The $2 Million Trade That Sent Bitcoin to $123,000 on One Exchange — Here's Exactly How It Happened

On September 6, 2026, a single account with roughly $2 million of sell orders pushed the BTC perpetual on Bulk Exchange, a Solana-based perps DEX, to print as high as $123,221 — a 55% premium to the real Bitcoin price of about $79,600 on Binance at the same moment. No exchange got hacked. No smart contract was exploited. Every liquidation, auto-deleverage, and matched trade executed exactly as the code said it should. That's the uncomfortable part.

This is a full reconstruction of what happened, why Bulk's own risk engine turned a thin order book into a 45-second price spiral, what the team has said about it, and why the fix matters more than the incident itself.

TL;DR

• What: Between roughly 21:32 and 21:37 UTC on Sept 6, 2026 (Bulk's own statement frames the wider window as 14:15–15:15 UTC), a leveraged short on Bulk's BTC-PERP market got repeatedly force-closed into a nearly empty order book, dragging the mark price away from the external oracle and triggering a self-reinforcing liquidation → auto-deleverage (ADL) cascade.
• Root cause: Three design gaps compounding at once — no price band on market orders, a mark-price formula that leans on the local book under volatility, and liquidation/closeout logic not fully anchored to the external oracle.
• Damage: Independent on-chain analysis puts total forced liquidations near $3.01M (33.67 BTC, 28 accounts) plus $2.82M in ADL across 248 profitable accounts. Bulk's own post cites smaller figures — a gap worth scrutinizing (see below).
• Fix: Bulk says its next network upgrade anchors mark price more strongly to external data and decouples account equity/liquidation eligibility from instantaneous local order-book depth, with affected users to be reimbursed.

The Setup: A Thin Book on a New Exchange

Bulk Trade is a Solana-native perpetuals exchange built around an on-chain central limit order book, targeting sub-400ms settlement and using its own leaderless BFT consensus (BulkBFT) across validators rather than routing through a centralized matching server. The project raised an $8 million seed round in 2025 and has been running a public testnet since November 2025, building toward mainnet with an incentivized "Aura points" campaign ahead of a token launch.

That matters here because Bulk itself describes the period this incident happened in as a "liquidity-bootstrap phase" — plain language for: the book is thin because the exchange is new. A market that's traded only ~7 BTC in the previous hour and 43 minutes does not have the depth to absorb a $2 million market sell without the price falling off a cliff. That's not a Bulk-specific flaw on its own — every young order book has this problem. What turned thin liquidity into a five-figure-per-BTC price dislocation was what Bulk's engine did in response to that thin book.

Timeline: How a Short Position Became a $123K Print

Independent analysis published by trader @Rutto02 on X reconstructs the sequence at the trade level. Bulk's own statement, published by co-founder @junbug_sol, frames the same event with account-level detail. Piecing both together:

The whole dislocation, from first liquidation to full mean-reversion, ran roughly five and a half minutes.

Why It Happened: Three Flaws, One Feedback Loop

None of these three issues is unusual in isolation. Combined, they built a closed loop where the exchange's own defense mechanism became the attack surface.

1. No price band on market orders. A market order on Bulk had no cap on how far it could walk the book. When real depth ran out, a stale ask sitting at $115,115 — presumably left over from a much earlier, wider market — was filled at face value. Most mature exchanges (and even most immature ones) implement some form of circuit breaker or maximum-slippage guard on market orders precisely to prevent this.

2. A self-referencing mark price. Bulk calculated its mark price as the median of a premium-adjusted external oracle, the live book price, and a 30-second smoothed book price — but under high volatility, the formula reportedly leaned toward the raw book price. On a book only a few BTC deep, that means a $2 million sell order doesn't just move the tradable price, it moves the reference price used to calculate everyone else's liquidation threshold. Every short's liquidation trigger moved with the local chaos instead of staying pinned to a robust external reference.

