ROI vs. holding compares your position's value to what you'd have if you'd just held the deposited tokens. It's a useful check on impermanent loss. Once most of the capital has been withdrawn, the percentage is measured against a small remaining base and stops meaning much.
An unpriced token in a position doesn't get averaged out or ignored, it gets left out of the total, which means the total is a floor, not a sum. We'd rather show a visible gap than fill it with a guess.
Every harvest you've made is sitting on chain already, timestamped and priced at the moment you claimed it. Most wallets never surface that. Sirius just reads it back to you, claim by claim.
After a large withdrawal, ROI percentage is measured against whatever small base of capital remains, so it can read absurdly high. The dollar figures underneath are still exact. Treat the percentage as context, not as the number.
Read-only access can't sign a swap, approve a contract, or move a token even if the tool were compromised. An analytics tool doesn't need transaction capability to do its job, so it shouldn't have it.
Sirius only reads Raydium CLMM positions on Solana. No Orca, no Meteora, no other chains. Narrower scope means every number is computed from real on-chain history instead of approximated across protocols it doesn't fully understand.
What actually happened to that position is specific to it: time spent in range versus sitting idle out of range, fees actually harvested, reward emissions actually claimed. None of that is in the pool estimate.
Take one open position on a Raydium CLMM pool. The pool page shows a single estimated APR, the same number shown to every LP in that pool regardless of when they entered.
Reward emissions are a separate token stream from trading fees, paid on top of them. Count only fees and your actual return looks low. Count rewards twice across tools and it looks high. Sirius folds both into one earnings figure, once.
An unpriced token in a position doesn't get valued at zero, it gets left out of the total. That total is then a floor, not a full count. Guessing a price would hide the gap. Showing it as unpriced is the more honest number.
Every harvest gets timestamped with its USD value at claim time, not marked to today's price. Add them up and you get what you actually took out, not what the tokens are worth in hindsight.
Raydium's APR is a pool-wide average across every position, weighted by capital and time in range. Your position has its own entry point, its own time in range, its own harvests. The average was never describing you.
ROI vs. holding compares your position to doing nothing: just holding the two deposited tokens. It stops meaning much once you've withdrawn most of the capital, the percentage is then measured against whatever small base is left in the position.
After a large withdrawal, ROI on what's left gets measured against a small remaining base. The percentage can spike even though nothing changed. The dollar figures stay exact. Read those, not the percentage.
Run those actual events through the position and the realized APR comes out lower than the pool page implied, sometimes by a wide margin. That gap between estimate and reality is the number worth knowing before the other one.
A pool page shows one APR for everyone in it. Take a single position: the pool's estimate assumes constant liquidity, full range coverage, and every fee reinvested. None of that describes what actually happened to your capital.
The real path is different: hours out of range earning nothing, fees harvested at irregular intervals, reward emissions that show up separately from swap fees and rarely get added back into the return figure.
Sirius only reads Raydium CLMM positions on Solana. No Orca, no Meteora, no other chains. A narrow tool that gets the math right beats a broad one that averages across venues it doesn't fully understand.
Reward emissions are separate from swap fees, but they land in the same wallet and get spent the same way. Count them once, in the same currency as fees, or your realized return is wrong in either direction.
When a token has no price source, we don't estimate it. It shows as unpriced, and it's excluded from your totals rather than guessed at. A total that's quietly wrong is worse than a total that admits what it's missing.