Positions
Up to three at once — vary the range, the pool, or both.
Side by side
Same window, same price path — so the differences are the pool and the range. Deployed is what each position actually costs at the entry price.
Position composition
Model & assumptions
The position maths is exact Uniswap v3. Liquidity L is solved from each
position's USD size at its pool's entry price, then balances follow
x = L(1/√P − 1/√Pb) and y = L(√P − √Pa), clamped at the
range edges. Position value, impermanent loss and final token composition are exact for the given price
path.
Volume and fees are real; only the liquidity split is modelled. Candles come from GeckoTerminal for the exact pool each position points at, so the swap volume, the fee tier and the current TVL are the pool's own numbers. Two things still have to be assumed:
- Pool volume — real, per candle, straight from the pool. Nothing is scaled or estimated.
-
Competing liquidity — no public API exposes the tick-by-tick liquidity curve, so
the rest of the pool's TVL is modelled as one position spanning ±c around spot, giving
L_pool = TVL / (2√P · (1 − (1+c)−1/2)). Your cut each candle isL / (L + L_pool). - Time in range — inside a candle the price is assumed uniformly distributed between low and high, so a candle straddling one of your bounds earns a partial fraction rather than all-or-nothing.
Positions are simulated independently: each holds its own liquidity and none of them dilutes the others. Fees accrue in USD and are not compounded back in — matching Uniswap v3, where fees sit outside the position until collected. No gas costs, no rebalancing, and no price impact from your own size.
Calibrating. One knob is left: the concentration width. Tune it by reading a pool's actual fee APR for a known range and adjusting until the simulator agrees. Note that TVL is read live, so a backtest over a window where a pool's TVL was very different will be off by roughly that ratio.