3. Closeout logic tied to instantaneous local liquidity. Position equity and liquidation eligibility were evaluated using an estimated cost-to-close based on the live book, not the external market price. In a deep market that's a reasonable, conservative assumption. In a market that has just had its liquidity stripped out by the same event causing the liquidation, it means a temporary, self-inflicted depth gap directly worsens the accounts being forced to close — a mechanical accelerant for the cascade rather than a brake on it.

Put together: a large forced seller drained the book → the drained book distorted the mark price → the distorted mark price triggered more liquidations at worse levels → those liquidations found no real liquidity and printed against a stale, uncapped market order → the resulting print made the deficit larger than any single account's margin could cover → the exchange's auto-deleveraging mechanism then took the loss out of 248 unrelated, profitable traders' positions.

What Bulk Says Happened

In an official statement, Bulk co-founder @junbug_sol confirmed the mechanics and gave account-level figures: the trigger account generated roughly $1.5 million in sell volume across 400 fills and was partially liquidated eight times before the cascade proper began. A protective short-close order activated near a mark price of $81,450 at 14:36:45 UTC, converted into a reduce-only market buy, consumed the available ask liquidity, and landed on that resting ask at $115,115. Three more protective closes fired shortly after near $82,000–$82,500.

Per Bulk's numbers, open interest on the BTC market fell from about 26.21 BTC to 6.82 BTC over the event. The company states ADL covered roughly $48,480 of the resulting shortfall, with the protocol's risk vault absorbing about $66,149 that ADL couldn't recover — a combined ~$114,600. Across the full incident, Bulk cites approximately 26.55 BTC in forced liquidations and 17.69 BTC in matched ADL reductions.

The Numbers Gap Worth Asking About

This is where the story gets more interesting for anyone tracking Bulk closely. The independent trade-level reconstruction puts total liquidation damage at $3.01M across 33.67 BTC and 28 accounts, plus a separate $2.82M ADL event across 34.75 BTC and 248 accounts — figures roughly an order of magnitude larger than the ~$114,600 in ADL-plus-risk-vault losses in Bulk's own statement, even though the BTC volumes moved (33.67–34.75 BTC on one side, 26.55–17.69 BTC on the other) are in the same rough neighborhood.

Some of the gap is plausibly definitional — "shortfall not covered by ADL" is a narrower number than "total value liquidated," and denominating losses in BTC-at-time-of-trade versus USD-at-time-of-report will diverge sharply when prices ranged from $78,000 to $123,221 within the same event. But the two accounts of the same incident don't reconcile cleanly from the public information alone, and that gap is exactly the kind of detail a platform doing a post-mortem under public pressure should close explicitly rather than let independent analysts fill in.

The Fix: Two Structural Changes

Bulk's announced network upgrade targets the two mechanical root causes directly, not just the symptom:

• A mark price anchored to external data, not local depth. During the liquidity-bootstrap phase, mark price will lean more heavily on robust external market data, require a minimum depth threshold before local book inputs can move it, and cap how far mark can diverge from the oracle.
• Equity and liquidation eligibility decoupled from instantaneous local liquidity. Rather than valuing open positions against an estimated cost-to-close on the live book, the upgrade evaluates equity and liquidation eligibility against the external market price. A temporary depth gap — even a severe one — should no longer by itself convert into a liquidation trigger.

Bulk has also committed to reviewing every affected liquidation and ADL account and reimbursing eligible users, and states it has zero tolerance for deliberate manipulation of thin-liquidity conditions, without publicly naming the trigger account as a bad actor at time of writing.

