---
name: rh-chain-radar
description: The Robinhood Chain early-project detection system — three-track radar, scoring rubric, the impersonation trap, chain data endpoints, and the verdict on Ben's 16 alpha tools. Read before screening any RH-chain token or building anything that watches that chain.
sources: [cowork]
aliases: [robinhood-chain-radar, rh-radar]
---

# Robinhood Chain Radar — built 27 Aug 2026

Companion to [[crypto-token-research]] and [[alert-bots]].

## The system
Toolkit rhlib.py / rhradar.py / rhscore.py + radar.db, delivered to Ben as files.

STAGE 1, three tracks:
- (A) smart money — balance-diff the 9 scored wallets, paid entries only (>=$1,000 AND a matching outflow in the same tx), tokens <30d old. DUPLICATES the live strongtrader.service — do not deploy this track, only B and C.
- (B) protocol deploy — poll Blockscout verified contracts, cluster bespoke names by stem, drop launchpad templates, require >=3 distinct system-shaped names sharing one deployer. FIRES BEFORE ANY POOL EXISTS. Earliest signal found on this chain.
- (C) survival — index new pools, gate at T+24h (v24>=$300k, liq>=$120k, holders>=800) and T+72h (vol>=$1M, liq>=$150k, holders>=1500). 96h cap on gate eligibility so stale pools cannot pose as fresh finds.

STAGE 2: kill checks K1-K6 (impersonation, fake liquidity, spoofed FDV, dead, no product, one-way market) run BEFORE the pitch, then a 100-pt score — product reality 30 / distribution 20 / market structure 20 / timing 20 / smart money 10.

STAGE 3: three differentiated QC passes — identity (am I on the right contract), source (can each number lie), inversion (what would make this false).

## Chain data surface, all free, no keys
GeckoTerminal network slug `robinhood` (the new_pools endpoint breaks with `?page=1`, call it bare) · DexScreener chainId `robinhood` · https://robinhoodchain.blockscout.com/api/v2 · https://rpc.mainnet.chain.robinhood.com · chain id 4663.
Base rate measured 27 Aug: ~6,990 new pools/day (200 in a 41-min window, 97 from pons-v2 and 73 from uniswap-v4) and ~2,100 verified contracts/day.

## The impersonation trap — the most valuable finding
Every major RH-chain winner has impersonator contracts carrying the same ticker and name, with huge FAKE liquidity and near-zero volume: Nektar-NET $48.2M liq / $0 vol / 2 holders; fake DELTA $4.7M/$0; fake PRINTER $2.9M/$2; fake DTF $7.7M/$1 with a $126B FDV, created six hours after Ben screened the real DTF. STONKBROKER has four contracts, three of them copies with 11k-32k holders each.
Ranking candidates by liquidity surfaces the fakes FIRST. Discriminators: turnover (real 0.3-10x/day, fakes 0.000x), pair count, and presence of socials + a website.

DTF has three live claimants with lookalike domains AND lookalike X handles — 0xeE5576Fa1Bcaa380e591D01245f406f3f384eb01 (downto.finance, @downto_finance, first pool 25 Aug 01:35 UTC, the one Ben screened), 0x2ec89afba136119c5252fc47d14e2bd2144b10d2 (downtofinance.cc, @downtofinancerh, first pool 5h50m later, now ~8x the original's 24h volume), 0x5Edb417509a869b1D8222dEa05257B8c8Ba00361 (dtf.finance, 30 Jul). Which is legitimate is UNRESOLVED — settle it before sizing.

## What "a real project" means here, empirically
The token contract tells you NOTHING — PONS, DELTA, CASHCAT and STONKBROKER are all launchpad templates (PonsLauncherToken, LaunchToken, CollectionToken), the same template behind 113 throwaway tokens in one six-hour window. What all eight winners share is a working dapp at their own domain, and most are native chain infrastructure (bonds, liquidity, launchpads, indices, lending).
First-24h volume spans $59k (CASHCAT) to $6.4M (PONS). CASHCAT and NET only clear a bar at 72h, so the screen must gate twice. INDEX ($94k over three days, now $21.9M FDV) is the acknowledged miss — no volume gate finds it.

