---
name: rbthreek-wallet
description: The hunt for rb3k (@rbthreek)'s Robinhood Chain trading wallet, the live PDEX timing test that killed two earlier candidates, and the state of the RBT3K alert bot. Continues the rb3k section of [[copytrade-targets]], which hit its size cap.
sources: [cowork]
aliases: [rb3k, rbthreek, RBT3K, sailboat]
---

# rb3k wallet hunt, continued 30 Aug 2026

Earlier passes are in [[copytrade-targets]]. Bot state is in [[alert-bot-fleet]].

## The live PDEX test, and what it killed
- [stated] Ben's instinct, unprompted: "I have a feeling that the wallets we have for him are not actually his wallets." He was RIGHT, and this is the second time in two days his push-back beat my screen
- On 29 Aug 12:26 UTC rb3k revealed **$PDEX, pTokens Index, 0x5C9A6a0353f1EDC67B22f2cdcfE9294f5A889B60** on Robinhood Chain ("Airdrops 3x long @arcus_xyz tokens for holding, first of its kind I believe"), quoting his own "might have found another gem on rh" teaser. Pool created 28 Aug 16:10 UTC
- 1,595 non-router wallets bought PDEX from a pool BEFORE that tweet. **Neither of the two wallets I had enabled in the RBT3K bot is among them.** 0xcd211569a108fdac74d729ba7358a743a3c2e08d and 0x0a47f7e62e39f28126c213addf99e6277901c3fd are both eliminated and are now disabled in the bot config with that reason recorded
- Also eliminated earlier the same day: 0x2cffed5d56eb6a17662756ca0fdf350e732c9818 and 0x0e331a293bb22e93e2e5e417d84d755ae185da02, which have never spent WETH on this chain and whose books are HOOD10 basket distributions

## The methodological error, stated plainly so it is not repeated
- **Current holdings are the wrong instrument for a trader who rotates.** Requiring a wallet to hold all seven of his 22 Aug names TODAY returns only 8 of the 1,595 PDEX pre-tweet buyers, and every one has a $0 to $821 book. The names he listed on 22 Aug have been rotated out; on 28 Aug he wrote he is "going to sit on core bags and flip shitters privately"
- The right test is EVER BOUGHT FROM A POOL, per name, with the buy timestamped against his post about that name

## Current leading candidate: 0xf154977eb88584660cb43467516126716938434a
- Plain EOA (eth_getCode returns 0x), RH nonce 12,309, 1.40 ETH on RH chain
- Bought PDEX **29 Aug 05:11 UTC**, 7h15m before his reveal and 13h AFTER launch, so it is a discretionary find, not a launch snipe. The 16:10-16:19 cluster on launch day is snipers and should be discarded
- Bought all seven of his 22 Aug stated names from pools, every one ahead of his posts: CASHCAT 4 Jul, TENDIES 8 Jul, PONS 14 Jul, AI 22 Jul, NET 22 Jul (23 days before his 14 Aug call), SB 23 Jul, DELTA 7 Aug (7 days before)
- 565 tokens received, 227 both received AND sent. The skill's separator: machines run 900-6,250 tokens both ways, humans 129-490. This is human range
- Cross-chain: Ethereum 414 txs, Base 2,273, Arbitrum 4, HyperEVM 472 with 0.4275 HYPE, matching his "one of my biggest pnl holds is on HyperEVM"
- **AGAINST IT, and why this is a candidate not an identification:** it holds none of his publicly named non-RH positions at this address (gSPEED, ALTT, AERO on Base, FWA on Ethereum all zero), and it never touched TTWO which he posted about on 28 Aug. Either he keeps a separate wallet per chain, or this is not him

## What changed strategically, and it matters more than the wallet
- [stated, from his 28 Aug post] "This will likely be my last ticker shill on the feed for a while going to sit on core bags and flip shitters privately (can see my scans on @Altcoinist)". **The public reveal stream that every timing test depends on is ending.** Future disambiguation tests will be rare, and his actual calls move behind the Altcoinist private beta, which gates on holding 100k $ALTT
- That makes the wallet worth MORE, not less, because the wallet becomes the only channel. It also means the next confirmation opportunity may not come for weeks

## Reusable, on top of the onchain-forensics skill
- A KOL reveal is a natural experiment with an exact timestamp. Get the token, get every pool-sourced buy before the tweet, and intersect with prior reveals. Two reveals with a shared buyer is close to proof; one reveal narrows 1,595 to about 100 when intersected with an older dated call
- X post timestamps: the `datetime` attribute on the `time` element belongs to the QUOTED tweet when a quote is present. Read the rendered "7:26 PM · Aug 29, 2026" footer instead and convert from the browser's timezone (Ben's Mac is Asia/Bangkok, UTC+7)
- Block-number-from-timestamp on RH chain must be a binary search on block headers. Block production went from ~10k/day in June to ~870k/day in late August, so any linear rate estimate is wrong by days