tekmaker
← all projects

Bagyard

Paste a wallet, see your dead memecoin bags: peak vs now, and the SOL rent left in the dirt

Open the app → Read the research
Token
$BAGYARD, reserved name, not launched. Planned venue: pump.fun.
Problem
Active pump.fun traders hold dozens of coins that are 90%+ below their peak. On 2026-10-09 only ~3.7% of ~28,500 daily launches bonded, and none of the day's runners survived 2–7 days with real flow. Wallets hide these bags, and the tools that reclaim their rent want a wallet connection and keep a fee.
What the MVP does
Paste a wallet, a Solscan link or a ?w= link. In about 1–3 seconds Bagyard returns a title (Fresh Grave … Necropolis), the dead bags' value at each coin's peak vs now, a death rate, the SOL rent in empty or dead accounts, and a tombstone for every dead coin with a factual epitaph. You can download the card, post it, or share the link.
Token role
$BAGYARD would be the brand of a self-roast meme with a working app behind it: every card shared is an ad. Creator fees would pay for the data bill and a full-history indexer, and holding would unlock a read-only pro tier.
What works today
  • Today's cemetery: a live board of the last 24 h of pump.fun graduates (share already 90%+ below peak, dead $100K+ peaks, median time to peak, biggest fresh graves), scanned server-side every 10 min and refreshed in the page every minute
  • Graveyard verdicts per coin: dead, fossil, bleeding, alive, dust (worth under $1 even at the peak) and unmarked (no price)
  • Peak vs now: each remaining bag valued at the coin's pump.fun all-time-high market cap vs today
  • Rent in the dirt: SOL locked in empty token accounts and under dead bags, read from each account's actual lamports
  • Tombstone yard with coin art, −% from the top, peak → now, launch and last-trade dates, and an epitaph built from real numbers
  • Title tiers by number of dead bags, plus biggest ghost, deadest coin, oldest grave and the necromancer coin
  • Share card PNG (canvas), prefilled X post, ?w= deep links, read-only wallet connect, and a live real-trader example
  • Your yards: dug wallets saved in this browser only; re-digging shows new graves since the last dig
  • Rejects mint and token-account inputs with a clear message; read-only with nothing to sign
Shipped 2026-10-09 · consumer · factory conviction 8/10 · data: Solana RPC via /api/rpc (Helius), Helius DAS getAssetsByOwner, pump.fun coins-v2, Jupiter price API, GeckoTerminal (live example only), pump.fun coins list (graduates, worker scan every 10 min), Bagyard worker /api/today and /api/history (SQLite snapshots, 14 days)

TL;DR

Every active pump.fun trader ends up holding a pile of dead coins. About 28,500 coins launched on pump.fun in the 24 hours before our 2026-10-09 digest, and only ~3.7% bonded. The coins that did run mostly died within a day. Wallets hide that pile, and PnL terminals bury it in tables. Rent-reclaim tools find only the empty accounts, and most of them take a cut. Bagyard is a read-only, paste-a-wallet "graveyard" for the bags you still hold. In about 1–3 seconds it shows which coins are dead or fossilized, what each bag was worth at the coin's all-time high vs today, the biggest ghost, and how much SOL rent sits in empty or dead token accounts. You get a tombstone share card and a ?w= link. It's a self-roast format, like a "wrapped" recap but for losses, and it has a concrete utility hook: there is SOL to recover. $BAGYARD would be the brand of the meme. Creator fees would pay for the build, and holding would unlock a read-only pro tier (planned; no token exists).

The user and the pain

User: a pump.fun or PumpSwap trader who has been in the trenches for weeks or months. They hold 20 to 900 token accounts, most of them coins they rode down and never sold because the bag was too small to bother with.