Track B's first live run found two real pre-token protocol deploys: "Arrow" (sArrow + ArrowStaking + ArrowBondingCalculator + ArrowAuthority — an OlympusDAO fork, deployer 0x16E7C1B229d5701e75Cccb68C13fcbf98FE5c027, routes through a Safe, $1.13 of ETH left) and "Trace" (TraceToken + TraceTimelock + TraceDescriptor, deployer 0x9aD53d4c4ED66D3a69f40816939D2515FF92597B). Neither had a pool. Independent deployers, so not one factory spraying forks.

## The X layer (added after Ben asked for the gap to be closed)
rhx.py, two backends, same output. AUTOMATED: twitterapi.io advanced search on the key xalert already uses — $0.15/1,000 tweets, about $3/month at radar volume. INTERACTIVE: with no key it prints the X search URLs; run them through the Chrome bridge on the Mac and paste rhbridge.js to get structured rows. Adds kill K7 (zero posts cite the contract AND zero cite the ticker in an RH context = no narrative owner) and K7b (ticker chatter cites a DIFFERENT contract more than this one = narrative migrated, you are screening the abandoned shell). Concentration and small-account share are reported, not scored.
X settled the DTF question: every organic post names @downto_finance (the original 0xeE5576Fa..., dev described as "ex ohm dev"); the downtofinance.cc lookalike returns five posts, one calling it a slow rug and four collab-spam bots. The contract doing $38.7M/day has NO narrative behind it — wash volume, not adoption.

## Verdict on the 16 alpha tools Ben supplied
RETRACTED — umi.bot is a SKIP, not a buy. Re-checked its 1.68MB bundle: zero hits for swapExactETH, exactInputSingle, UniversalRouter, swapRouter, tokenIn, amountOutMinimum; it carries ERC-721/1155 handling and its routes are /mints /wallets /customs /disperse /tags /terminus /plans /billing with no /swap or /trade. It is an NFT minting bot and CANNOT buy an ERC-20. Its chain-4663 tuning is real but exists to win NFT mint races, which is irrelevant to a weeks-to-months holder. The error was letting "real chain engineering" stand in for "useful".
purealpha.app — downgraded to TEST ONE MONTH at $29, NOT the $299 lifetime. The insider follow-graph convergence signal is genuinely distinct and not reconstructable from X search, but whether it fires BEFORE a launch is unverified without an account and it has no chain filter. Pass condition agreed: within 30 days it must surface at least one RH-chain project Ben had not seen, before that token has a market.
The real answer to the X gap is twitterapi.io, which Ben ALREADY PAYS FOR (xalert holds a websocket against it) — ~$3/month for the radar's entire X layer.
FREE TIER ONLY: cryptorank.io's Robinhood ecosystem directory (188 projects; it is what identified Pons and Memecoin.Fun as the chain's launchpads).
SKIP the paid tiers of moni.ai ($120-250/mo), rootdata ($78-218/mo), frontrun.vc ($49-99/mo) and waypoint — no chain-4663 dimension at all. Also skip 985monitor, alphatrack, watchoor (farming), snipenowbot (no public surface).
RH-native but NFT-only: guap.wtf, mintgo.fun, neverfuckingtrade.com (142 RH entries, free Telegram alerts).
Unverifiable without login: alphagate.io, orbitalhq.
NONE of the 16 is an RH-chain token-launch scanner. GeckoTerminal, DexScreener and Blockscout beat all of them, free.

## Deliverable
Artifact "Robinhood Chain Radar": https://claude.ai/code/artifact/e3b7ab15-c0b7-4fcf-879d-0c17e140852b

## STANDING RULE — mandatory on every detection run and every deep dive (Ben, 27 Aug 2026)
- [stated] Ben now holds a purealpha.app PRO subscription. USE IT EVERY TIME on project detection and on every deep dive — check the insider-follow feed for the project's X account before forming a verdict. In-session this needs the Chrome bridge on his Mac (he is logged in there); purealpha has no public API.
- [stated] USE twitterapi.io ADVANCED SEARCH EVERY TIME — it is part of every deep dive, not an optional extra. Key lives in the environment as TWITTERAPI_KEY (same key xalert uses); never ask Ben to paste it into chat.
- [stated] Tracks B and C must feed ONE merged candidate list, not two separate outputs. Detection method is only a source tag on the row; the deep dive and the score run over a single queue and produce a single ranked list.

