Slippage

Two sources cause slippage on Solana: market movement between quote and fill, and the size of your trade relative to pool depth. The first is the cost of latency. The second is the cost of consuming liquidity as you walk price. CLMMs feel both because the price step is continuous. AMMs feel the second heavily because reserves move with every fill. DLMMs feel only the cross-bin component since intra-bin trades have zero slippage by design.

Every production swap instruction has a slippage param. On Raydium AMM v4 it is minimum_amount_outfor swaps. On Orca Whirlpools it is other_amount_threshold combined with a sqrt-price limit. On Jupiter aggregator routes it is a basis-points value applied across each leg.

Common values: 0.5% for blue-chip swaps, 1% to 3% for memecoins during normal volume, 5% to 15% for new launches and high- volatility tokens. If you set zero slippage tolerance and the block ahead of you moves price by even one wei, your transaction fails and you pay the priority fee anyway.

One opinion: a lot of failed transactions during congestion are not failed because of compute-unit limits. They are failed because the trader set 0.5% slippage on a memecoin during a launch event. Tune slippage by token volatility, not by a "safe" default.

Two patterns help in production. First, quote against the same on-chain pool state you will execute against; do not quote on a stale snapshot. Second, simulate before you submit for any trade above a few hundred dollars; simulation runs the same swap math the validator will and tells you the exact min-out at current state, which lets you set slippage tight without false rejects.

Slippage params show up on every DEX swap instruction. Watch for transaction logs containing SlippageToleranceExceeded or program-specific errors like Raydium's ExceededSlippage. Those are the failures you can fix with config, not with code.

Our parsed swap events on Raydium, Orca, and Meteora include both the quoted and executed amounts, so you can compute realised slippage per swap and tune your bot in production.