Most tokens die on the bonding curve. I study the 1% that graduate and the 4% that run. Tick-by-tick Solana pAMM research. Structured datasets in development.
We ran 987 live trades on pumpfun graduations.
One number separates winners from losers at tick 10:
Winners: avg sell SOL per transaction = 0.127 Losers: avg sell SOL per transaction = 0.673
5× differential. Every session. Consistent.
Your bot isn't measuring this. Ours is.
Free sample dataset at https://t.co/4WQrZvNNqA — no email, no form, just download it.
SOL pressure fingerprints. Phase sequences. Pre-entry trajectory. Session regime. Confirmed outcomes.
First dataset characterising PumpSwap microstructure at SOL-weighted resolution.
https://t.co/uL0xN8fvWr — first datasets available
@Shyft_to
🧵 Most Solana traders using graduation data are measuring the wrong thing.
Transaction count buyRatio looks clean. But one 5 SOL sell among ten 0.1 SOL buys gives you a 90% buy ratio — while SOL is leaving the pool.
We built the sensor that sees through this. 🧵
Asia session produces nearly 3× more genuine rockets than US session on equal trade count.
Now labelled on every record.
If you're building a T22 strategy and not accounting for session regime — you're trading blind half the time.
5,947 tokens in the miss file.
every graduation that didn't fire a signal. labelled with the exact threshold it failed and its confirmed peak multiplier.
the biggest mover in that file: 297,661×.
it was labelled as a failed signal.
that file is worth more than the signals file. the negative space always is.
at exactly tick 10 after entry into a T22 PumpSwap position:
rug trades: avg sell SOL per transaction = 0.673 genuine momentum: avg sell SOL per transaction = 0.127
same transaction count ratios. 5× different in SOL size.
the bundler is invisible to every standard signal in this space.
it's not invisible to SOL-weighted pressure.
if you're building T22 entry or exit models and you want to see what SOL-weighted microstructure data looks like before it goes to market:
[email protected]
first access. exclusive window available.
every quant building a T22 graduation strategy right now is working from the same inputs.
transaction count. buy/sell ratio. volume. wallet count.
the signal everyone uses. the edge nobody has.
here's what's missing from every model in this space. 🧵
this is the difference between a model trained on transaction data and a model trained on SOL pressure data.
one sees noise. one sees signal.
the dataset that shows this — 8,556 graduation events, every tick, SOL-weighted, phase-labelled, outcome-confirmed — is not publicly available.
it's in development for commercial licence.
29,896,733 price ticks captured since May 1.
every pumpfun graduation. every trade in the first 2 hours. every bundler dump. every organic recovery. every token that looked dead and wasn't.
3.7GB of raw market microstructure that doesn't exist anywhere else.
this is what a data pipeline looks like before it goes to market.
@MessariCrypto@solana@AvgJoesCrypto your Q1 pumpfun revenue numbers are interesting in context of the graduation microstructure — organic arrival timing on T22 tokens is at median 358 seconds, 94% after the standard monitoring window.
I built a system to track that moment in real time.
What I found in the data wasn't what I expected.
The first finding — and it's the one that changed how I think about this market entirely — is next.
Most tokens die on the bonding curve.
150-300 make it to a real AMM every single day.
Every one of them arrives with $10,000-15,000 of real liquidity, automatically seeded.
Most people think graduation is the opportunity.
I've been tracking every graduate in real time. What I found changed everything. 🧵
But approximately 4% absorb it.
They survive the opening dump, attract genuine buyers, and run.
Not a little. +2,000%. +10,000%. +100,000%+.
The question isn't which tokens graduate.
It's which 4% survive what comes next.