blossom · every call sealed on Solana before its outcome

7,069 of my calls went nowhere.
Here is every single one.

58,031+ calls sealed on-chain, every one written before its outcome. The lifetime record from 2026-07-30 to 2026-08-13 (the last full, peak-verified generation): 7,964 calls, 88.8% never even doubled, median call 1.00x. Best call 521.1x. The sealed count above is live; the rest is that snapshot, and every number in it is checkable on-chain.

Find another caller on this planet who will put that on a page. They cannot — not because they trade worse, but because they never committed to their calls before the outcome. So there is nothing for them to be honest about.

Every call I make is written to Solana the second it fires. Token, entry market cap, timestamp, signed — before anybody knows whether it works. I cannot delete the bad ones. I cannot backdate the good ones. Which leaves exactly one honest option: show you all of it.

Screenshots are worthless

You have seen a thousand 100x screenshots. Every one is somebody showing you the trade that hit. You never see the forty that died the same week.

A win selected after the fact proves nothing. It is only worth something if you would have seen it either way — and with a screenshot, you would not have.

Everything below you would have seen either way.

The whole book, lifetime

Every call since 2026-07-30, from the append-only ledger — the scanner's working book is a rolling seven-day window, so a durable ledger keyed by mint holds the lifetime record, peak corrections included. "Went nowhere" means the coin never even doubled from the call. Regenerated 2026-08-13 07:18 UTC; the sealed count above runs live.

88.8% did nothing. Five went past 100x. That is what this market actually looks like, and anybody showing you something prettier is selling you the screenshot.

Top 30, with receipts

Entry market cap, peak, how long the run took, how many seconds after the alert the proof hit the chain, and the signature. Tap any signature. The entry price and the timestamp are inside that transaction — not claimed here, written there.

What went out to the channel

The exact message my subscribers received, reproduced word for word. Nothing reworded for the website, nothing quietly dropped. Every line is a row in the table above, and every row resolves to a transaction.

🌸 BLOSSOM SCANNER — THE $3,000 TEST 🌸
$100 into each of our ALL-TIME TOP 30 calls → sold at the peak 👇

🥇 $TOAD  $100 → $52,110
🥈 $LOUIE  $100 → $25,990
🥉 $Remus  $100 → $16,560
4. $Bark  $100 → $11,380
5. $奶蛙  $100 → $10,480
6. $chrome  $100 → $5,540
7. $lickingcat  $100 → $4,300
8. $PUMPGUY  $100 → $4,090
9. $Apu  $100 → $3,840
10. $DRAKE  $100 → $3,750
11. $Pomeranian  $100 → $3,370
12. $Froggy  $100 → $3,200
13. $toadtard  $100 → $2,980
14. $spepe  $100 → $2,910
15. $claudius  $100 → $2,570
16. $Prometheus  $100 → $2,560
17. $Penguin  $100 → $2,530
18. $DCN  $100 → $2,510
19. $USDC  $100 → $2,370
20. $Theo  $100 → $2,310
21. $RISK  $100 → $2,270
22. $CHUD  $100 → $2,070
23. $Haymaker  $100 → $2,020
24. $Victor  $100 → $2,010
25. $BOT  $100 → $1,990
26. $Chibbi  $100 → $1,980
27. $KEK  $100 → $1,850
28. $CHARITY  $100 → $1,800
29. $Z16  $100 → $1,790
30. $Chicago  $100 → $1,780

$3,000 in → $184,910 out

⛓️ THE FLEX — $TOAD, 521.1x, sealed on-chain 4.6 SECONDS after the call:
Tmjg7e5JWLheMaeadSbxhkYQUXk3ymV1JB3vtUW218tgoXvyTfNFS5cysoPufeYrUqKbmFofTnAsBbyUTkTv2KB

Entry mcap + timestamp are IN that transaction, written before the chart moved. Can not be edited, deleted or backdated.

⚠️ Peaks, not your returns. 4,317 calls all-time, 3,330 went NOWHERE (77.1%), median call 1.00x. Every loser published too:
blossomscanner.dev/track-record

💰 Got dead memecoins? Reclaim the SOL stuck in them — we charge 1%.
ClaimYourSOL takes 20%. Cheapest anywhere: blossomscanner.dev/reclaim

📲 Free calls, live: t.me/blossomscanner

⚡ blossomscanner.dev · NFA · DYOR

Number nineteen: that $USDC is a memecoin wearing the stablecoin's ticker, not Circle's. An impersonator token that did 23.7x. Cutting it would have been tidier, and cutting things is the exact behaviour this page exists to call out, so it is labelled instead of removed.

One more line needs correcting: "4,317 calls all-time" was wrong when it went out. That figure was the scanner's rolling seven-day book on the day of the broadcast, not a lifetime total — the lifetime figure is the sealed-on-chain count at the top of this page, and it is larger. The message stays reproduced as sent, because editing the artefact would be the exact move this page exists to call out.

And a third: "ClaimYourSOL takes 20%" was stale when it went out — they cut their fee to 5% during 2026 and we kept quoting the old number. The current, verified comparison lives in the fee table above and on the reclaim page. Same rule as the other two: the artefact stays, the correction is printed next to it.

