Every launchpad token starts on a bonding curve, and almost all of them die there. The curve is a simple pricing mechanism. Buyers push the price up along a fixed formula, and once enough SOL has come in, the token graduates. It migrates into a real DEX pool with actual liquidity, and from that point it trades like anything else. The gap between a token sitting at forty percent of its curve forever and one that fills and migrates is where the whole game lives, and most of what you need to judge it is on-chain if you know where to look.
I am not going to pretend I can pick winners from this. What I can do is throw out the obvious fakes before I waste attention on them, and that filter alone is worth more than any single indicator. The tokens that graduate on manufactured flow tend to look identical to the ones that graduate organically right up until the curve fills, and then the organic one keeps a floor and the manufactured one round-trips to zero in an afternoon.
The three numbers that actually matter
Bonding-curve progress is the headline, and it is the least useful of the three on its own. Progress just tells you how much SOL has been deposited relative to the graduation threshold. A token at eighty percent looks close. But progress says nothing about who put that SOL in, and a single wallet can push a curve from zero to graduation by itself if it wants to burn the capital. Watch the rate of change more than the level. Curves that climb in smooth, continuous steps over hours usually have real buyers behind them. Curves that jump in big discrete chunks and then flatline are one or two actors, not a crowd.
Unique buyer count is the number I lean on hardest. Pull the list of distinct wallets that hold or have held the token and look at the distribution, not just the total. A hundred holders where the top ten wallets account for most of the supply is a different animal from a hundred holders spread evenly. What you are trying to catch is the wash pattern. One funder splits SOL across twenty wallets, each buys a slice, and now the token shows twenty unique buyers that are really one checkbook.
Migration velocity is the third leg. This is how fast the curve is filling and, once it graduates, how quickly and deeply liquidity actually shows up in the new pool. A token that fills its curve in minutes on a handful of wallets and then has thin liquidity post-migration is a trap that completed successfully. A token that takes longer to fill but pulls in fresh wallets the whole way, then lands with liquidity that holds through the first wave of sells, is the rarer thing worth sizing into.
Telling one funder from a real crowd
The tell for manufactured flow is almost always in the funding graph. Genuinely independent buyers arrive from different sources. Some fund from centralized-exchange withdrawals, some from other wallets they already held, some bridge in. When you trace the funding history of the top buyers and they all trace back to the same parent wallet within a few hops, that is your answer. It does not matter how clean the chart looks. Twenty wallets funded from one address in the same block window are one participant wearing costumes.
A few specific things I check before I treat buyer count as real:
- Funding source overlap. Trace the top buyers back two or three hops. Shared parent wallets are a hard no.
- Timing clustering. Buys landing in tight bursts, same block or adjacent blocks, at similar sizes, read as scripted rather than organic.
- Buy-size uniformity. Real crowds buy messy amounts. A ladder of near-identical buys is a bot loop.
- Holder behavior after buying. Wallets that buy and immediately sit dead, never adding, never trimming, are placeholders, not conviction.
- Sell-side presence. A curve with buys and almost no sells is suspicious. Real interest generates two-way flow.
None of these is decisive alone. Two or three of them stacking on the same token is enough for me to walk away. The failure mode I have burned myself on more than once is treating a high unique-buyer count as a green light without checking where the wallets came from. The number was real. The independence was not.
Setting thresholds before you look at the chart
The mistake I keep watching people make, myself included, is deciding the thresholds after they already want the trade. You see a token pumping, you feel the fear of missing it, and suddenly forty organic-looking buyers is plenty and the funding overlap is something you will check later. Later never comes. So write the numbers down cold, when you have no position and no opinion.
Mine are rough rules of thumb, not laws. I want to see unique buyers comfortably into the dozens with no single wallet dominating supply, funding sources that do not collapse to one parent, curve progress that climbed steadily rather than in two whale-sized jumps, and post-migration liquidity that survives the first real sell pressure before I add. If a token misses more than one of those, size stays tiny or zero. The point of a threshold is that it holds when you least want it to, so decide it while you are bored.
The tooling here is a grind if you do it wallet by wallet, so most of what makes this practical is having the flow already parsed. Whale movements, unique-buyer distributions, and funding traces are the kind of thing our platform at Blockcircle exists to surface, but the discipline matters more than the dashboard. You can do this manually with a block explorer and patience, and honestly doing it by hand a few times is how you learn to feel when a launch is one person talking to themselves.
Most of these tokens are going to die no matter how carefully you read them. That is the base rate and no amount of on-chain forensics changes it. What the flow lets you do is skip the ones that were never real to begin with, and keep your size for the small number where the buyers are actually a crowd and the liquidity actually holds. That is a narrower job than picking winners, and it is the one worth being good at.