Protocol
How FLUXION works
FLUXION is a liquidity intelligence layer for Solana. It finds every pool holding a token across the major venues, normalizes their data, and models how a liquidity position would perform under assumptions you control.
Observed is not modeled
Market data and estimates are labeled differently everywhere. Cyan marks observed data; violet marks modeled estimates.
No silent gaps
Missing inputs show as “Insufficient data”. Nothing is interpolated or invented.
Stateless by design
Serverless requests on demand, with no accounts, wallets or database. Nothing runs until you ask for it.
Data sources
Where the numbers come from
Each request has a timeout and retries with exponential backoff. Responses are cached at the edge and in memory. If a source fails, the report shows which one and continues with the rest.
Token identity
Verification before analysis
Input must be a valid Solana public key: base58 text that decodes to exactly 32 bytes. The mint is then looked up in Jupiter's token registry. Search shows verified tokens first and hides unverified ones unless you choose to show them.
Paired tokens are verified as well. A pool whose other token is unverified stays in the map and table, but it is excluded from allocations. This guards against look-alike tokens that copy the names of well-known assets.
Methodology
The position model
Every modeled figure is compared against simply holding the tokens you would deposit, so it excludes the direction of price. Inputs: your amount A, holding period, volume basis and multiplier, range width w, strategy (hold range or rebalance on exit), transaction cost and volatility σ: the observed 30-day realized volatility of the pool price from GeckoTerminal, or Orca's 7-day price history when that is unavailable, unless you override it.
Utilization
24h volume ÷ pool liquidity
ObservedEffective fee rate
observed 24h fees ÷ 24h volume (falls back to the fee tier)
ObservedYour fee share
A·k ÷ (TVL + A·k), where k = 1 for full-range pools, (your range efficiency ÷ observed active-liquidity multiple) for Orca, and your stated fee-share multiplier elsewhere
ModeledFees earned
daily volume × effective fee × LP share × your fee share × days × expected time in range
ModeledCapital efficiency
2 ÷ (2 − √(1−w) − 1/√(1+w)) for a ±w range
ModeledTime in range
average over the horizon of P(price stays in range), with log-price as a driftless random walk at realized volatility
ModeledDivergence
expected value vs. holding the initial tokens, integrated over the lognormal price distribution at the horizon
ModeledRebalances
horizon ÷ expected exit time, where E[exit] = a·b ÷ σ² for log-distances a, b to the range edges
ModeledSwap cost
size × fee + size² ÷ (depth + size), a constant-product upper bound applied to half the position
ModeledFragmentation
HHI = Σ shareᵢ²; effective venues = 1 ÷ HHI
ObservedThe optimizer funds pools one increment at a time, always choosing the largest modeled net gain. It respects your caps on pool share and pool count, and leaves capital unallocated when no pool improves the net outcome. Its result is a starting point for the allocation sliders, not a recommendation.
Risk management
How the optimizer handles risk
The optimizer never ranks a pool by its fees alone. It maximizes a risk-adjusted score: projected fees discounted for uncertainty, minus expected divergence and costs, minus penalties for volatility, concentration and thin liquidity. It never has to invest the full amount, so capital that cannot clear the bar stays unallocated.
- Volatility tiers use the annualized volatility of the pool price: low below 50%, moderate 50–100%, high 100–150%, extreme 150% and above.
- Fee uncertainty. Projected fees are multiplied by a confidence factor. It is reduced when 24h volume diverges from the 7-day average, when only a 24h snapshot exists, when fees are modeled rather than observed, when fees are dynamic, and when competing liquidity concentration is assumed. Daily LP fee yields above 0.2% of TVL are scaled by √(0.2% ÷ yield), because unusually high fees rarely persist.
- Volatility penalty: the profile's risk aversion (0.5 / 0.25 / 0.1) times the divergence at the 5th or 95th percentile price outcome. It is 1.5× for extreme pools and 1.25× when volatility comes from fewer than 14 days of data.
- Liquidity penalty: 1% × $500k ÷ (TVL + $500k) of capital, plus 0.25 × amount² ÷ TVL for depth. Concentration penalty: κ × amount² ÷ total, with κ = 0.04 / 0.02 / 0.01.
- Pair exposure. Each deposit is the analyzed token plus a paired asset, and outcomes are measured against holding that pair. Because that comparison ignores the paired asset's own price risk, the optimizer only compares and funds pools whose paired asset is in your exposure policy: USD stablecoins, SOL & liquid staking tokens, or other verified tokens.
Limitations
What the model does not know
- Liquidity distribution per tick or bin is not published by these APIs. Active-liquidity depth is derived only for Orca.
- Real prices trend and jump. A random walk at realized volatility understates tail events.
- Volume and fees change quickly. The model assumes the selected volume level holds for the whole horizon.
- Dynamic and scheduled fees are captured only through the observed effective rate.
- Smart-contract, oracle and depeg risks are outside the model.
FAQ
Common questions
Does FLUXION execute trades or hold funds?
No. FLUXION reads public market data and runs calculations. It never connects to a wallet, requests signatures, holds funds or sends transactions. The amount you enter is used only as a model input.
Why don't you show pool APYs?
Annualizing one day of fees treats a single day as permanent and ignores who earns those fees. FLUXION shows observed fees for the last 24 hours, not annualized, and models your share separately from assumptions you can see and change.
Why is a pool marked “Insufficient data”?
A model input is missing. The venue may not publish the fee tier or volume, the paired token may not be verified, or volatility history may be unavailable. FLUXION does not fill these gaps with guesses.
How fresh is the data?
Sources are queried when you request an analysis. Results are cached for up to 60 seconds, and price history for 30 minutes. If a refresh fails, the last good data is shown and marked stale with its age.
Are incentives and farm rewards included?
No. Only trading fees are modeled. Reward programs change often and are paid in other tokens.
Run your first analysis
Choose a token and an amount. The report is built from live data.
FLUXION provides analytics for informational purposes only. Nothing on this site is financial, investment or tax advice. Modeled values are estimates that depend on stated assumptions and will differ from actual results. Providing liquidity can lose money.