## Built 27 Aug 2026 — the merged queue (rhqueue.py)
- ONE candidate table. Tracks are only a source tag in a VIA column. A token flagged by two tracks is one row with two sources and gets +8 on the score; two separate lists would have hidden that. A pre-token track-B row absorbs its token via `rhradar.py resolve` rather than being re-filed by track C as a new find.
- Mandatory evidence is enforced in code, not remembered: twitterapi + purealpha must both be on record or the row is PROVISIONAL and CANNOT reach a buy verdict. Proof it works: PCC scored 70 (clears the buy threshold) and was held at WATCHLIST because the X check was missing.
- Commands: rhradar.py b | c | resolve, then rhqueue.py run N | list | checks | evidence <token> <source> <json>.

## purealpha is TRACK P, not just a deep-dive check — upgraded 27 Aug after first Pro use
Read through the Chrome bridge (no API; he is logged in on his Mac). On its first run it added FIVE RH-chain projects neither B nor C had: HOOD10 Index @hood10xyz (index token paying dividends from trading volume, $120k distributed in 24h — structurally the same shape as NET; 3d old, $2.0M/24h, $322k liq, 27 pairs, scored 68), VozDex AI @Vozdexai (liquidity layer, 17h old, $341k/24h, but ZERO posts cite its contract), Aureon @buildaureon (on-chain AI agents, +1328), Bons @bonsfamily (agentic layer on Pons via Grok, +176 — killed K4, zero volume in 6h after a +387% day), PLTR GOYBEAM (+1254). It sees projects before they have any on-chain signature because it watches the insider follow-graph, so it belongs at detection, not at verification.
Extraction note: purealpha's feed is in the DOM but not in innerText (React streaming shell stays "Loading pure alpha…"). Do not read document.body.innerText — query a[href*="x.com/"] and walk up to the row container.

## Defect ledger (each found by RUNNING it, not reading it)
1. Bespoke-contract filter would have rejected 4 of 8 winners — token contracts are launchpad templates. Dropped as a token-level test.
2. Airdrop dust read as conviction — fixed with a $1,000 floor plus a matching outflow in the same tx.
3. Stale pools promoted as fresh finds — fixed with a 96h cap on gate eligibility.
4. GitHub/X 403 read as a dead link — whitelist hosts that block datacenter IPs.
5. THE WORST ONE: resolve_pending ranked matching contracts BY LIQUIDITY and resolved Arrow onto 0x1564491895E51C95bB624B957F828526D4Ca12F4, the decoy with $10.6M pooled and $0 traded — the tool walked into the exact trap the process document warns about. Fixed: rank by traded volume, require $5k+, re-apply the fake-liquidity test, reject single orphan pairs with no socials. NEVER RANK BY LIQUIDITY ON THIS CHAIN.
6. Junk stems (Test/Demo/Mock) alerted as protocol launches — blacklisted.

