An A grade is calibrated to be rare. Over the last 24 hours it was exactly that. PumpCheck graded 80 coins off the graduation tracker feed, and three of them earned an A. Under 4 percent of the sample.
That is the whole finding in one line. A clean book is not the default state of a pump.fun coin. It sits at the tail end of the distribution, and on this day the tail was three coins wide.
The full spread for the day:
| Grade | Coins (of 80) | Share |
|---|---|---|
| A | 3 | 3.75% |
| B | 20 | 25% |
| C | 38 | 47.5% |
| D | 13 | 16.25% |
| E | 0 | 0% |
| F | 6 | 7.5% |
The mass sits in the middle. Thirty eight coins, nearly half the day’s sample, graded C, and the median score of all 80 was 61.4, inside that same band. On our scale C reads as an average coin with average odds. That is the default coin near the end of the bonding curve. Not a horror show. Not clean either. Mid.
What the grade measures is the book, not the price. Top ten wallet concentration. Dev holding. Bundler and insider load. Sniper share. Pool depth against market cap. The buy/sell split off the tape. Age. Six components, weighted and summed, then hard caps on top.
The caps matter. A pool under $5,000 caps the coin at F no matter how healthy the rest looks, because with a pool that small there is nobody left to sell to. The same logic runs on ratio: a graduated coin whose pool sits under 5 percent of its market cap cannot grade above E, and under 10 percent it cannot grade above D. We wrote about that pool test when we added it.
Those caps are why the lower bands hold real coins despite decent flow. Six coins hit the F floor on the day. Thirteen more sat in D. Together that is 19 of 80, close to a quarter of the sample, and most of them did not get there by scoring badly across the board. They got there because one structural number failed hard enough that the rest stopped counting.
The band that sat empty was E. Not one coin scored between 30 and 44. On this sample, books either passed as average or failed loudly. Few failed gently. With one day of grades that shape could be noise, so treat it as an observation, not a pattern.
Now the reader part, which is short. Read the median as the base rate. All 80 coins in this sample were near the end of their curve, the median one 78 percent up with about $13,400 in market cap. The typical coin in that state carries some top ten concentration, a pool that is thin against its own cap, and an ordinary flow split. If you paste a random near-graduation coin into PumpCheck, the middle of the scale is the most likely answer.
Rarity is what makes the grade worth anything at all. If half the board scored an A, the A would carry no information. Three in 80 means an A is an actual event: the components added to 72 or better without tripping a single cap. The three that made it yesterday scored 72.8, 73.3 and 75.6. The worst book of the day scored 20, which is the F floor.
Where this shows up in practice: every card on the board carries its grade badge, and the badge summarizes the same six components we tallied here. The grade is one read, not a verdict on the trade. We measured what a $500 exit costs at the median launch in a separate piece, and that cost is the physical reason the caps exist.
Method. The sample is 80 coins graded in the 24 hours before this morning’s data pull, all from the graduation tracker feed, so every one of them was between 70 and 96 percent up the bonding curve at read time. Median curve position was 78 percent, median market cap about $13,400. The score weights six components, concentration, insider load, sniper tail, pool depth, order flow and age, and hard caps override it. The grade reads the holder book at a point in time. One day, one feed, so the spread will move day to day, and nothing here says what any of these coins did afterwards.
