Hyperliquid's Weekend Edge, TradeXYZ Rumors & Why 97% of Tokens Fail

Hyperliquid's RWA volume hit an all-time high of 63 percent of total perpetual volume in the prior week, while pure crypto volume in July 2026 ran roughly 50 percent below the same month a year earlier. SK Hynix became the first non-crypto asset to overtake Bitcoin in trading volume on Hyperliquid, surpassing crude oil and silver. Research conducted with the Hyperliquid Policy Center found that when Hyperliquid moved more than 100 basis points during traditional market closures, it was directionally correct 95 percent of the time in predicting whether a market would open positive or negative, and it cut the median reopening error between Friday close and first traditional market print by more than 50 percent. Pushback from incumbents against weekend trading was attributed to settlement and banking hour requirements rather than principled opposition to 24/7 markets.

During the transition from Hyperliquid's internal weekend regime to the pre-market regime for SK Hynix, a single trader printed a manipulated price on the thin pre-market book that dropped the reference price 30 percent. Because Hyperliquid's Oracle stitches together pre-market, market, and post-market regimes and the mark price derives from that Oracle combined with a smoothing function and local midpoint, the print triggered liquidations that cascaded prices lower. TradeXYZ chose to refund all adversely liquidated traders, paying out over one million dollars. Shonda attributed fault to neither party but described the refund as the right decision because it reinforces trader trust and helps explain why TradeXYZ markets outperform those of other deployers on the platform.

Shonda dismissed rumors of TradeXYZ leaving Hyperliquid as unfounded, arguing it would require rebuilding all infrastructure from scratch and would amount to reputational suicide. She noted it is more plausible that Hyperliquid could in-house TradeXYZ's function than the reverse. An unconfirmed open secret holds that TradeXYZ raised capital from Paradigm, though Shonda stated she had zero confirmed information. She argued that TradeXYZ's minimal communications make rumors easy to spread and that greater transparency would be warranted given the firm's importance to the platform. If TradeXYZ were to launch its own token or equity, Shonda warned it would be bearish for Hype by fragmenting liquidity unless revenues continued funding Hype buybacks. She also noted that HIP-3 assets carry a 50-50 revenue split with market makers, and predicted that within one to two years, 90 percent of Hyperliquid volume could be real-world assets, meaning roughly half of revenue would flow to market makers rather than the protocol.

Carlos set out to quantify the effect of buybacks on token performance and surfaced a broader finding instead. Less than five percent of tokens outperform Bitcoin over a long window, and only 1.7 percent of tokens with at least 24 months of history outperform Bitcoin. The median token lost 97 percent from the month it first qualified above a 50 million dollar market cap, a result that held with robustness checks at 100 million and 250 million dollars. Using a vintage analysis framework borrowed from credit, 86 percent of the 2024 token class had lost at least 90 percent of entry value by month 24, compared with 70 percent of the 2021 class and 18 percent of the 2020 class. The dataset covered L1s, DeFi, governance, and gaming tokens rather than primarily meme coins, making the finding more significant. The deterioration was attributed to venture capital capturing private upside before token generation events and launchpad platforms creating hundreds of thousands of tokens without sufficient liquidity.

Exchange tokens were 16 times over-represented among long-window Bitcoin outperformers, making up 32 percent of that group despite being only two percent of the universe. Recurring fee-funded buybacks and burns were identified as the defining feature, with Hyperliquid cited as a confirmed example. High momentum and high volatility tokens subsequently underperformed on both one-month and three-month horizons, and trailing 30-day outperformance preceded forward 30-day underperformance, indicating alpha is mean reverting. Diversification across many tokens was described as providing little protection due to high correlation, making a concentrated position in one or two high-conviction names alongside Bitcoin more sensible than holding 10 to 15 tokens.

Only approximately one percent of tokens launched from 2023 onward ever delivered a 10x return in public markets. The only three non-meme tokens from that cohort to achieve it were Hyperliquid, Virtuals, and Mux Finance, with Hyperliquid the most successful by a significant margin and Mux noted as underperforming at the time of recording. Most infrastructure tokens including alternative L1s and L2s are expected to go to zero, and only roughly three blockchain ecosystems are considered relevant today. A small group of around 20 DeFi tokens are described as having genuine fundamentals but trading at discounted valuations due to poor disclosure and weak sentiment. USDC revenue begins accruing to Hyperliquid on August 26, with the protocol receiving 90 percent of yield generated and Coinbase retaining the remainder.