-closed at the live market price on the next tick. The loss (or gain) is realized immediately — the fill is recorded like any other sell and the proceeds return to cash. Holding a dying token is never a way to freeze a losing position, and a stale price can never hide a loss from the drawdown check.
Until the auto-close lands, the token stays pinned in your token list with its real market cap, so a position is never something you have to hunt for. If the coin climbs back above the floor later, it becomes buyable again on its own.
# accounts & integrity
An account is an EVM wallet. Signing in means signing a nonce-bearing EIP-191 message with that wallet — there is no password, and the address you sign in with is the address any payout is sent to.
For a signed-in wallet the run is executed on the server, not in the browser. Fills are priced from server-side marks, and the exposure cap, drawdown floor, minimum-fill count and challenge expiry are enforced in the engine before anything is written. The browser holds no authority over the outcome: it renders a mirror of a state it cannot modify. Requests without a valid session are rejected outright.
the preview mode is not an account
# price infrastructure
The feed is layered, each layer replaceable without touching the game:
- Discovery. The eligible universe is rebuilt every couple of minutes from GeckoTerminal's new, trending and volume-sortedRobinhood Chain pools plus DexScreener's boosted tokens, filtered by the listing floors, deduplicated to the deepest pool per token.
- Marks. Live prices arrive by polling — a single batched DexScreener call for the hot set — every second, coordinated across instances through a database lock so a thousand terminals cost one upstream call. Pools quoted in USDG are converted to ETH at the live WETH/USDG rate.
- Candles. OHLCV history loads per pool and timeframe, then the current candle is painted forward in real time from the mark feed between full resyncs every 20 seconds. Clock skew between the client and the data source merges into the last candle rather than dropping ticks.
- Isolation. Every upstream call is proxied and cached server-side, so a thousand open terminals cost the same upstream quota as one.
planned
# the stack
One deliberate constraint: every game rule lives in a single constants file that both the engine and every page import. Change a number once and the terminal, the checkout, the landing page and this document all follow. Nothing here is hand-copied.
# the prize
Clearing all three challenges pays a fixed cash prize in USDG, sent straight to the wallet you signed in with. There is no funded account to manage and no profit share — you win, the firm pays, it is over. Two entry tiers, one gauntlet:
The rules are identical in both tiers — same targets, same floor, same fills minimum. The free roll exists so the gauntlet can be attempted without spending anything; the prize scales with the entry, not the difficulty.
The economics are a prop-firm classic: the prize is 15x the entry, so the pool of entry fees is only solvent if fewer than one attempt in 15 clears all three. The targets ( +50%, +100%, +200% back to back, under the drawdown floor and the exposure cap) are set to make that the case. Most attempts fail, by design.
# payouts
The prize is paid automatically from a dedicated payout wallet — deliberately not the treasury — to the address you signed in with, after a short safety window.
The prize is recorded against the winning run the instant challenge 03 is cleared, and a unique constraint on the run means a replay or a double-submit can never pay it twice. The delay and the ceiling exist so an anomaly can be caught before money moves rather than after.
preview build
# the HOODFUND token
HOODFUND is the platform token, and unlike most, it has exactly one job:
No governance theater, no gated formats, no claim on prize pools. Holding HOODFUND does one thing: it makes every attempt cheaper. That keeps the token's value proposition honest — it is a discount coupon with a market price, and demand for it scales exactly with demand for seats.
Entries paid in HOODFUND are burned by the firm — sent out of circulation, verifiable on-chain. The treasury never re-sells entry tokens into the market: every seat bought with HOODFUND is supply that permanently disappears. Demand for seats becomes deflation, not sell pressure.
not deployed yet
# limits & risks
Stated plainly, because a whitepaper that only lists strengths is marketing:
- Simulated fills are not real fills. A result here demonstrates timing against live prices, not that the same size could have been executed on-chain at the same price.
- Market data is third-party. Prices, pool statistics and eligibility checks depend on external indexers. If they are wrong or unreachable, quotes and floors are affected.
- The universe depends on public indexers. Eligibility is read from GeckoTerminal and DexScreener's free tiers, which are rate-limited and can change without notice. Two sources cover for each other, but a day both are down is a day the universe stops refreshing.
- One wallet is not one person. Nothing today proves a single human controls a single wallet, so the free tier is farmable by someone willing to spread across addresses. It is capped in reward for that reason, and per-identity limits are the obvious next step.
- Payouts depend on an operator. They are automated, but from a wallet this firm controls, on infrastructure this firm runs. Settlement is not trustless and this document does not claim otherwise.
- Floors are a deterrent, not a proof. They raise the cost of manipulating a thin market above the prize on offer; they do not make manipulation impossible.
- Entry fees will be at risk. Once payments go live, a losing run forfeits the fee. Nothing here is investment advice and no return is promised.
# verifying this document
Every number above is rendered from src/lib/rules.ts — the same constants the fill engine, the risk checks and the checkout execute. If the product changes, this page changes in the same commit, or the build fails. Where something is simulated or unimplemented, this document says so in an amber box rather than implying otherwise.