A SafeMoon-style token charges a fee on every transfer, redistributes part of it to all holders ("reflections") and sends part to a liquidity pool. The mechanics are clever but come with real costs to integrations and users, and SafeMoon itself ended in bankruptcy and criminal convictions for its leadership. Anyone copying the model should understand both.
How reflection tokenomics work
SafeMoon launched on BNB Smart Chain in March 2021, building on an earlier "reflect" contract design. Its original version charged a 10% fee on transfers, roughly split in two:
- Reflection: a share is redistributed to every holder in proportion to their balance, without any transaction to each holder.
- Auto-liquidity: a share accumulates in the contract and is periodically swapped into BNB and added to the DEX pool, with LP tokens held by a designated address.
The aim was to reward holding and punish selling, and the marketing leaned heavily on that: "hold and your balance grows".
The reflection trick
Paying thousands of holders individually would be impossibly expensive. Instead, the contract keeps balances in two units: a large internal "reflected" supply and the visible token supply, linked by a conversion rate. When fees are taken, the reflected supply shrinks, the rate changes, and every holder's visible balance rises at once. balanceOf computes the balance on the fly from the holder's reflected amount and the current rate.
Some addresses, typically the DEX pair, the contract itself and exchanges, are "excluded" from reflections and tracked separately. The original pattern looped over excluded accounts when computing the rate, so gas rose with each exclusion. Variants fixed this in different ways, with different bugs.
The practical problems with the model
| Issue | Why it matters |
|---|---|
| Fee-on-transfer breaks integrations | Many DeFi protocols assume the amount sent equals the amount received. Lending markets, vaults and some routers mis-account or reject such tokens. Uniswap v2-style routers needed special functions just to swap them. |
| Holders' "yield" comes from other holders | Reflections are paid by people who sell or transfer. There is no external revenue; it is a redistribution among participants. |
| Owner powers | Typical contracts let the owner change fees, exclude addresses and set the liquidity recipient. These are trust assumptions users often missed. |
| Honeypot risk | Copies with adjustable fees can be set to tax sells at near 100%, trapping buyers. Scanners flag this, but variants evolve. |
| Gas and complexity | Swapping and adding liquidity inside transfers raises gas costs and creates reentrancy and price-manipulation risk. |
| Exchange support | Centralized exchanges must handle balances that change without transfers, which many prefer to avoid. |
What happened to SafeMoon
SafeMoon's price rose sharply in 2021 and then fell steeply. In March 2023, an attacker exploited a bug in an upgraded contract that left a burn function publicly callable, draining a large amount from the liquidity pool. In November 2023, US prosecutors charged CEO Braden John Karony, creator Kyle Nagy and CTO Thomas Smith with fraud and money-laundering offenses, and the SEC brought parallel civil charges. Authorities alleged that, while telling investors liquidity was locked, insiders withdrew millions of dollars from it for personal use. The company filed for Chapter 7 bankruptcy in December 2023. Smith pleaded guilty, and in May 2025 a federal jury in Brooklyn convicted Karony. Nagy was reported to remain at large.
The lessons for builders are direct:
- "Locked liquidity" claims mean nothing unless the LP tokens are verifiably held by a time-lock contract that nobody can override.
- Upgrades to live contracts need the same audit rigor as the original deployment.
- Marketing a token as a way to grow your holdings invites securities and fraud claims.
If you still want a fee-on-transfer token
There are legitimate uses for transfer fees, for example funding a project treasury transparently, or a game currency with a sink. If you go ahead:
- Use a fixed, low fee cap written into the contract, so the owner cannot raise it later.
- Renounce or time-lock owner powers, and put remaining roles behind a multisig.
- Lock LP tokens in a well-known time-lock contract and publish the transaction.
- Avoid swapping inside transfers where possible; accumulate fees and process them in a separate, permissioned call.
- Test against target DEX routers and wallets, and document fee-on-transfer behavior for integrators.
- Get an independent smart contract audit covering upgrades and admin functions, not just transfer logic.
Often a standard token with a separate, opt-in staking contract achieves the same goals more cleanly. For standard token builds on BNB Smart Chain, see the BEP-20 token guide and the broader BNB Smart Chain overview. For how fee tokens interact with AMM pools, see the decentralised exchange software guide.
Frequently asked questions
Do reflection tokens generate real yield?
No. Reflections are fees paid by other holders when they transfer or sell. Without outside revenue, total value is only redistributed, not created.
Why do some DEX swaps of reflection tokens fail?
The token takes a fee during transfer, so the pool receives less than expected. Users must raise slippage tolerance, and routers need fee-on-transfer-aware functions.
Is it legal to launch a SafeMoon-like token?
The mechanism is not illegal in itself. Misleading claims, undisclosed insider withdrawals and selling it as an investment can lead to securities and fraud liability, as SafeMoon's case showed.
How can buyers check a tax token for a honeypot?
Read the verified contract for adjustable fees, blacklist and max-transaction functions, check owner status, and confirm LP tokens are locked. Simulated sells on test tools help but are not proof.