fair is an autonomous venture fund. every research, evaluation, and trade decision is made by an AI agent system running on Base.
most funds rely on insider deal flow and slow-moving committees. fair runs on signal — live market data, social discourse, on-chain truth — processed by specialized agents that decide what to back, when, and how much.
profits from trades buy back the $Fair token. better signals → better trades → harder buyback. holders share in the outcome.
scouts are the human edge in fair's signal pipeline.
agents are fast and systematic — they scan thousands of tokens, check builder wallets, parse social feeds. but they can't recognize a builder from a prior project. can't feel a narrative forming before it has a name. can't catch insider info that lives in DMs not in tweets.
your single goal: surface what agents miss. you find the signal. fair's judgement agent evaluates it. when the signal is strong, fair acts on it.
be as early as the best hunters, but autonomous. you bring the spot. fair brings the firepower.
one submission, from tweet to performance payout.
two paths. both public. both equally valid.
tag @FAIR_VC on X with:
saw a KOL post strong alpha? reply to their tweet with:
@fair_vc must appear in the first 280 characters of your tweet. if it lands only in the long-form tail (beyond char 280), x's api blocks our agent from replying — your submission goes unprocessed, no points.
but DON'T start your tweet with @fair_vc. tweets that begin with @ are treated by x as direct replies — reach drops sharply, and the algorithm deprioritizes them.
right: "$AEON — dev @aaronjmars shipped a16z-tier framework, @pmarca just followed. @fair_vc 0xbf8e..."
wrong: "@fair_vc $AEON dev shipped..." (starts with mention → reduced reach)
wrong: tagging @fair_vc only after a 1000-char essay (lands in long-form tail → no reply possible)
spotting alpha in someone else's feed before broader attention catches it is exactly what scouts uniquely do. agent reads the parent tweet too — if it's solid (thesis ≥ 8/10), you earn the same thesis bonus as if you wrote it. minimum text from you is fine if the parent has substance.
different cap = different bar. plus a third path for writers.
edge: finding a real project before the market notices.
what scores high: named builder with track record, real product (not just roadmap), early in a live narrative, you arrived before the move.
performance reality: +1000% to +5000% realistic on a real find. bonus: 50–120 points on top of base.
edge: insider info that isn't already on page one of search results.
what scores high: wallet accumulation by tracked addresses, KOL follow that means something, integration confirmed pre-announcement, mechanism analysis the market is mispricing.
performance reality: +100% to +300% great outcome at this scale. bonus: 5–15 points. compensated by larger position sizing on signal/trade.
thesis quality is graded separately from token outcome. strong analyses earn ≥8/10 (+3 pts) and ≥9/10 (+7 pts) bonuses regardless of whether the token pumps. what wins: specific numbers, verifiable claims, mechanism described, peer comparison, no filler.
on low-caps, a named builder is enough. on high-caps, alpha is required — public info on a known project will only earn "saved" at most. lies cost: 3 falsifications = permanent tier downgrade.
your submission lands in one of four buckets. points awarded immediately.
weak, duplicate, or unverifiable. no performance window opens. doesn't damage reputation unless pattern of false claims.
fair adds to research corpus. doesn't act now. performance bonus still pays — a KB call that does 20x earns big.
fair publishes to its feed. may take a small position. credits your handle in the post.
fair takes a real position. high conviction. rare. credits your handle. execution moves price.
performance bonus scales with how your call performs, regardless of verdict tier. a knowledge-base call that peaks at +2000% earns 1 + 60 = 61 points. a trade-tier call that goes flat earns 20. the token's outcome matters more than the bucket it landed in.
s0.5 is the calibration phase. the agent learns from every call. the team reviews every TRADE verdict. the system is still being tuned and is self-improving with each cycle.
early in the season, a TRADE verdict may not always result in actual on-chain execution while we sharpen calibration. that does NOT unearn your points — TRADE base (20 pts) + bonuses are credited regardless of execution. performance window opens the same way.
our target for the season: at least one scout-sourced trade actually shipped onchain. that's the bar we're working toward — be the scout who clears it.
two metrics tracked per scout. one resets, one is forever. one pool pays out at season end.
season points. resets every 2 weeks at season end.
determines current rank and your slice of the reward pool. at season end, the 100M pool splits proportionally to points earned during the season.
what fills it: base points + performance bonus + thesis quality + impressions, all accrued during the active season.
impressions: the reach of your submission post earns points too. 1 point per 1,000 views, read once about three days after you post. only your first call in a given thread counts, so give each call its own post instead of burying them in one reply chain.
lifetime score, 0–100. never resets. accumulates across all calls.
in S0.5 it's tracked but mostly passive. in next seasons reputation becomes a multiplier on submission evaluation — the higher your rep, the more weight your future calls carry through scoring.
build rep now → bigger payouts later. the earliest scouts get the longest compounding runway.
recomputed weekly (Sunday 02:00 UTC) and on-demand.
top-3 in any season's leaderboard → GP review. doesn't matter how you got there — trades, signals, KB-class mega-calls, thesis quality — what matters is season rank. final invitation is team call.
your token's FDV is tracked daily for 30 days from submission. we record the peak — once you've earned a tier, dumps don't unearn it.
| peak fdv gain | bonus added | cumulative |
|---|---|---|
| +100% | +5 | 5 |
| +200% | +5 | 10 |
| +300% | +5 | 15 |
| +500% | +5 | 25 |
| +1000% | +5 | 50 |
| +2000% | +5 | 60 |
| every +500% above 2000% | +10 | unbounded |
a knowledge base call (1 pt base) that peaks at +2000% earns 1 + 60 = 61 pts.
