What Is a Bonding Curve? How Memecoin Launches Actually Work

If you've bought a memecoin in the last two years, you've almost certainly traded on a bonding curve, whether or not anyone told you that's what was happening.

It's the mechanism behind pump.fun on Solana, four.meme on BNB, and the equivalent pads on every newer chain. And understanding it explains three things that otherwise seem arbitrary: why the price moves the way it does, what people mean by a token "graduating," and why wallet trackers sometimes show a trader selling millions of tokens they apparently never bought.

The basic idea

A normal token needs someone to seed a liquidity pool before anyone can trade it. Someone has to put up real money on both sides, and that's a barrier — it's why launches used to involve presales and team allocations.

A bonding curve removes that. Instead of a pool, there's a formula. The contract holds the entire token supply and sells it according to a fixed mathematical relationship: the more tokens that have been bought, the higher the price of the next one. Sell back, and the price falls the same way.

Nobody provides liquidity. There's no counterparty. You're trading against the curve itself, and the price is entirely determined by how far along it you are.

That's the whole thing. Everything else is a consequence of it.

Why early buyers get better prices

This is the part people intuit but don't quite understand, and it matters because it's not luck or timing — it's arithmetic.

The price at any point is a function of how many tokens have already been sold. Buy at 1% of the way along the curve and you get a much lower price than someone buying at 50%, because the formula says so. There's no discovery, no market, no other traders needed. The first buyer always pays the least.

On pump.fun, tokens launch with a supply of 1 billion, of which roughly 800 million are available on the curve. So the difference between being in the first few transactions and being in the first few minutes can be an order of magnitude on price.

This is also why the wallet stories you see — the $120 into six figures type — almost always involve someone buying in the opening moments at a market cap of a few thousand dollars. They weren't smarter about the token. They were earlier on the curve, and the curve rewarded that mechanically.

The flip side, which gets mentioned far less: the same maths means a large buy on a thin curve moves the price a long way against you. Curves have brutal price impact early on, which is a real cost separate from your slippage setting.

Graduation

When enough tokens have been bought — the curve completes, or "bonds" — the token graduates. The accumulated funds get used to seed a real liquidity pool on an AMM, and the bonding curve shuts off permanently.

On pump.fun this used to mean migrating to Raydium; since PumpSwap launched it goes to their own AMM instead. Other pads have their own equivalents.

After migration, the token trades like any other. Real pool, real liquidity, real market makers, normal price discovery. The curve is done — its reserves are zeroed and it can never be traded again.

Graduation is a meaningful signal, because most tokens never get there. The overwhelming majority of launches stall somewhere on the curve and quietly die. A token that bonded at least had enough buyers to complete it, which is a very low bar but not a zero one.

The two-pool problem, and why your data looks wrong

Here's the practical consequence almost nobody explains.

A graduated token has existed in two separate places: the bonding curve, and the AMM pool it migrated to. Those are two distinct addresses with two distinct trade histories.

Most screeners and aggregators index them as two separate pairs. Their "top traders" and PnL views are calculated per pair. So if you look at the post-graduation pool and check who made money, a wallet that accumulated on the curve and sold after migration appears as a seller with no buys — or with a cost basis of almost nothing.

That's why you'll occasionally see a tracker claiming someone turned $75 into $93,000. They didn't. The entry happened on the curve, on a different pair, and the tool simply isn't looking there.

If you're doing any kind of wallet analysis, this is the single most common source of wrong numbers. The fix is either a tool built natively around the launchpad — GMGN, Photon and Axiom all index the curve as part of the token's lifecycle rather than as a separate pair — or going to the chain directly and reading the wallet's actual token balance changes, which capture every acquisition regardless of where it happened.

What this means if you're trying to trade launches

Being early is mechanical, not clever. The curve rewards position, and position is determined by speed. That's why serious launch trading is pre-configured rather than manual — amounts, slippage and tax limits decided in advance so execution is one action. The specifics are in how to snipe a token launch.

Bonding curve progress is a signal worth watching. A token at 90% is close to graduating; one that's been sitting at 30% for six hours has stalled. Most tools built for launchpads surface this, and it tells you more about a token's trajectory than the chart does at that stage.

Graduation is often a local top. Plenty of people buy the curve specifically to sell into the migration liquidity. If you're buying at the moment of graduation, consider who's selling to you.

The risks specific to curve launches

The low barrier that makes bonding curves useful is the same thing that makes them dangerous. Anyone can deploy in thirty seconds with no capital, which is exactly why tens of thousands of tokens launch daily and almost all of them are worthless.

Two things to look at that are specific to this format:

Bundling. If a large share of the supply was acquired in the opening block by wallets funded from the same source, the launch was farmed before the public saw it. Those wallets need someone to sell to, and the curve's price mechanics mean their entry was vastly better than yours. The full set of checks is in the rug pull guide.

Deployer behaviour. The wallet that created the token has a history. If it's deployed forty tokens this week and dumped every one, that's the most useful piece of information available about the forty-first.

And the base rate is worth restating plainly: the vast majority of curve launches go to zero, usually within hours. Being early on a curve improves your entry on a distribution of outcomes that's still overwhelmingly negative. Discovery matters far less than position sizing, which is covered in more depth in how to find new memecoins early.

Trading them

Axxel's sniper works on both EVM chains and Solana, with configurable slippage and maximum buy and sell tax limits so orders won't fire on tokens that breach your thresholds. Market, limit and trailing orders across Ethereum, Base, BSC, Solana and Robinhood Chain, flat 0.9%, non-custodial, no subscription.

The multi-chain part matters more for launches than for anything else, because the pads move. The dominant launchpad on a chain can be replaced within a week — it happened on Robinhood Chain when the leading pad shut down and the flow moved to competitors almost immediately. Being set up across chains means a pad dying is an inconvenience rather than a migration.

Crypto trading carries risk. Most memecoins lose value. Nothing here is financial advice. Axxel is not available in all regions.