Altitude

The feed answers one question: what just graduated, and how did it score? It is ordered by recency, so a launch leaves view within hours no matter what it goes on to do.

Altitude answers the other one. Of everything we have already scored, what is still worth something today?

What qualifies

Three things, and all of them are filters rather than opinions:

• It cleared the bonding curve. Around 13,000 tokens a day reach a curve on Solana. Most never finish one.
• It survived the first 24 hours. The day after graduation is when almost everything dies, and nothing enters this list before it is over.
• It still holds a real market cap. Above $40,000, measured continuously.

A token that drops away leaves the list. One that climbs into it appears. Neither happens instantly — see Why it does not flicker below.

Where the market cap comes from

Not from the price feed's own figure.

We measured this: across 5,841 tokens holding both a cap and a price, 1,931 — 33% — disagreed with price × known supply by more than 20%. One read $27,287 against a real $7,843, implying a supply of 3.48 billion for a mint whose own metadata says exactly 1 billion.

So the cap shown here is computed — the price, which is well sourced, multiplied by the supply, which is a fact we store. Where we do not hold the supply we cannot compute it, and the token is left out entirely rather than admitted on a figure we know to be wrong a third of the time. A market cap nothing would price is unpriced, not zero.

How the ranking works

Market cap is the entry requirement, not the ranking. Sorting by cap would put the tokens with the least room left to run at the top, which is the opposite of the question this page is asking.

What ranks them:

Liquidity against the cap. The heaviest term, and the one that catches the most expensive mistake. A market cap is what the last trade implied. Liquidity is what you could actually get back. A $50,000 cap sitting on $200 of liquidity is not a $50,000 token — it is a door too narrow to leave through, and it ranks below a smaller token you can actually sell.

Turnover. Volume measured against the cap. A cap with nothing trading behind it is a chart nobody is participating in.

How much of its peak it still holds. The section is named for this. The question is not how high it got, it is how much of that it kept.

What we already knew. The launch score, the cabal read, whether the crowd was organic, how much supply the launch block still held, and whether that crew has ever produced a real multiple before.

Recent price direction is in there too, and deliberately weighted below all of the above. It is the noisiest input and the easiest to over-trust: it breaks ties, it does not decide rows.

These weights are provisional

They have never been tested against a forward outcome, because nothing has been ranked by them until now. Every pass is recorded so that in a few weeks "which of these actually predicted a higher cap" is a query rather than an argument — and the weights will change when that measurement exists.

This is the same discipline the alpha score is held to, and it exists because two scoring rules have already shipped here on plausibility alone and had to be withdrawn.

Why it does not flicker

A single threshold would make a token sitting near $40,000 join and leave on every refresh — sixty times an hour, re-ordering the page under whoever is reading it.

So a token enters above $40,000 and only leaves below $36,000, and a low reading has to repeat before it counts. One bad print from one data provider cannot take a row off the page.

Why the list is short

This is the part worth being direct about.

A trading terminal ranks whatever moved in the last five minutes and refreshes forever. That is not an accident of design — a trader who checks thirty times a day and acts on impulse pays thirty times the fees, and on those platforms the fee is the product.

We take nothing from your trades. No commission, no spread, no percentage of a win, no cut of a loss. There is therefore nothing on this page tuned to make you click: the list is capped at thirty, it changes when the evidence changes, and most days it barely moves.

A short list that rarely changes is a worse business model and a better answer. If only four things qualify today, you will see four.

The DexScreener Boosted tab

Altitude has a second tab: DexScreener Boosted. DexScreener's own board is ranked by paid boosts — pay-to-appear — and this tab shows that same board, in the same layout as our own list and in the same boost order a viewer sees there. It exists to make one comparison visible: which of the tokens someone paid to surface, our own algos had already flagged first — earlier, and for free. Where we scored a token, its row carries our score and the same signal tags the Our Own Algos list shows, next to the boost that was paid for it.

Rugged boosters are kept, greyed out. A token that paid to appear and then collapsed is not removed — a board that hides the ⚡500 that died is the brochure this tab exists to argue against.

We keep a lifetime record of the big boosters. Every token paying a large boost is remembered even after it leaves the board, so the banner can answer of every big-boost token we have ever tracked, how many have since rugged and never recovered? A booster that pumped and then had its liquidity pulled — so it can no longer even be priced — counts as rugged. Paying to appear is not the same as being worth appearing, and this is the number that says so.

What it is not

It is not a buy list, and it is not a forecast. A high altitude score means a token with fewer things wrong with it than the others in the window — measured, shown with its working, and open to your disagreement. What it does next is not something this or any other page knows.