Why Exchanges Lose Money in a Falling Market - and How Hedging Fixes It

An exchange can charge a fee on every trade and still end up in the red. This happens not because of operator errors or competitor dumping, but because of inventory risk - the price movement of the asset held in reserve. Below we break down three exchange operating models using concrete numbers, show exactly where the loss hides, and explain how a hedging module stabilizes profit.

For all calculations we use the following common assumptions: the spot market price of ETH is 2000 USDT, the exchange fee is 1% of the trade amount, and slippage plus exchange commission when buying or selling on spot is 0.4% (an average for trades of 1–5 ETH, of which 0.1% is the taker/maker fee).

Model 1 - Spot purchase at the moment (broker model)

In this model the exchange holds no inventory. For each client request, it buys or sells the asset on the spot exchange and executes the trade immediately.

This is the simplest model and is risk-free with respect to price fluctuations. It suits early-stage exchanges with low volume. However, it carries a "broker" risk: if the spot exchange account becomes unavailable, the operator can no longer execute trades or earn profit.

Trade direction	Exchange action	Cost / income	Client fee	Net profit
Client buys ETH (pays 2000 USDT)	Buys 1 ETH at 2008 USDT (2000 + 0.4%)	−2008 USDT	+20 USDT	+12 USDT
Client sells ETH to exchange	Sells ETH at 1992 USDT (2000 − 0.4%)	+1992 USDT	+20 USDT	+12 USDT

Result per trade: profit of about 12 USDT - the 20 USDT fee minus 8 USDT on slippage and exchange commission.

Model 2 - Using your own reserves

Here the exchange holds its own ETH inventory and fills trades from it. An own reserve eliminates slippage and exchange commission, which increases margin by that amount.

Trade direction	Exchange action	Cost / income	Client fee	Net profit
Client buys ETH	Gives 1 ETH from reserve	- (uses reserve)	+20 USDT	+20 USDT
Client sells ETH to exchange	Takes 1 ETH into reserve	- (grows reserve)	+20 USDT	+20 USDT

Result per trade: profit of 20 USDT - the full fee, with no slippage. On larger trades (for example 10 ETH), slippage and commission in the first model grow, so using an own reserve raises margin by roughly 67%.

Comparison of the two models (per 1 ETH)

Metric	Model 1 (spot at the moment)	Model 2 (own reserve)
Profit per trade	12 USDT	20 USDT
Price-movement risk	Very low	High
Foregone profit on a rising price	High	Low
Scalability	Good	Limited

Model 1 delivers stable but smaller profit (roughly 40% less due to slippage and exchange commission) at near-zero risk. Model 2 delivers 8 USDT more per trade but adds the risk of inventory price changes. That risk is the central subject of the rest of this analysis.

Where the loss hides

To see inventory risk in action, consider Model 2 under a price change. The exchange holds 1 ETH in inventory, bought earlier at 2000 USDT, with a 1% fee. We examine a drop to 1900 USDT and a rise to 2100 USDT.

Situation	ETH price	Client action	Received / paid	Real ETH value	Net result
Price fell	1900	Client buys	+1919 USDT	2000	−81 USDT
Price fell	1900	Client sells	−1881 USDT	1900	+19 USDT
Price rose	2100	Client buys	+2121 USDT	2000	+121 USDT
Price rose	2100	Client sells	−2079 USDT	2100	+21 USDT

Three conclusions follow. When a client buys ETH from the exchange (crypto leaves the reserve), a price drop produces a loss (−81 USDT - four times the fee), while a rise produces extra profit (+121 USDT - six times the fee). When a client sells ETH to the exchange (crypto enters the reserve), profit is almost always just the fee (~19–21 USDT) regardless of price movement.

The most dangerous scenario is when the price falls while clients actively buy ETH. That is precisely when a real loss appears. In Model 2 profit is unstable: on a rising price the exchange earns more than the fee; on a falling price with active buying it loses money.

Model 3 - Reserve plus hedge

To remove inventory risk, a hedge is added to Model 2: on each trade a corresponding position in perpetual futures is opened, offsetting the price change.

Situation	Client action	Hedge action	Received / paid	Hedge effect (PnL)	Net profit
Price fell to 1900	Client buys	Short opened earlier	+1919 USDT	+100 USDT	+19 USDT
Price fell to 1900	Client sells	Open short	−1881 USDT	0 USDT	+19 USDT
Price rose to 2100	Client buys	Short opened earlier	+2121 USDT	−100 USDT	+21 USDT
Price rose to 2100	Client sells	Open short	−2079 USDT	0 USDT	+21 USDT

Model 2 with and without a hedge

Metric	Without hedge	With hedge
Price falls, client buys	−81 USDT	+19 USDT
Price falls, client sells	+19 USDT	+19 USDT
Price rises, client buys	+121 USDT	+21 USDT
Price rises, client sells	+21 USDT	+21 USDT
Profit stability	Highly variable	Almost always ≈ fee
Risk of large losses	High	Very low

Without a hedge, profit depends heavily on price movement, and a fall combined with active buying produces a real loss. With a hedge, profit stabilizes and almost always equals the fee. The hedge offsets inventory price changes: the exchange neither loses on a drop nor gains excess profit on a rise, when the hedge is fully on. The main effect is protection from loss on a falling price when clients are actively buying.

What a hedge costs at volume

A hedge is not free. To estimate its real cost, consider a two-month period with the following conditions: daily volume of 20 ETH, a 1% fee, an initial ETH price of 2000 USDT, a final price of 2600 USDT (a 30% rise), an average price over the period of 2300 USDT, and an average inventory of 60 ETH under linear accumulation. Fee profit over the period is +27,600 USDT.

Metric	Without hedge	Hedge (min cost)	Hedge (medium cost)	Hedge (high cost)
Fee profit	+27,600 $	+27,600 $	+27,600 $	+27,600 $
Profit from inventory rise	+36,000 $	0 $	0 $	0 $
Hedging cost	0 $	−3,174 $	−4,500 $	−7,000 $
Net profit over 2 months	+63,600 $	+24,426 $	+23,100 $	+20,600 $

Breakdown of the minimum hedging cost (~3,174 $): funding around 1,490–1,518 $, commissions 1,104 $, slippage 552 $.

The flip side is visible here too. Even at a high hedging cost (up to 7,000 $), the exchange stays profitable (+20,600 $ over two months). However, in a strongly rising market (+30%), profit without a hedge is significantly higher (+63,600 $), and the difference is 39,000–43,000 $ in favor of operating without a hedge.

In other words, a hedge is not a way to earn more - it is a way to remove uncertainty. In a rising market, going unhedged would have earned more, but no one knows in advance whether the market will rise or fall. A hedge converts unpredictable profit carrying the risk of a large loss into stable profit at the level of the fee.

Conclusion

It is prudent to run all trades through a hedging mechanism. In its normal state - default hedge - the module performs baseline hedging and delivers stable profit equal to the fee, protecting the exchange from the most dangerous scenario: a falling price with active client buying. When needed, the degree of hedging can be adjusted, leaving partial market exposure if the operator deliberately chooses to carry that risk for potential extra profit.