## BOTTOMED-INFRA SCREEN, 6 Sep 2026 — 33 infra tokens, price shape + insider ledger + revenue
[stated] Ben: "find tokens on Robinhood Chain that have bottomed or are close to bottoming and could be key infrastructure players. PONS was the ideal example, it pulled back while usage did well, and insiders weren't selling at the bottom. I'm hoping DELTA is next?"
Scripts `/tmp/bt/universe.py` (shape), `/tmp/bt/insider.py` (extended cohort), revenue from DefiLlama `overview/fees/Robinhood%20Chain?dataType=dailyRevenue` (150 rows).
- **Bottoming is a shape and it is checkable: days since the lowest close, multiple off that low, higher lows in the recent third vs the middle third, and volume returning.** A low that is hours old is not a bottom. That one filter removed most of the chain: DELTA, PAIR, MANCER, RIPE, HOOKR, INJOH, PROLOGUE, NET, SLVR, TWO, WALL3 all printed their low within a day of the screen.
- **THE TWO THAT PASS BOTH, price shape AND insiders gone:**
  - **WIRE `0x8ECEA3d0E648DB646d824AA51EedeB16aC3d6878`** mcap $630,140 · 85% off ATH · low **31.0 days ago** · 3.63x off it · higher lows 2.42x · volume 1.62x. Insider network 33 wallets sold 20.37% and **holds 0.101% = $631**, 0.000% sold in 24h. The only tokenised bot on the chain. GAP: no DefiLlama revenue line, so its usage is unverified.
  - **SESTRI `0xC1a0C3929f5e12b5CfD6Ae36f197a45beDf79417`** mcap $490,069 · 90% off · low 18.2 days ago · 2.89x off it · higher lows 1.80x. Insider network 29 wallets **holds $0**. GAP: volume trend 0.92x, so no bid has returned. Ben already holds ~$2,400.
- **LUTE / Alandale `0xD1e861CC5Eee7eA88649206b74504D78CCD7AEeA`** is the best USAGE candidate and the unresolved one. mcap $4,247,446 · 52% off · low 3.7 days ago · higher lows 2.39x. Alandale V3 revenue $71,644/30d with **$51,995 of it in the last 7 days = 3.1x acceleration**, a real DEX. Its early cohort has sold **0.000% in 7 days**. BUT the cohort is only **2 wallets holding 24.11% of supply ($1,017,513)** and their identity is unresolved (Blockscout 500s on its holders endpoint). Resolve that before sizing; it is the whole question on LUTE.
- **RVH / Ravenhood `0x96765066f6a040a21EB027167D2315B707c82633`** looked strong on shape (off-low 1.46x, higher lows 2.12x, volume trend 12.96x, revenue +3.7x) and **fails on insiders: 42 wallets sold 1.100% of supply in the last 24h.** Same live-seller shape as TWO. Its top holder is `0xAc558b558E228DE033Cd97C580618C4403CB05a6` RVHStakingPool at 24.71%.
- **DELTA IS NOT IT YET.** Its lowest close was the hour of the screen. The good half is real (insiders stopped, 0.004% in 24h) but there is no base, and DefiLlama does not track its revenue at all. More important, DELTA has **no holder mechanism** (`owner()` reverts, no staking, no fee share), which is exactly the trait that puts it in this chain's worst-performing category.
- **CORRECTION TO BEN'S PREMISE ON PONS, worth saying plainly.** PONS is not a current pullback setup: its ATH was 05 Sep 17:00, it is only 10.6% off, and Pons V2 revenue is $8,986,967/30d with **$6,615,424 in the last 7 days**. Its launch cohort held **0.00% of supply at the ATH**, which is why it did not dump. PONS is what the pattern looks like AFTER it works, not a template to buy today.
- **APPLY THE REGIME FINDING ABOVE.** "Infrastructure the holder must OPERATE" is the WORST category on this chain at +29% median 30d, against +1,925% for tokens that pay the holder automatically. So a bottomed infra token with no payout mechanism is fighting a 66x category headwind. Screen for the payout mechanism BEFORE the chart.
- **The chain's fastest-growing earners have no token.** o1 Launchpad did **$768,417 of its $792,035 30-day revenue in the last 7 days** (4.16x acceleration) and the only "o1" ticker on DexScreener is a $2,896 dust contract. Snuggle $230,940/30d, no token. That is the recurring shape here: revenue accrues where there is nothing to buy.

