The line is how big the trenches were on each day. Dotted marks are celebrity coin launches. Each label is the reading from the day before launch. That is what the grifter saw when they pulled the trigger.
Each input is how big the room is against its own trailing year. A reading of +1 means about one standard deviation bigger than usual. The composite is the plain average of the two. Direction is shown next to the reading and is not part of it.
How big pump.fun bonding-curve volume is right now against its own last year on a 7-day average. This is the room on the chain where nine of the ten celebrity coins on record launched. LAPTOP was the first on Base.
Solana only for now. The other three chains are counted in the fee input to the right. They don’t enter this one because their launchpads publish fees and not volume. PONS on Robinhood Chain also has under three months of history against a scoring window that wants a year. As volume series appear and the venues age they go in here too.
How much the launchpads themselves are earning right now against their own last year. Seventeen venues on Solana, Robinhood Chain, BNB and Base summed. This is the room everywhere else. It is the input that sees a crowd that moved chains.
Robinhood Chain is half of this input on under three months of history. Treat its share as young.
Reading is from the close of the day before launch. Under it is whether the room was filling, holding or draining at the time. Trench volume 20 days later is the 7-day average versus launch day. Token outcomes are approximate and come from public reporting. Every one of them went down.
| Launch | Coin | Gauge | Trench vol, 20d later |
|---|
Findings as of the September 2026 backtest. The gauge above refreshes live. These numbers don’t.
How big the trenches are today against the last year on Solana and across the other three chains. Size is the whole gauge. On the tenth of days it reads highest, the following three days carried 3.6 times the trench volume of a median day and 1.8 times the launchpad fees. That held in both years of data and in every decile. Rank correlation between the reading and money in the room over the next three days is 0.84.
Seven of ten celebrity coins launched with the room at or above the 80th percentile the day before. Median 83rd. A random ten days does that about half a percent of the time. LAPTOP launched on Sept 9 2026 into a reading of +2.21, the 96th percentile, the second-highest launch reading on record, and was down 98% within 90 minutes. The two exceptions are YZY and NYC. Both launched into ordinary rooms that were filling at the fastest pace in the sample. Grifters see a crowd. The gauge measures the crowd.
Whether the room empties right after your launch. A full room does not drain in three days. The correlation between today's reading and the room holding through the launch window is zero. Over 20 days it says almost nothing. Over 40 it does. After any reading above +1 the room was a median 0.7 smaller 40 days later and it was smaller two times in three. After a celebrity launch into a room that size it was 1.1 smaller. Solana volume fell 55 to 70% in the 40 days after the Trump family. Five of those six launches are the same January 2025 drain measured five times. Hold it as a template. It is a posture for the month after a launch and not a three-day signal.
Whether growth matters. Every version of this gauge that included how fast the room was filling did worse at the question than size alone. Twenty-day volume growth on its own correlates 0.22 with money in the next three days. Size correlates 0.84. Averaging them together pulled the gauge down to 0.60. Growth is an annotation here and not an input. LIBRA and CAR launched into rooms at the 80th percentile that were draining fast. They still extracted. Milei brought his own crowd. The reading tells you who the exit liquidity is. It does not tell you how much of it there is. Every coin on this list went down 70 to 95% regardless.
How it's built. Two inputs. Solana trenches is the log of pump.fun bonding-curve volume on a 7-day average. Launchpad fees is the log of the summed daily fees of seventeen named launchpad venues across Solana, Robinhood Chain, BNB and Base on a 7-day average. On Solana they are pump.fun, bonk.fun, LaunchLab, Meteora DBC, Bags and Believe. On Robinhood Chain they are PONS v1 and v2, noxa.fun and Stonkbrokers. On BNB they are four.meme, flap.sh and Binance Alpha. On Base they are clanker, Zora Coins, Flaunch and o1. Those seventeen carry 93% of everything DefiLlama classifies as launchpad fees on the four chains. Each input is converted to a z-score against its own trailing 365 days with a 90-day minimum. The reading is the equal-weighted average of the two. Direction is the change in the reading over 20 days. Filling above +0.4, draining below −0.4, holding between.
Why size and not growth. The first versions of this gauge copied the structure of Deutsche Bank's pressure index. That is the index the TACO trade gets read off. It averages equal-weighted 20-day changes in each input. It measures pressure, which is a change. A grifter deciding whether to launch needs a level. Is the money there. So every candidate input was scored against the question that person asks, which is how much money will be in the room over the next three days. That was run on 700 days with a year-one and year-two split. Size of the room correlated 0.84 with the answer and 0.62 out of sample. Twenty-day volume growth scored 0.22. Fee growth scored 0.11. A 5-day acceleration term scored 0.17. Every composite that mixed size with growth scored lower than size alone. The growth inputs are mostly noise on that question and they dilute the one input that isn't. Growth stays on the page as an annotation because it says something size doesn't, which is whether the room is filling or draining. YZY and NYC both launched into rooms that read ordinary on size and extreme on growth. It stays out of the number because putting it in made the number worse.
Zones. Above +1.5 is extraction season. +0.7 to +1.5 is frothy. −0.8 to +0.7 is lukewarm. −1.3 to −0.8 is quiet. Below −1.3 is grim. The cutoffs are round numbers near the historical 88th, 75th, 27th and 10th percentiles. They are not fitted to anything.
What it's tested against. Thirteen celebrity coin launches from May 2024 to September 2026. Three fall before the index has enough history to score. For the ten that don't, the reading is taken from the day before launch so the launch's own volume can't contaminate it. Timing is scored as the percentile of that reading against all days. The p-value comes from a bootstrap of ten random days. The all-days test ranks every day by reading and buckets them into deciles. It then measures the money in the room over the following three days in each bucket, separately for each year. Both tests are in the notes along with every composition that was tried and rejected on the way here.
What it isn't. A prediction. A top call. Financial advice. It's a gauge. It reads the room. It does not tell you when the room empties in the three days that matter to a launch. It does not tell you the coin will hold. LAPTOP launched into the second-highest reading on record and lost 98% before lunch. It does say that big rooms tend to be smaller 40 days later. In the one large episode on record, a celebrity launch into a big room was the start of the drain.
Source is DefiLlama, pump.fun daily bonding-curve volume and per-venue daily fees. The page refreshes from the API on load with eighteen requests. If a venue fails it falls back to an embedded snapshot for that chain and says so above. Built by @that1618guy, stateofblocks.com.