Burn
Buy the token and destroy it. Favoured when the pool is already deep enough for the trades it sees and buying is steady.
a token with a model in its fees
The answer comes back as three probabilities, and the fee is cut by exactly those numbers: some is burned, some deepens the pool, some is paid to the wallets that never sold.
fee = p(burn) + p(pool) + p(holders)
eal is a token with a model wired into its fees. When someone swaps, the fee from that one swap is handed to a model along with the facts of the trade: its size, the depth of the pool, who is buying and who is leaving.
The model does not write a reply. It returns a probability for each of three routes, and the three always add up to one. Those numbers are the split. Nothing runs on a schedule here; a swap is the only thing that moves it.
| contract | not set |
|---|---|
| market value | — |
| swaps read by the model | — |
| wallets with weight | — |
| swaps in 24 hours | — |
| value swapped in 24 hours | — |
One dot per swap. A dot in a corner means the whole fee went that way; a dot in the middle means an even split. No swaps have been read yet.
| route | share | amount |
|---|---|---|
| burn | — | — |
| pool | — | — |
| holders | — | — |
| flow | — |
|---|---|
| pool stress | — |
| confidence | — |
| answered in | — |
Buy the token and destroy it. Favoured when the pool is already deep enough for the trades it sees and buying is steady.
Add the fee to liquidity. Favoured when one trade moves the price too far, or the pool is thin against the token's market value.
Pay the wallets that kept their tokens. Favoured when someone is selling hard or leaving, and the wallets that stayed are absorbing it.
The holders' share of each fee is not divided by balance alone. It is divided by weight, and weight remembers whether you sold.
weight = balance × kept
Balance is what a wallet holds now. Kept is the fraction of everything it ever bought that it has not sold.
A wallet that bought and never sold has kept 100% and earns on its full balance. A wallet that sold half of what it bought earns on half of what remains, until it buys back. There is no lock, no timer and no penalty period: the number is recomputed from the wallet's own trades every time a swap lands.
The wallet making a sale takes no part of the holders' share from that sale. Splitting a bag across several wallets gains nothing, because the pieces add up to the same weight.
| wallet | holds | kept | weight | earned | paid |
|---|---|---|---|---|---|
| No wallets have weight yet. | |||||
Earned is what the rule assigns to a wallet. It only counts as paid once a transfer is recorded.
Every trade the model has read, next to the answer it gave. This is the page to judge it by.
| time | side | value | fee | burn | pool | holders | wallet |
|---|---|---|---|---|---|---|---|
| No swaps have been read yet. | |||||||
Should lean toward holders, and the larger the sale against the pool, the further they should lean.
Should lean toward the pool, whichever side they are on.
Should lean toward the burn. If the log does not look like this, the question needs rewriting.
Why a token would let a model decide what each fee does, and what holders should expect from it.
A token that charges a fee has to decide what the fee is for. Most decide at launch and never again: always burn, always pay holders, always send it to a treasury. Each of those is right in some markets and useless in others. Burning does little for a pool so shallow that a single seller can knock the price down. Paying holders does little when nobody is selling. Adding liquidity to a pool that is already deep changes nothing anyone can feel.
eal makes the decision again for every swap. The facts of the trade are given to a model, and it is asked which of the three routes the market needs most right now. It answers with a probability for each. A sure answer sends nearly the whole fee one way; an unsure answer spreads it. Either way the entire fee is assigned the moment the trade is read.
The intended effect is that the fee works against whatever just happened. A heavy sell into a thin pool is the worst moment for a small token, and it is also the moment that produces the largest fee. Read as stress, that fee should go back into the pool that was drained and to the wallets that did not leave. Calm buying into a healthy pool should send it to the burn instead. The model is not given this as a rule. It is shown the state and asked.
The holders' share is divided by weight: balance multiplied by the fraction of a wallet's purchases it has kept. Two wallets holding the same amount do not earn the same if one of them got there by selling down a bigger position. The seller is left out of the holders' share of its own sale, so a large exit pays the people it would otherwise hurt.
A model that replies in sentences would need its reply parsed and checked before money could move on it, and it would be too slow and too costly to ask on every trade. A model that can only return probabilities over the options it was given has an answer that is already in the shape of a split, and that always sums to one.
The model can be wrong, and a wrong answer sends real fees the wrong way. Each answer is stored with the facts it was given so it can be checked afterwards. Balances are built from the trades this site has watched, so tokens moved between wallets by transfer are not counted as weight until they trade. Amounts shown here are what the rule assigns; they are paid only when a transfer is recorded. Nothing on this site is financial advice.