Training note, written on 27 September 2026 from data as of 24 September 2026. Not part of the published series.

Raydium AMM is growing more than the other DEXs

Luca Valente — dictated note, lightly edited

The use of Raydium AMM is growing in fees, in volumes and in deposited capital, and it is going much faster than the other DEXs in its group. This is a very solid signal, and it matters precisely because it rises so much against the competitors. Raydium AMM is a decentralized exchange on Solana and it has its own token, RAY, which however counts for the whole of Raydium, so I will not take it into consideration in this analysis.

Let's look at the measures I checked. First of all, fees for the month rose 295.8%, while the median is 38.2%, so there is a spread of 257.6 points. Then I looked at volumes, which more than doubled, +155.1% against 27% for the group, and here too there is a spread of 128.1 points. Finally I checked deposited capital, which rose 18.2% against 9.2% for the group, and here too we are at double, for a total of 1.3 billion dollars deposited. Of all these indicators, the first to move were fees, then capital, and finally volumes.

The growth in fees makes it unlikely that this is only a movement of the whole sector, even if it does not rule it out. One would have to look at the wallets, at how many different wallets pay those fees. Volumes too make it unlikely that this is a spike of a few days; they do not rule it out, but they make it unlikely. Capital too makes it unlikely that this is only passing activity. In any case we have to check what capital does after this peak. But the most important thing I said at the start of the analysis: there is an enormous difference between Raydium AMM and its reference group, which is 26 DEXs, as I said before; the spreads are enormous.

The unusual fact I want to point out concerns fees, which are growing more than volumes, 295.8% against 155.1%, so average fees per dollar traded are higher, if both measures cover the same pools and periods.

If we want to look at those who think the opposite, someone could say we are facing a spike and not a trend, and this thesis could rest on the fact that on 24 September volumes were 343.2 million, well below the 573 million of 21 September. But it also has to be said that the engine does not have enough similar episodes to predict how this kind of movement ends.

These data do not say one thing: they do not say who is trading, because we have not measured users and wallets, so it may be growth that comes from a few traders or from many traders, and this obviously makes a difference. And they do not say either why fees are running more than volumes, and any cause I could pull out would be invented. We cannot verify development either, because we have no code repository for this platform.

So what I will want to see in the coming days is whether 30-day fees and volumes fall, grow or slow down. We will look at daily volumes, whether they stay above 343 million, so that the peak of the 21st was a high point of a climb and not an exception. And for fees, today we are around 2 million a day, and here too we will have to see whether they hold, to see whether the growth becomes stable income. So, to conclude: if within a month, that is by 24 October, fees and volumes fall below the 30-day average up to 24 September and deposited capital falls below the 24 September level, the judgment on this Raydium run falls, and so we will check these three indicators over the next month.

Data: DefiLlama, GitHub; 30-day sums for flows, end-of-window for stocks; peer group: 26 exchanges, one product per family; as of 24 September 2026.

Not financial advice. This is a research note based on public on-chain data.