Pains:

  1. Out of sight, never counted. Phantom and other wallets show balances and hide spam. Nobody shows "you hold 37 coins that are 99% below their top, and here is what they were worth at the peak". Traders feel the loss but never see it in one place. That is also why the format is funny and shareable.
  2. SOL stuck in the dirt. Every token account locks a rent deposit. Empty accounts and dead-dust accounts can be closed to get it back. Most people don't know how much is there, and the tools that close accounts want a wallet connection and keep a fee.
  3. No honest recap. PnL terminals (GMGN, Axiom, Cielo, Birdeye) are built for copy-trading research. Their numbers depend on cost basis, transfers and airdrops. A "peak vs now of what you still hold" view needs no cost basis, so the chain alone can answer it, and it reads as a story rather than a spreadsheet.

Evidence

  • Almost everything dies. In the 24h before 2026-10-09 07:30 UTC, ~28,510 coins launched on pump.fun and 1,050 bonded, a 3.7% bond rate. 305 of those 1,050 were instant self-bonds: ATH within 2 minutes of creation, then down 50%+ (memeradar digest, our own collector).
  • Runners die fast. Of the 12 organic volume leaders that day, the trajectory pass classified 3 as "dead / −70%+ from 24h high" (TikTok −83%, Gary −70%, Novita −82%) and 3 as "bleeding" (OWLNIGHT −46%, TWEETCRAFT −59%, HeeHaw −67%). It found zero "survivors" 2–7 days old with real two-sided flow (memeradar deep pass, 2026-10-09).
  • Industry graduation stats agree: 0.84% of tokens graduated in April 2026 (6,379 of 763,350), per PumpFunData (2026-07-16). Scorp Trader measured 1.71% (11,877 of 693,331) for 2026-06-23 to 07-19 (2026-07-26). Solana Compass, citing The Block and @adam_tehc's Dune dashboard, reported ~42,000 launches in one day, under 2% of tokens ever graduating, and 11.9M total launches (2026-06-10).
  • Dead coins pile up across crypto: CoinGecko counts 53.2% of tokens tracked on GeckoTerminal as dead, with 11.56M dying in 2025 alone (updated 2026-04-17).
  • Most traders lose: 62.5% of 4.2M+ pump.fun wallets lost money over six months, per crypto.news citing Dune query 5233371 (2025-06-05). Only ~0.41% of wallets had realized more than $10K profit, per Cointelegraph citing @adam_tehc's dashboard (2025-01-11). These are older figures; we found no fresh 2026 trader PnL stat.
  • Rent reclaim is a real, used behaviour. Sol Incinerator had returned more than 470,000 SOL by November 2025, per Four Pillars (2026-01-12; the article cites no upstream source). Jupiter shipped a portfolio "rent reclaim" (launch post 2026-02-26). Solana's SIMD-0437 rent cut went live in stages from September 2026: step 2 hit mainnet 2026-09-11, and Solana Compass estimates ~612K SOL is now reclaimable from overfunded accounts (2026-09-12). So "how much SOL is in my dirt" is a timely question.
  • Our own wallets, 2026-10-09: a random PumpSwap trader (HzBh…QcKp) held 30 coins in 84 token accounts. 53 accounts were empty, and 8 coins were dead (3 fossils), worth ~$2.4K at their peaks and ~$15 now. A heavy bot-like trader (6bCT…4g9m) held 719 coins in 928 accounts, with 0.44–0.52 SOL of rent sitting in empty or dead accounts (varies run to run with how many pump.fun lookups land).
  • Unverified signals: traders sharing PnL and loss screenshots is a known pattern (Axiom's "Share PnL" card, viral loss screenshots). We could not fetch the specific posts, so we treat this as UNVERIFIED and don't rely on it.