What This Means Beyond Bulk

This incident is a textbook illustration of a problem every perpetuals venue with an on-chain, low-liquidity order book eventually runs into: mark price and liquidation logic that reference the same book they're supposed to protect create a feedback loop, not a safety mechanism. This is not a Bulk-specific defect — it's the recurring failure mode of the category, and Hyperliquid has hit variants of it repeatedly despite being the market leader:

• The JELLY incident (March 2025): attackers used a low-liquidity token's isolated spot feed to inflate its price roughly 8x with a relatively small outlay, manipulating the liquidation process on a large short and forcing Hyperliquid's validators into an emergency manual override of the oracle to avoid protocol insolvency.
• SPACEX-USDH oracle error (May 2026): a single bad print from one offchain data provider wiped $1.5M and liquidated 405 users across 1,393 positions in under 30 minutes on a thinly traded pre-IPO perpetual.
• SKHX flash crash (July 2026): an unusually small trade in the underlying stock during thin premarket liquidity produced a distorted print; Hyperliquid's oracle reflected it and triggered a cascade that briefly pushed the platform's 4-hour liquidation volume above Binance's.

The instructive contrast is that Hyperliquid's mark price is deliberately built as a median of several external CEX feeds plus its own book — specifically so one thin, isolated venue can't wick traders out on its own. Bulk's formula did something closer to the opposite at the worst possible moment: it's designed to lean toward the raw local book price precisely when volatility is highest, which is exactly when the local book is least trustworthy. Hyperliquid's incidents show that even a more conservative, oracle-weighted design doesn't fully eliminate this risk category; Bulk's incident shows what happens when the design leans the other way.

For traders, the practical takeaway from the independent analysis is blunt: sizing leverage to book depth, not to account equity, is the only defense that works before an exchange fixes its own mechanics. Shorting into an illiquid, self-referencing perp isn't arbitrage against the "true" price — it's standing inside the blast radius of the next cascade.

FAQ

Did Bulk Exchange get hacked? No. Every liquidation, matched trade, and auto-deleveraging event executed according to the platform's documented logic. The incident was a mechanical design flaw in how mark price and liquidation eligibility were calculated under thin liquidity, not an exploit of a bug or a breach of the protocol.

Will Bulk reimburse affected users? Bulk has stated it is reviewing every affected liquidation and ADL account and will reimburse eligible users for losses tied to the dislocation.

What caused the BTC price to print at $115,000+ on Bulk while Bitcoin traded near $79,600 everywhere else? A large forced seller drained the order book, which distorted Bulk's mark price (calculated partly from the local book), which triggered further liquidations that had to buy back short positions with no real ask-side liquidity available — eventually filling a stale resting ask because Bulk's market orders had no price cap.

What is Bulk Exchange (bulktrade.exchange)? Bulk is a decentralized perpetual futures exchange built on Solana, using an on-chain central limit order book and its own BulkBFT consensus mechanism to target sub-400ms trade settlement. It raised an $8 million seed round in 2025 and has been operating a public testnet ahead of a token launch.

Has this kind of thing happened on other perp DEXs? Yes — it's a recurring failure mode for low-liquidity, on-chain perpetuals, not unique to Bulk. Hyperliquid, currently the largest perp DEX, has had multiple related incidents: the March 2025 JELLY manipulation that forced an emergency oracle override, a May 2026 offchain-data error that wiped $1.5M on a thinly traded pre-IPO market, and a July 2026 SKHX cascade driven by thin liquidity in the underlying stock. Each involved a different specific trigger, but the same underlying pattern: thin liquidity plus leverage plus a pricing mechanism that can be distorted by the very conditions it's supposed to price around.

Sources: independent trade-level reconstruction by @Rutto02 on X (Sept 6, 2026); official incident statement by Bulk co-founder @junbug_sol on X (Sept 6, 2026); Bulk Trade project background, CoinLaunch; Bulk Trade overview, Wikiquote; Bulk Trade architecture notes, junbug's Substack; Bulk Trade guide, KuCoin; Hyperliquid liquidation mechanics, Hyperliquid Support; Hyperliquid JELLY incident summary, GitHub; Hyperliquid SPACEX-USDH oracle error, KuCoin; Hyperliquid SKHX flash crash, Cryptometer.