Between our October 14 and our October 15, 2026 records, BNB moved from a B at 69.5, in position 10 of our ranking, to a C at 59.9 in position 16. Our record attaches no cause to that move, so we are not going to invent one.

What happened next in the table is the interesting part, and it is arithmetic rather than opinion. Six assets each rose exactly one position, and not one of them moved on its own score. PancakeSwap 11 to 10, Arweave 12 to 11, Convex Finance 13 to 12, Polkadot 14 to 13, Raydium 15 to 14, Decred 16 to 15. They did not improve. The asset above them left.

That matters for a practical reason. Logged out, our ranking at https://cryptograde.ai/tokens shows you the top 10. So one asset changing letter moved the line, and the assets immediately past it are exactly the ones a free account exists to show you.

Here are three of them:

ARWEAVE, GRADED B, SCORE 64.0, POSITION 11

Arweave sits one place past what a logged out visitor sees, and it got there without its score changing: 64.0 on October 14, 64.0 on October 15.

Our model's assessment on that record, in its own words: "The supplied data support a token with limited dilution risk, a differentiated storage use case, and ongoing developer activity. Offsetting that, the evidence for strong organic usage, deep ecosystem moat, and attractive relative valuation is incomplete or modest. The result is a stable but not high-conviction profile from the available record."

Note what that paragraph refuses to do. On tokenomics our note says circulating supply is 65.65M of a 66.0M maximum with FDV roughly equal to market cap, which is a concrete reason dilution risk reads low. On valuation it says the opposite kind of thing: market cap and FDV are both about $239.6M, the price is far below its all time high, and "relative value cannot be confirmed from supplied metrics."

A score of 64.0 with that note attached tells you something a number alone cannot: which part of the case is evidenced and which part is a gap.

CONVEX FINANCE, GRADED B, SCORE 62.0, POSITION 12

Convex Finance is 2.0 points below Arweave and makes an entirely different claim on an investor. Our note on the October 15, 2026 record puts real protocol activity on the table, and then declines to credit the token for it:

"Convex continues to show strong protocol-scale usage, with $559.4M TVL and $2.77M in 30-day fees supporting relevance in the DeFi stack. Tokenomics are acceptable because supply is near capped, but the supplied record still does not establish direct token-holder fee entitlement or strong cash-flow capture."

That distinction, between a protocol earning fees and a token holding a claim on them, is one of the most common ways a crypto investment thesis goes wrong, and it is the sort of thing a framework has to be willing to say out loud about an asset it still grades B.

POLKADOT, GRADED B, SCORE 61.3, POSITION 13

Polkadot is 0.7 points below Convex Finance, a third kind of asset again, and scored on the same six dimensions.

Our note: "Development activity in the canonical SDK repo is strong and recent, which supports usage and ecosystem durability. However, the record still lacks direct on-chain usage, fee/revenue, audit, distribution, and peer-valuation data, so tokenomics and valuation remain weakly supported."

On usage specifically our note cites 268 commits and 73 active contributors in that repository, and then says plainly that the dimension "rests on protocol development rather than measured on-chain usage." On valuation, with market cap about $1.97B and no peer ratios or fee inputs available, "only a low relative-valuation score is supportable."

THREE ASSETS, 2.7 POINTS, ONE SCALE

A permanent storage network, a layer sitting inside the DeFi stack and a layer-1 protocol all land between 61.3 and 64.0 on the same framework, and each one carries a different reason for being there. That is the whole point of a comparable scale: the assets do not have to be alike for the comparison to mean something.

The weights are published, so you can argue with the result precisely. Tokenomics and value accrual 28 percent, on-chain usage and economics 22 percent, security, decentralization and durability 18 percent, ecosystem and moat 12 percent, relative valuation 12 percent, market structure 8 percent. On top of that sit nine pass or fail screens, and a tripped screen places an asset in our F grade at any score, including a score high enough for a much better band. That is why we always publish the grade our record carries and never work a letter out from a score, and why we do not tell you which screen applied to a given asset: our record does not say, so we do not guess.

The published bands are 80 and above for A, 60 to just under 80 for B, 40 to just under 60 for C, and below 40 for F, with those screens overriding all four. Read the full framework at https://cryptograde.ai/framework

WHAT THE FREE ACCOUNT ACTUALLY OPENS

Three things, stated without decoration.

It costs an email address. There is no paid tier to decline, nothing is charged, and nobody we grade pays us anything: CryptoGrade does not work commercially with the assets or projects it covers, and it does not sell data.

Create the account at https://cryptograde.ai/sign-up

ONE HONEST LIMIT

A grade is a reading of an evidence record on a date, not a verdict for all time. Every figure in this piece is stated against our October 15, 2026 record, and positions in particular can move without the asset doing anything, which is exactly what six assets just demonstrated. Not every entry in our ranking carries a letter grade either: where our evidence record is incomplete for an asset, our model publishes no grade for it rather than guessing one, and we will not write a letter next to an asset that does not have one. The three assets above do have one, which is why they are the three we named.