## Deep dive 27 Aug 2026 — five candidates, contract-level. Artifact: https://claude.ai/code/artifact/c50ca5c0-7cfb-4261-998a-1aab4353068a
- **ARROW / Arrow Finance 0xf2915d1e3C1B0c769d0c756Ec43F1c1f6c99cD03 — the pick. STARTER $5-7k.** Overcollateralised CDP (Liquity-shaped) native to RH chain: deposit crypto, stablecoins or TOKENISED STOCK, mint aUSD, keep upside without selling. Docs cover vaults, collateral tiering, stability pool, liquidations, surplus buffer, governance, and a "Market-hours Economics" section on stock-token collateral outside trading hours. It is the one project aimed at the chain's actual reason to exist. NO TAX. liq $2.90M > mcap $2.34M because 58.5% of a 10M fixed supply sits in the Uniswap V3 pool; team SafeProxy 10.0%; largest EOA only 1.37%, then 0.88%, 0.81% — float genuinely spread. $15k moves price 1.02%, round trip ~1.4% — by far the cleanest exit on the chain at Ben's size. BUT NOT LIVE YET: aUSD has ZERO market on RH chain, and ArrowStaking/sArrow/ArrowBondingCalculator (verified today) show 0/3/0 transactions and zero balance. Site claims "Swap, Borrow and Launch are live now" — the chain does not agree. Add to $12-15k on ONE trigger: aUSD gets real liquidity, or ArrowStaking tx count goes above zero.
- **HOOD10 0x0D257cA40d40090BE60C2d2Ed5bB3535392838cc — real project, uninvestable price. WATCH, no position.** EIP-1167 clone of CashCatTokenV2; fee hook CashCatHookV2; 5% BUY + 5% SELL tax confirmed by eth_call (taxBps=500). I got it wrong TWICE before getting it right: (1) the token contract has no dividend logic and the hook's claim() is gated to the creator emitting CreatorFeesClaimed, so I nearly called it marketing; (2) I scanned 300 transfers on the token DEPLOYER (0xb3e36d3a) and found zero outgoing — wrong wallet. The real distributor is 0x9f3edbfae8014d55e328b6dd966c6c95e75f6210 (7,643 txs, 76,125 transfers), found by tracing what HOOD10 HOLDERS received. It is genuinely running: the last 400 transfers paid 40 wallets exactly 10 tokens each, equal-weighted across YOLO/TENDIES/INDEX/HMM/PIPEDOG/AI/STONKBROKER/PONS/CASHCAT/DELTA, pushed not claimed. Scale is tiny — that batch was ~$7.55 total, about $0.19 per wallet; the site's own numbers say $185,623 across 16 epochs and last epoch paid $2,794. Purealpha's "$120k distributed in 24 hours" blurb does NOT reconcile with the project's own dashboard — treat purealpha blurbs as project marketing, not data. Round trip at $10k = 21.2% (5%+5% tax plus 5.6% impact each way), so a 1.27x just to break even. Distributor contract is UNVERIFIED (0 bytes source) — mechanism observable, logic unauditable. Triggers: distributor gets verified, or pool passes ~$1M.
- **VAI / VozDex AI 0x33316E5930C659fEDF2d3C4cCa7079461F05E3a1 — PASS.** Contract NOT VERIFIED (zero source) for a protocol whose whole proposition is custody of liquidity. Zero X posts cite its contract address. Round trip at $10k = 55%. Mcap fell $413k -> $295k and liq $62k -> $52k within one session.
- **AUREON 0xd293291060334D42E5DBEa6fb854C231AF527777 — PASS, un-sizeable.** EIP-1167 clone of FlapTaxTokenV3, 1%/1% tax, real site + public SDK repo, but $46k pool means a $10k round trip costs 63%. Revisit only if the pool passes $400k.
- **PCC / Pussy Cat Club 0xbf3e53713a53e9c3d5d1ddc25dd2c65669244663 — PASS, it is a meme by its own admission.** Its homepage reads "No roadmap, no promises, no dogs… No utility. No promises." It scored the highest on the board (70) which exposes a REAL SCORING FLAW: product_reality awards points for a website resolving and rendering as a dapp WITHOUT READING WHAT THE SITE SAYS. Fix needed = a content check on landing copy, not a fetch check.
- Book context applied: RH chain is $15,300 = 9.7% of Ben's ~$158k book (STONKBROKER $12,900 + SESTRI $2,400). A $5-7k ARROW starter takes the chain to 13-14% and does not diversify. Direct overlap: HOOD10's basket INCLUDES STONKBROKER, which Ben already holds at $12,900 — buying the index doubles him into a name he is already sized in, at a 5% entry tax.

