Even if you buy the right thing, it can still kill you.

I shorted RIVER around the 50 mark. It went up to over 80. The price didn’t kill me; first the fees drained my margin. When I got out, the direction was still correct. Later it dropped to 1.2. The high point on January 26 was around 87, and by October 2 it had fallen by 98% or more. The price won, and my account was gone.

This isn’t a matter of one coin on Binance. Centralized exchanges and on-chain exchanges are all using this same setup.

The spot market is shallow and the liquidity is concentrated; the futures market is deeper than the spot. If you watch it, you think it either pumps or dumps. Out of a hundred, only a few can really get pulled away. And even when it does, it still leaves you with less money you can take out.

If longs crowd in, the funding rate gets set positive—longs pay. If shorts crowd in, it gets set negative—shorts pay. Price can stay put. You’re waiting for direction; it’s collecting rent. On-chain, the rules are looser, so you die faster.

First, see how big the cap gap is.

Most of Binance’s altcoins have a per-period upper and lower limit of 2%, and they settle once every 4 hours or 8 hours. On Hyperliquid, it settles once every hour: the cap is 4% per hour, and every coin is the same—intentionally set wide. Lighter is also settled hourly; the cap is about 0.5% per hour. Some Aster contracts have already changed to hourly settlement. With the same crowding, on-chain you can lose several percentage points of your notional position within just one hour. A 1x still can’t stop it; the liquidation price will keep moving toward the current price.

From January to September this year on Binance: out of more than 600,000 settlements, 146 times were pushed to -2% across 82 contracts. There were zero times pushed to +2%. On the centralized side, the main thing they openly capped and killed were shorts. On-chain doesn’t have such neat ledgers, but the caps are higher, hourly settlement, and the same overcrowding is even harsher.

It’s already played out—those few trades.

RIVER is my own statement. The high was 87; now it’s $1.2. In the middle, I died to the funding rate—not to liquidation price. By October 1st, on Binance the 4-hour funding rate could still annualize to about 74%, and the position kept adding. When it just moves sideways and doesn’t move, longs are still paying.

RAVE this year surged to more than 19 in April. Someone was short at the top; the direction was right. Binance’s hourly funding rate hit -2%: shorts paid 2% of the position each hour—about 48% in a day. Someone called the timing right, but the account still lost to over 3,000 points. When the funding rate backed off, the price dropped from the high too. By October it was about twenty cents.

DEXE this year touched 48 in July, then in ten-odd days it fell to around 2 dollars. It dropped about 96%. The front part was a mismatch in cross-site funding rates forcing a squeeze. After topping out, the custodial addresses sent coins into the exchange; the spot couldn’t keep up. The leverage then blew up as well. A year ago it was still 12 bucks. That middle leg of the rally is the move meant for people to enter.

The earliest on-chain case was JELLY. On March 26, 2025, on Hyperliquid someone opened a big short, then pulled out the margin to push themselves into liquidation; their position ended up entering the platform treasury. At the same time, the spot surged by about 250%, and the treasury’s floating loss was over $10 million. The validators changed the oracle price, and only then were things settled at a fixed price. Outflows over the next three hours were about 140 million U. This isn’t the funding rate grinding slowly—it’s that shallow order books broke the liquidation mechanism itself. There’s no customer service on-chain to reverse it for you.

On October 2nd, it was still lit on both sides—everything churned together.

The funding rate gap for SAND between Bybit and BloFin: annualized about 2100%, with positions around 39 million. On Hyperliquid, SAND’s hourly funding rate reached -0.29%—shorts were paying. PRL moved over to Lighter: annualized gap about 1970%, and the position was only around 3 million. QNT and BOT are both on-chain and on centralized exchanges; the price difference annualizes to over 300%. On Aster, some smaller coins had 8-hour funding read as low as -0.8%. Out of 178 Hyperliquid contracts, on that day 175 funding rates were positive—longs were paying.

Later I saw someone get knocked out of a position for several million on a market like this, so I never got hands-on again. It’s not that the profit wasn’t tempting. It’s that the seed has already been planted. Price can kill you a bit later; the funding rate can kill you tonight. A 1x isolated position only pushes the liquidation price further away. You think you’re waiting for it to drop back—while they’re waiting for your margin to hit zero. Switch to Hyperliquid: the cap is several times Binance’s, so you don’t even need to bet on whether it will smash or not.

Before you enter, only look at three things: look at both centralized and on-chain together. How deep is the spot liquidity. Whether the futures position is larger than the spot. Whether the fees are always skewed to one side, and they’re charged by the hour. If all three are glowing, then a sideways market is cost, not an opportunity. If you bought right, it only means you die a little more knowingly.

For funding rates, deposits, and depth in a market like this, I’ll break it down next. Don’t want to pay tuition again—follow me. Take fewer detours.

#Contract #Funding Rate #Perpetual Futures #Hyperliquid #Trading Recap #币圈避坑