Market now

From the memeradar digest and deep pass of 2026-10-09 07:30 UTC (24h window), plus our own tests the same day:

  • SOL $110.42. Launches ~28,510/day. Graduations 1,050 (43.8/h), a bond rate of ≈3.7%. Instant self-bonds: 305 of 1,050.
  • Graduates by ATH: 544 reached $100K+, 109 reached $1M+, 33 reached $10M+. Only 51 organic graduates were still at 50% or more of their ATH. Median organic ATH: $192.4K.
  • Clone clusters print and fade: "fomo" bonded ×29, "Google Gemini" ×15, "Claude AI" ×12, "MrBeast" ×11 and "Tiktok Coin" ×9 in one window. FADING names, with nothing bonded in the last 6h: Meta, NVIDIA, Nike, X Coin, Youtube Coin and Instagram Coin. Every clone that fades becomes a dead bag in someone's wallet.
  • Words lifting on graduates included "app" (lift 10.3), "trading" (3.2), "research" (4.1) and "games" (10.3). Utility- and app-flavoured names are bonding, which suits a meme with a working app behind it.
  • Survivors (2–7 days old with real two-sided flow): none.
  • Our own test, 2026-10-09: the HzBh…QcKp wallet's 14 pump.fun-history coins break down as 8 dead or fossil (57%), 1 dust, 0 bleeding and 5 alive or near the peak.

Competitors and the gap

All checked 2026-10-09.

  • Sol Incinerator (sol-incinerator.com): Closes empty accounts and burns tokens/NFTs; its blog targets "rugged meme coins". Price/fee: ~0.00004 SOL per standard account / ~0.00007 per Token-2022 account, from the reclaimed rent; its blog calls this a "2% base recovery fee" (post dated 2026-07-02). Gap vs Bagyard: Needs a wallet connection; no peak-vs-now values, no dead/alive verdicts, no share card.
  • RefundYourSOL (refundyoursol.com): Closes accounts and burns dust; ranks/XP, referrals. Price/fee: "starting at 15%, matchable down to 2%" of recovered SOL. Gap vs Bagyard: Wallet connection, high fee, no recap.
  • Solana Tracker Refund (solanatracker.io/refund-sol): Previews reclaimable rent for any address. Price/fee: 1% of recovered rent. Gap vs Bagyard: Closest on paste-and-preview, but no per-coin peak values, verdicts or card.
  • Jupiter portfolio rent reclaim: One-click close of empty accounts inside Jupiter. Price/fee: not stated (UNVERIFIED). Gap vs Bagyard: Wallet-bound; no graveyard framing; free built-in threat to the rent half.
  • solgraveyard.com: "Bury or burn your dead, rugged, or worthless meme coins" with gravestones and AI obituaries (from a search snippet). Price/fee: UNVERIFIED (site returned 502). Gap vs Bagyard: Closest theme; appears to need on-chain burning. Peak-vs-now unverified.
  • GMGN / Axiom / Cielo / Birdeye: Wallet PnL, holdings, tracking. Price/fee: free tiers; Axiom ~1% per trade, Cielo paid tiers (both UNVERIFIED). Gap vs Bagyard: Built for trading research; cost-basis PnL; no dead-bag recap.
  • Step Finance: Used to have "Wallet Cleanup" + PnL. Price/fee: —. Gap vs Bagyard: Shut down, announced 2026-02-23 after a breach on 2026-01-31.

The gap: none of these tools takes a pasted wallet with no connection, values each remaining bag at its coin's all-time high vs now, sorts the coins into dead / fossil / bleeding / alive, and turns the result into a screenshot-worthy card. Bagyard is free, read-only and never signs. It doesn't compete on closing accounts: it shows what's there and leaves the closing to your wallet or a tool of your choice.

The product

What it does: paste a wallet, a Solscan link or a ?w= link, connect a wallet read-only to fill in your own address, or hit "Try a real trader's wallet". About 1–3 seconds later you get:

  • A title by number of dead bags: Clean Soil, Fresh Grave, Gravedigger, Undertaker, Cemetery Owner, Necropolis.
  • Peak → now of the dead bags, the number buried (and fossils), the death rate of the pump.fun coins held, and the SOL rent in empty and dead accounts.
  • Facts: biggest ghost, deadest coin, oldest grave, and the "necromancer" coin still within 25% of its top.
  • The yard: a tombstone per dead bag with the coin image, −% from the top, peak → now, an epitaph built only from real numbers ("Peaked before you blinked." for instant self-bonds, "Never left the curve.", "No trades for 41 days."), and launch → last-trade dates. Each stone links to the coin.
  • On life support (50–90% down), Still breathing (alive or regular priced tokens), unmarked graves (no price anywhere) and dust graves (dead bags worth under $1 even at the top, usually airdrops).
  • A share card (1200×630 canvas PNG, via the native share sheet where supported), a prefilled X post (never auto-posted), and a copyable ?w= link.