## CORRECTION 27 Aug 2026 — ARROW, and the two defects Ben's chart question exposed
Ben asked why the ARROW chart looked crazy (ran to ~$32M early July, died to ~$2M, sideways since end of July). He was right and the deep dive was wrong.
- ARROW real history from GT DAILY candles: launched 3 Jul on V2; V3 WETH pool 10 Jul at $1.62 ($16.2M mcap); **ATH $2.9507 on 15 Jul = $29.5M**; bleed to $0.7389 by 22 Jul; **23 Jul it fell -74% IN ONE DAY on only $114k of volume** (low $0.1276); five-week base $0.13-0.34 with volume down to $1.6k/day; volume returning 25-27 Aug ($158k/$50k/$56k) as staking/bonding shipped. Now $0.2558 = $2.56M = **-91.3% from ATH**.
- **DEFECT 7 — ATH LOOKBACK TOO SHORT.** rhscore used 200 HOURLY candles (an 8-day window) so it reported ARROW as "-29.3% from an ATH set 40h ago" when the true drawdown was -91.3%. On a chain whose interesting tokens are 6-8 weeks old this is severe. FIXED: rhscore now pulls DAILY candles (limit 180) and falls back to hourly.
- **DEFECT 8 — I SUMMED reserve_in_usd AS IF IT WERE EXIT DEPTH.** Third time falling into the chain's core trap. reported liq is NOT tradeable depth; the only honest measure is `balanceOf(quoteToken, poolAddress)`. New file `depth.py` in the toolkit does this. ARROW reported $2.9-3.0M aggregate; the "main" V3 ARROW/WETH pool (reported $1,491,096) actually holds **1.47 WETH = $3,675** plus **5,851,894 ARROW = 58.5% of total supply** — a V3 LP that got run over on 23 Jul and converted entirely into the falling asset. It gives no exit and by standard V3 mechanics it SELLS into any recovery. Could not read its tick range (this Uniswap fork emits no standard Mint events) so the price at which the wall activates is UNRESOLVED — that is the open question on ARROW.
- **But the correction cut both ways — do not over-correct.** The USDT/ARROW 1% V3 pool created 21 Aug holds **1,248,084 REAL USDT**. I first scored it phantom because I assumed 6 decimals for a stablecoin; **USDT on Robinhood Chain has 18 decimals** (0x991b3ec20e8aaf19ff2473c751c05ec167b2b4e1). It is IN RANGE (internal price 3.9746 ARROW/USDT = $0.2516 vs spot $0.2558) and tick math gives **$125,607 sellable for a 5% move, $254,566 for 10%, $523,712 for 20%**. Someone placed $1.25M of stablecoin bid at the base, a month after the crash and three weeks before staking shipped — the most interesting single fact about the token.
- **ARROW routing rule: real depth is on the USDT pair, which carries only $2/24h — routers do not use it.** Default routing sends you to the V2 pool (33.4 WETH = $83,346, $10k sell = 10.7% impact). REVISED TRADE: $5k, entered MANUALLY against USDT, never on default routing.
- Method limits to remember: balanceOf(pool) works for v2/v3 (pool holds its own reserves) but NOT for Uniswap v4, where assets live in the PoolManager singleton. And token/token pairs (e.g. UP/STONKBROKER) understate as "$0 real quote" — they are real but require a second hop. STONKBROKER (Ben holds $12,900) direct-to-WETH depth is $280,733 across two pools; the ~$2M of UP/YARD/MANCER/CLOCKIN-paired "liquidity" is routed, not phantom.

