Short answer: GMX charges an open/close fee, a continuous borrow fee based on pool utilization, and price impact based on pool depth — none of which map one-to-one onto a CEX’s maker/taker fee and periodic funding rate. For an algo trader, the real comparison requires modeling both structures over the position’s actual expected duration, not comparing headline percentages.
Why ‘GMX fees vs. CEX fees’ isn’t a single number
A common mistake is comparing GMX’s advertised fee percentage directly against a CEX’s taker fee, as if they were the same line item. They’re not: GMX bundles an open/close fee, a continuous borrow fee, and trade-size-dependent price impact into the real cost of a position, while a CEX typically separates a flat maker/taker fee from a periodic funding rate.
The only fair comparison is total cost over the position’s actual expected holding period — a fee structure that looks cheaper per trade can cost more over a multi-day hold, and vice versa.
Borrow fee vs. funding rate: different mechanics entirely
GMX’s borrow fee accrues continuously based on how much of the relevant GM pool’s liquidity your position is utilizing — it applies regardless of whether you’re long or short, and scales with how concentrated that pool’s usage is. A CEX’s funding rate is periodic (commonly every 8 hours) and driven by the imbalance between aggregate long and short open interest across the exchange.
This means a position held for exactly the same duration can accrue meaningfully different total financing costs on each venue, and the direction of that difference depends on current pool utilization on GMX versus current long/short skew on the CEX — neither is fixed, so this has to be checked at trade time, not assumed from a table of historical averages.
Price impact: the cost order-book slippage doesn’t quite capture
On GMX, price impact is a deterministic function of trade size against current GM pool depth, computable before you submit the trade. This is arguably more transparent than order-book slippage on a CEX, which depends on resting orders that can be pulled or filled by other traders in the milliseconds before your order lands.
For an algo strategy sizing positions programmatically, this transparency is an advantage: GMX’s price impact can be estimated from on-chain pool state ahead of time, while CEX slippage estimation typically relies on historical order book snapshots that may not reflect current conditions.
The comparison that actually matters
Rather than asking ‘is GMX cheaper than exchange X’, the useful question for an algo strategy is: for this specific position size and expected holding period, which venue’s total cost — fees plus financing plus impact — comes out lower right now. That answer changes with pool utilization and market conditions, which is exactly why it needs to be modeled per trade rather than assumed once.