How it works (engine.js, testable from Node):

  1. getAccountInfo rejects mint and token-account inputs. Helius DAS getAssetsByOwner (showFungible) returns the wallet's fungible holdings with balances and Jupiter-sourced prices. getTokenAccountsByOwner for SPL and Token-2022 finds empty accounts and each account's actual lamports, so the SIMD-0437 rent change is handled.
  2. For every mint with the pump.fun "pump" suffix (max 200 per dig, 10 concurrent, retried), pump.fun coins-v2/{mint} gives ath_market_cap, usd_market_cap, created_timestamp, last_trade_timestamp, ath_market_cap_timestamp and complete.
  3. Current market cap = DAS price × supply when Helius has a price, otherwise pump.fun's own. Peak value = (balance / supply) × ATH market cap. Drawdown = 1 − now / ATH. Broken ATHs (seen: 1.5e21) are discarded.
  4. Verdicts: dead = 90%+ down, or 70%+ down and idle 7+ days. Fossil = dead, 30+ days old and idle. Bleeding = 50–90% down. Alive = under 50% down. Dust = dead with a peak value under $1. Unmarked = no price.
  5. Non-pump tokens with no DAS price are priced through Jupiter (up to 150). Stablecoins are left out.

Backend (server/worker.mjs): every 10 minutes the worker reads every coin that graduated from the pump.fun curve in the last 24 h (≈17 pages of pump.fun's coins list via ctx.px, ~100 px calls/hour of the 2,000 budget) and stores a snapshot in SQLite: graduates, share now 90%+ below peak, $100K+ peaks that died, median minutes from launch to peak, the biggest fresh graves and the survivors. GET /api/today serves the latest snapshot to the "Today's cemetery" board on the first screen (refreshed every minute without a reload); GET /api/history?h= serves the death-rate series (14 days kept). It stores public coin data only: no wallets, no IPs. Coins with caps over $10M are excluded because pump.fun reported $10M–$450M caps on a family of day-old junk launches (SARP, USDP, VSOF…) on 2026-10-09. Calibration on 2026-10-09 ~13:00 UTC: 1,013 graduates in 24 h (37 excluded as broken caps), 89.4% already 90%+ below peak, 444 of 502 $100K+ peaks dead, median peak 6 minutes after launch; the biggest graves were brand clones (Instagram, Nike, Meta, Google, TikTok, FOMO).

Your yards: wallets you dig are kept in localStorage only (max 12). Re-digging one shows the new graves since the last dig, so the loop is "check your yard after a bad week".

Data sources: Solana RPC + Helius DAS via /api/rpc; pump.fun coins-v2 and Jupiter price via /api/px; GeckoTerminal PumpSwap pools and trades (live example only, 300 s cache); pump.fun's image CDN for coin art.

Calibration (2026-10-09, live mainnet):