## WHAT THE MARKET IS PAYING FOR — regime analysis 27 Aug 2026. Artifact: https://claude.ai/code/artifact/b2061525-b5bd-401d-8791-80ba47d3d148
Method: 200 GT pools by volume + tx count -> 120 distinct tokens -> filtered >$1M v24 -> daily OHLCV 180d -> categories assigned by READING each landing page.
**Median 30-day return by category — the headline number:**
- Pays the holder automatically / equity exposure: **+1,925%** (n=3: NET, INDEX, Artificial Inu)
- Wears Robinhood's identity: **+378%** (n=5: JUGGERNAUT, STONKBROKER, GOOD, HOOD, CHUMP)
- Meme with a website: +341% (n=9)
- Infrastructure the holder must OPERATE: **+29%** (n=7: ARROW, PONS, HOOKR, DELTA, VIRTUAL, IF, INVEST)
- Tokenised equity itself: +13% (n=3) but the largest volume block on the chain
**Payout tokens returned 66x what infrastructure tokens returned over the same 30 days.** Both groups have real products, docs and teams. The difference is whether the token asks the holder to DO anything.
- Individual receipts: NET +3,481% (bonds+dividends), Artificial Inu +1,925%, JUGGERNAUT +530% ("Vlad's favourite meme, now onchain"), CASHCAT +341%, INDEX +249% (homepage literally reads "hold it, get paid in stocks"), PONS +232%, STONKBROKER +226%, GOOD +171% 7d, HMM +7,877%.
- **ARTIFICIAL INU 0x2E8c31162b855A2ffa90F6F8634643Ad6F111e18 (ticker AI, artificialinu.com) is the template**: a dog coin whose LP pair is tokenised NVDA. 42.7% of its $14.7M daily volume is quoted in NVDA, not ETH. "Every trade feeds the Vault, every fee burned or locked. The dog is long compute." FDV $54.7M, launched 14 Jul. Impossible on any other chain — no one else has real tokenised equities beside memecoins in the same AMM.
- **The gravity well: NVDA alone does $66.7M/day on a $7.2M FDV (9.2x turnover), SPY $16.2M, SPCX $13.3M — and they barely move (+13%/+3%/+18% over 30d).** More daily volume than CASHCAT, the chain's biggest memecoin. Every outperformer is a pipe between that boring money and the speculative money.
- **ARROW is the control group and it fails the thesis I used to recommend it.** Best-specified protocol on the chain, -91% from its July high, flat five weeks, in the +29% category. To benefit you must deposit collateral, mint aUSD, manage a ratio, watch a liquidation. To benefit from INDEX you hold it and stocks arrive. This chain's buyers came from a retail brokerage app; they did not arrive wanting to run a CDP. Second strike: it is a Liquity fork that could exist anywhere, and every large winner either could ONLY exist here or is ABOUT here. Trade not withdrawn ($1.25M bid at the base still stands) but it is explicitly a bet AGAINST the prevailing preference and should be the smaller of two ideas.
**DEFECT 9 — ATH PENALTY MISCALIBRATED FOR THIS CHAIN.** rhscore zeroed the timing block past -45% off ATH. Median drawdown among tokens UP >100% in 30 days is **-41%**, and 4 of 12 are worse than -45% (INDEX -62%, CLIPPY -67%, STONKBROKER -50%, JUGGERNAUT -49%). A deep drawdown is the normal resting state of an uptrend here. FIXED: bands are now >-25% / >-55% / >-80%.
**NEW SCORING BLOCK — market_fit, 18 pts** added to rhscore (max now 118). Reads the landing COPY (fetch_site now keeps 6,000 chars of body text) and scores: +8 holder-directed payout phrasing, +6 plumbed into stock tokens, +4 Robinhood identity, **-6 if the copy requires the holder to operate something**. First version was too loose and gave ARROW 18/18 because CDP docs also say "vault", "distribution", "backed by" — tightened so the payout phrase must be holder-directed and the action penalty always applies. Now: ARROW 4/18 (total 68, WATCHLIST), INDEX 8/18 (total 74, BUILD).
**The six-question screen this turns into:** (1) does holding it pay you automatically, no claim step? (2) is it plumbed into the stock tokens? (3) does it wear Robinhood's identity? (4) could it exist unchanged on Base/Solana — if yes, downweight. (5) does the holder have to operate anything — inverse signal. (6) is the payout mechanism visible on chain?
Next radar change argued for: a KEYWORD LAYER on landing copy of every queue candidate — dividend, paid in stocks, in kind, every epoch, vault, backed by, paired with, plus any tokenised-equity ticker. That would have surfaced INDEX and Artificial Inu in July before their moves.
CAUTION: this describes a 30-day regime on a two-month-old chain at one moment. Re-run monthly; when the payout trade gets crowded the edge moves to what the crowd left behind.