at +5000%, the same KB call earns 1 + 120 = 121 pts. outearns a flat trade call.
| thesis quality | bonus |
|---|---|
| ≥ 8 / 10 | +3 pts |
| ≥ 9 / 10 | +7 pts |
awarded regardless of token outcome. writers and finders both have paths to the top.
visible publicly in scout chat. every broadcast shows your thesis grade out of 10 + bonus earned. writers learn from each other's grades.
we sample once per day. if your token spikes intra-day and dumps before our snapshot, that peak is missed — we read the daily close, not the wick. submission timing also matters: a call entered right before a big day captures the move.
reputation in S0.5 has three active components. one good hit can carry the whole score.
| component | cap | what it measures |
|---|---|---|
| accuracy | 60 | hits weighted by magnitude. +100% = ×1.0, +1000% = ×2.0, +10000% = ×3.0. one mega-call can max this alone. |
| longevity | 20 | seasons active + non-reject submission count. 3 seasons = roughly max. |
| volume × quality | 20 | submissions that pass the thesis-quality gate (≥7/10). 4 quality submissions = cap. |
recency boost ×1.5 on activity components if active in last 2 seasons. decay starts after 1-season grace, -5% per missed season, floor 0.7.
three channels. nothing specific goes public. cohort sees a shortened version. full analysis stays internal.
agent replies to your tweet with a single templated message — same for everyone, no specifics. doesn't reveal what your info was, what the verdict was, or what the edge looked like.
why generic: anything specific leaks the scout's edge to the broader market, and the verdict + fair's read stay internal.
each submission posts to the cohort chat with a short summary:
cohort sees what's flowing + how the writing graded. no token-rubric scores, no dim breakdown.
the complete breakdown — builder check, on-chain signals, narrative phase, catalyst verification, dim scores, reasoning — only the team sees this.
used for calibration, decision-making, and reviewing edge cases. does not reach scout chat in any form.
public stays opaque so fair's edge doesn't leak. cohort gets enough to see what's being surfaced and learn from peers' patterns. full reasoning stays internal so the rubric can't be reverse-engineered.
what the agent scores you on. specific verifiable claims beat vague hype on every axis.
three patterns that score high. study what they share.
$DONUT by @Glaze_Corp @heeshilio 0x1234...abcd @qrcoindotfun auction winner validated. 8,967 casts / 2,669 authors in 30d. king glazer throne mechanics: 4 $donut/sec, 5d to first halving, 80/15/5 splits fund buy+burn. treasury $86K, $486K total earned. frame-native attention game.
named builders, verifiable metrics, mechanism described, treasury data, halving as upcoming catalyst.
$AEON 0xbf8e... dev: @aaronjmars — NeuroAI researcher, ex-Tela Ventures VC. prior projects: Hyperstitions, soul.md (356★), OpenDia (1788★). product: autonomous AI agent framework, 68 skills, MCP adaptor. catalyst: @pmarca (a16z general partner) just followed @aeon — first a16z engagement with this project.
builder fully named with track record we can verify. product specifics. catalyst is verifiable insider-tier alpha — we check pmarca's follow list.
$CLANKER 0x1bc0... DWR signaling second buyback on sun/mon. if mirrors first cycle, scale jumps 7x. ~$130M vol on 24-25 oct pushed 1% fee to treasury ($1.3M). half tokens, half eth. one-third taxes, two-thirds recycled = ~$430K round, up from first $65K. closest fee-flow rival is Pump, gap down to 20-30%, mcap differs ~38x. repricing inevitable.
specific catalyst (buyback), dollar amounts derived from on-chain data, peer comparison, timing edge.
three patterns that get rejected. avoid them.
$CRED — next 100x, dev promising big stuff 0x1234...abcd
no specifics, no verifiable claims. instant reject.
$CLANKER — Neynar just bought Clanker. buybacks huge. OG launchpad. repricing inevitable.
every word public knowledge. for established projects, this only earns "saved" — no alpha means no signal. resubmit with insider data ("wallet X just accumulated 3%") and you'd cross the bar.
$X is at intersection of AI + DeFi + social. early stage. big things coming.
no builder named, no on-chain evidence, no specific catalyst. narrative buzzwords without substance.
scheduled posts to the scout chat. standings twice a day, daily recap, weekly recap on sundays. (live submission posts — see §11.)
current season points + lifetime reputation top of the chart. who's leading.
today's submissions list + best-of-day callouts. catch what you missed.
refreshed standings after the day's submissions land.
top performers + best submissions and their performance + days left in season.
the scout chat is your real-time view into everything fair is evaluating from the cohort. every KB-save, every signal, every trade — all of it lands in the same room. coordinate without leaking to broader market.
the loop fair was built for. activates in S1 if s0.5 proves out.
if the loop works — agents calibrated, scout signal proven, buyback shipping onchain — we open the gate. next phase: permissionless season. no curated cohort, no allowlist. anyone with a wallet can scout and earn from a trading engine that learned from this round.
s0.5 is the proof. every TRADE shipped, every performance bonus paid, every reputation point earned is evidence the system holds when we hand it to the world. early scouts who carried the calibration phase get the longest compounding runway in everything that comes after.
they earn rewards from their calls and from the buyback pressure their calls eventually create. the loop is the product.