  • HzBh…QcKp (random PumpSwap trader): 30 coins, 84 accounts, 53 empty → Gravedigger. 8 dead (3 fossils): DONO −99.6% ($1.2K at peak → $5), PRINTR −99.1% ($964 → $9), MOONDOGECOIN −99.98%. Rent 0.123 SOL. On the first run it held WORLDCUP at its ATH ($37K), correctly shown as alive / necromancer. It had sold WORLDCUP by the next run, and the yard updated.
  • 6bCT…4g9m (heavy bot-like trader): 719 coins, 928 accounts. Before the dust rule it showed 205 "dead" coins, mostly worth cents at their peak, which overstated the yard. After adding dust (<$1 at peak): 11–14 dead, ~160–200 dust, death rate 84–85%, 0.44–0.52 SOL in empty or dead accounts. The broken ATH on ACT (1.5e21) was caught and discarded.
  • 5ZRU…r7Xg and 7oeU…rq5y (wallets in the WHUF farm cluster): 14–19 coins, mostly unpriced airdrops. Their dead coins had sub-$1 peaks, so they read correctly as Clean Soil with "just N dust graves" instead of fake losses.
  • 3hqv…Gcf6: 2 coins, 57 empty accounts → Clean Soil, with Agency 83% down on life support and 0.109 SOL of rent in empty accounts. That's the "rent hook" case.
  • EPjF…Dt1v (USDC mint pasted as a wallet): it used to "dig" junk sent to the mint address; it now returns a clear "That's a token (mint) address" error.
  • Speed: 0.5–3 s for normal wallets; 3–25 s for 700+ coin wallets, depending on how warm the pump.fun cache is.

Limits:

  • The first 1,000 assets are read, and up to 200 pump.fun lookups run per dig.
  • Spam airdrops of real pump coins count as bags, because the chain doesn't say who bought what.
  • "Peak value" is not cost basis or realized loss.
  • Pump coins without the "pump" suffix aren't looked up.
  • The live example prefers a fresh trader from today's top PumpSwap pools, verified to have graves. When none qualifies (top pools are bot-heavy), it falls back to one of two real traders seen today, still dug live.

Token design

Planned, not launched. No token exists unless it's listed at tekmaker.xyz/p/bagyard.

  • Role: $BAGYARD is the brand of a self-roast meme with a working app behind it. The share card is the marketing: every "I'm a Cemetery Owner" post points back to the app.
  • Creator fees fund the build, stated plainly: RPC/DAS credits, and an indexer for full transaction history (true cost basis, sold-at-the-bottom detection).
  • Holder tier (read-only balance check, no signatures): unlimited heavy-wallet digs (beyond 1,000 assets / 200 lookups), multi-wallet yards, a monthly "obituary" recap, and custom card themes.
  • Community leaderboard: an opt-in "biggest yards" board where wallets submit their own dig. No doxxing; only wallets that opt in are listed.
  • We dig ourselves: the project page would show Bagyard's own dig of the dev wallet. No promises about price, yield or returns.

Launch plan

  1. Ship the MVP as a free tool. Seed it by replying to "rekt" and loss-porn posts with the poster's own Bagyard card, only for wallets they've posted publicly themselves.
  2. Daily "yard of the day" post: a pseudonymous card of a public trader wallet that opted in, or one of our own wallets.
  3. Lean on rent-cut timing (SIMD-0437 steps 3–5, expected November 2026): "check how much SOL is in your dirt".
  4. Launch $BAGYARD on pump.fun only after the tool has organic usage, with the creator-fee use stated on the project page.

Roadmap

  • Full history mode: every coin ever bought, sold-at-the-bottom vs held-to-zero, "paper hands hall of fame".
  • Optional, fully previewed account-close transaction built client-side: close empty accounts only, each instruction listed, signed in the user's wallet, no fee to us and no approvals or delegations.
  • Multi-wallet yards and a monthly obituary email/X card.
  • A creator's view: every coin a dev launched that is now in the yard.

Risks and kill criteria

  • Risks: pump.fun API changes or rate limits (mitigated by caching and retries). Helius DAS price gaps. Spam airdrops inflating yards (mitigated by the dust rule). Jupiter or wallets shipping an equivalent recap. The roast tone could read as mocking losses; the copy keeps it self-deprecating and never names other people's wallets.
  • Kill criteria: under 300 digs/week four weeks after launch, or a share-card click-through under 2%, means the format doesn't travel, so stop. Also stop if the pump.fun data source becomes unreliable and no replacement for ATH data exists.

Sources