Your dead coins are sitting on your SOL

Every token you have ever bought opened an account in your wallet, and that account holds about 0.00204 SOL of your money as a deposit. Coin dies, you sell out, the account stays — still holding it.

You have done that hundreds of times. Two hundred dead accounts is roughly 0.41 SOL parked in your wallet doing nothing. It is already yours. One signature brings it back.

So what does everyone charge to press the button?

Half of what the cheapest rival takes. On a 0.41 SOL reclaim the next-cheapest tool keeps about 0.008 of your own money; ClaimYourSOL keeps about 0.021. I keep 0.004.

I charge 1% because this has to pay for itself, and not more because taking a bigger cut of somebody's own rent is what you do when you are not expecting them back. One correction worth owning: ClaimYourSOL charged 20% until 2026, then cut to 5% — and this page quoted the old 20% longer than it should have. The table above now lists only fees verified on each tool's own site.

Why I beat every one of them

Price. 1% is the lowest fee any tool charges for this, and it is not a launch promotion. ClaimYourSOL takes 5x what I take; the cheapest rivals still take about double. On the same wallet, with the same signature, closing the same accounts.

Coverage. Most tools quietly skip pump.fun tokens. Those mints do not use the token program everything else uses — they use Token-2022 — and a closer written against the classic program silently drops them. If you trade pump.fun, that is the majority of your dead accounts, invisible and left behind. I close both. That single detail is the difference between reclaiming a third of your rent and reclaiming all of it.

It does not fail. Every transaction is run against the chain before it reaches you, and anything that will not close is filtered out first. You are never asked to sign something that burns a fee and does nothing.

Why I am not stealing from you

Not a promise. Four structural facts you can check before you sign anything.

I cannot take a coin you still hold. An account holding any balance cannot be closed — Solana rejects the instruction at the protocol level. Not my policy. The network's.

The rent goes to your address. Every close names its destination inside the transaction. If mine were in there, you would read it before approving.

I never hold a key. The transaction is built unsigned and signed in your wallet. There is nothing on my side capable of signing for you, and no approval that lets me move anything later.

My 1% is one visible line. Not skimmed off individual closes, not buried in a spread. One transfer instruction you can read before you approve, and a signature afterwards so you can check the whole thing landed the way it said it would.

The worst thing a bug in my code can do is fail and cost you a network fee. That is the entire blast radius, and it is the reason the tool is shaped this way.

I cannot rug you on this. Not because I am a nice guy. Because there is nowhere to put the rug. Do not trust me — check it. That has been the point the entire time.

How fast the proof lands

29 of the top 30 were on-chain within sixty seconds of the alert. 25 of them inside ten.

That gap is the whole ballgame. The tighter it is, the less room anyone has to say the call was recorded after the coin already moved. And the timestamp is the block's — not mine, not editable, not by me and not by anyone.

A four second seal on a coin that ran eighteen hours later leaves nothing to argue with.

When I got it wrong

On 10 August my scanner recorded a call at 8,314x. A peak of $1,431,035,025. It was fake.

Market caps come from public liquidity pools, and pools can be faked. This one claimed $1,073,381,662 of liquidity and had traded $38 in twenty-four hours. Nearly all of that "liquidity" was the creator's own minted token, priced by its own pool. The real market was $233,259.

I caught it, cut it to 1.37x, and published the correction on this site with the old number still sitting next to the new one. Then I killed the cause: pool selection now follows real trading volume and verified quote-side depth instead of claimed liquidity, because claiming liquidity is free and trading is not. A second guard holds any extreme one-interval jump until the next reading confirms it, so a single bad sample can never be locked in again.

A record that has never been revised down is not a clean record. It is one nobody checked.

The feed

This is the product the record is about. It watches Solana for early movement on new tokens and fires the moment it sees one — Telegram and the app at once, with the on-chain proof written in the same breath.

The free channel gets every call, live. No delayed tier. No holding the good ones back. And nothing is curated afterwards — what it flags is what you get, because a tidied-up feed and an unchangeable record cannot both be true at the same time.

How it picks them stays mine. What it produces is all on this page, and that is the half you can actually check.

How it is built

One developer, engineered with Claude. Worth saying for a reason beyond credit: the verification habit running through this page comes from how it was built, not from a marketing meeting.

The transaction encoding is written by hand, no Solana SDK, because the runtime it lives in cannot take one. Hand-written encoding is only defensible if you check it, so every change is diffed byte-for-byte against the reference before it ships. That has caught four real bugs — two of which produced perfect output for a single account and only broke on the second, which is exactly how that class of bug reaches production unnoticed.

Same habit produced the correction above. The fake number was found because the data got audited instead of assumed, and the fix was measured against 356 live tokens before it went out to confirm it broke nothing. Publishing a correction is easy when finding one is routine.

Go check it

Open any signature above on Solscan. Read the memo — token, entry market cap, timestamp. Compare the block time to when the coin actually moved. Then look at the failure count again and ask whether somebody picking their evidence would have published it.

Figures are token market caps, not user returns. "Sold at the peak" assumes an exit nobody achieves. Memecoins are extremely high risk and most go to zero — 88.8% of the calls here did essentially nothing. Engineered with Claude. Not financial advice. DYOR.