Spot, options, lending, futures. We borrowed all of it from TradFi and brought it on-chain. But a position with no expiry and no settlement, one that just trades around an index and pays funding to stay there, that didn't exist before crypto.
With $85.3 trillion in volume on CEXs last year alone, and the DEX-to-CEX ratio [growing to 11.7%](https://www.coingecko.com/research/publications/dex-to-cex-ratio#:~:text=What Is the Ratio of DEX to CEX Perps Trading,gap with CEXs this year.) by November 2025, perpetual markets are among the most successful use cases in our space. They're also the major financial instrument actually native to crypto.
And still, we're not running them properly (yet).
What’s actually broken
Traders know how to read a loss. You can usually tell if the setup was off, the timing was wrong, or the thesis just didn't hold. What's harder to account for (and fix) is when the market moved as expected, but the outcome didn't behave predictably.
What follows is where things break: five places where the venue decides how your trade ends, not the market. That's an infrastructure problem.
We think that's a problem worth solving, and that Solana is where it gets fixed.
The platform breaks when you need it most
Speed matters, but on its own it isn't enough. Ask a trader how 25 milliseconds versus 60 changes their PnL, and most will shrug. Ask them about the impact of their trading venue going down while the market is crashing, and you'll get a very different answer.
In July 2025, Hyperliquid went dark for about 37 minutes. A spike in traffic overwhelmed order delivery to its nodes, and the venue stopped responding. Traders sat on leveraged positions, suddenly unable to place an order, fire a stop, or add margin while the market kept moving without them. All of this while the status page showed 100% uptime.
This is far from a DEX-only problem. During the crash on October 10, several major exchanges experienced outages, delays, and trading halts at the worst possible moment, including Binance, Coinbase, and Robinhood.
What makes this so damaging is timing: venues fall over right when volume spikes and price moves fast, which is the exact window you need to add margin, cut, or get a stop in.
That's not bad luck, it's load: uptime measured over quiet days doesn't say much. What matters is whether you can act when it counts.
Force-closed, winning or not
On October 10, 2025, crypto saw the largest liquidation event in its history. Roughly $19 billion gone in a day, across more than a million accounts. Months later, no one's offered a clear account of what actually caused it. When execution breaks, the system rarely owes you an explanation.
Cascades like that aren't new in our space, and they all run the same playbook: price moves fast, leverage unwinds, forced selling feeds on itself. But the cascade isn't really the story. Two mechanisms decide who actually gets hurt, and neither has much to do with whether you called the move:
- First, auto-deleveraging. When liquidations come in faster than a venue can absorb, as a last resort, it closes other positions to cover the gap. Picture the market dumping while you're short with size. The longs on the other side blow through their margin faster than the insurance fund can handle, so the venue reaches for the profitable shorts, yours included, and force-closes them to fill the hole. More often than not, it's the hedged or profitable positions that get pulled in. You read the charts right, covered your risk, and the exchange unwinds the leg that was protecting you to balance its own book.
- Second, the price you're closed at. Plenty of traders get liquidated at a level the market never touched, because the number your position is judged against is the venue's internal mark, not the open market. That candle never actually touched your stop, but the number on the DEX did. And that was all it took.
Losing a bad trade can happen. Losing a good one to the platform you're trading on shouldn't.
The fix is a process bound by rules you can see in advance. Clear and predictable, rather than discretionary.
Front-running and MEV
There's a subtler version of the same pain point. This time it's the price you actually get filled at, not the price you're judged on. This one usually doesn't make headlines. It shows up as slippage you can't quite explain, fills a tick worse than they should be, a stop taken right before the move that justified it.
You trade alone in a crowded room. Between the moment you fire an order and the moment it fills, anyone sitting in the order flow can see what's coming and move first: validators, bots, the market makers quoting against you.
We could argue that these actors are part of the exchange’s machinery, and in a sense they are. Still, each one plays its own game. The cost is real even when it's invisible, and there's a second cost that shows up in no number: trust.
Traders assume they're being picked off with no way to confirm it or rule it out. A faster matching path doesn't fix this. Fair and deterministic ordering does, so the edge those actors rely on stops existing in the first place.
Priced on a feed you can't see
Positions are marked, funded, and liquidated against the venue's price feed, not the open market. On deep crypto pairs like BTC or SOL-USDC, that feed is liquid and tightly arbitraged, so it rarely drifts far from reality.
The problem shows up the moment a platform lists anything thinner: long-tail tokens, commodities, forex, equities, RWAs. There, feeds go stale, lag the real market, or stop updating on weekends, and you end up paying funding you don't owe or getting liquidated on a level that never traded anywhere real.
As internet capital markets keep growing, this matters more over time. The general problem is opacity: which sources the feed draws from, how it's calculated, and how often it updates. And partly the quality of the oracle infrastructure itself: stale prices, accidental drifts from reality, feeds that can't keep up with the market they're supposed to track. The fix is visibility: a price you can watch being formed and check against the market before it decides whether you keep your position.
If the book's too thin, none of it matters
Before fairness, before speed, every serious trader screens for one thing first: liquidity. Can I get in and out at size without moving the price against myself?
A thin book is a hard ceiling. The matching can be clean, the uptime perfect, the feed honest, and none of it helps if pushing real size through costs you on the way in and again on the way out. This is why depth is the first filter for anyone trading meaningful volume, and everything else comes second. Depth is also self-reinforcing: liquidity pools where liquidity already is, so thin venues stay thin and deep ones pull further ahead.
So, where does a book like that get built? Increasingly, the answer is Solana. On-chain SOL-USDC spot markets are now matching Binance and OKX on price. The performance that used to live only on CEXs is showing up on-chain, on the network that already has the users, the activity, and the design to support it.
It all comes down to execution
Liquidations that close the wrong positions, front-running, outages, bad feeds, thin books. They're five faces of the same issue: the quality of execution, and whether you can trust it.
That's what perp DEXs will compete on: fair and predictable ordering, reliability when the market moves fastest, and outcomes that match your trade. And that's the gap we built Parasol to close: Solana's trading engine to accelerate perpetual markets. Trustless by design via ZK, transparent end to end.
Around that sits the market structure serious traders expect: the order types a real desk needs from day one, latency that holds steady when workload spikes, and a price drawn from multiple sources instead of a single point of failure.
So the only variable left in your PnL is whether you were right, or not.
