Why does Sezga put news, liquidations and whales in one feed?

Sezga's event feed is a single list that puts news, Binance and Upbit announcements, large liquidations, sudden price moves and large whale positions on the same timeline. The aim is to see what else happened just before and after a price move without switching between separate lists.

In crypto markets, events rarely arrive alone. A news story moves the price, the price move triggers liquidations, a large trader closes a position. Following these pieces separately loses both the order and the context. This article explains which events enter the feed and under what conditions, how noise is filtered out and how to read the feed.

Which events enter the feed?

Each event type has its own condition for entering the feed:

A liquidation wave event also states which side (long or short) dominated. The whale accounts are roughly 60 of the most profitable large accounts, selected every hour from the leaderboard of Hyperliquid, a futures exchange whose positions are publicly visible on-chain.

Why one timeline instead of separate lists?

In separate lists each type of data makes sense on its own, but the order between them is lost. A single timeline provides three things:

1. The order is visible. Seeing whether a liquidation wave came before or after a news story helps you think about what started the move.
2. Context forms by itself. Seeing a listing announcement for the same coin next to a large sudden move in an altcoin is quicker than searching separately for a possible explanation.
3. Comparison becomes possible. Event memory measures the price of the related pair 5 minutes, 1 hour and 24 hours after each event. Because the same measurement is applied to news, announcements, liquidations, sudden moves and whale events alike, what the price did afterwards can be compared on the same scale.

One more detail matters for reading the order correctly: the time an event was published and the time Sezga saw it are kept separately. That way, a story that reached a source late does not blur whether it was published before or after the price move.

How is noise filtered out?

A single feed that takes in everything becomes unreadable. So a few rules apply:

• Thresholds: Liquidations of $50,000 or more are recorded individually, but only those of $1 million or more enter the feed. The long/short liquidation total for the last hour is shown separately.
• Repeat suppression: No second sudden move alert is generated for the same pair within 30 minutes.
• Clustering: Repeats of the same story across different sites are grouped under a single entry.
• Severity level: Every event gets a level of Critical, Important or Info. The level is rule-based; for example, news of a hack or a withdrawal halt and a liquidation of $5 million or more are Critical, while news about the Fed is Important.
• Language filter: Chinese and Japanese flash news feeds are also collected, but because their headlines are not translated, they appear only when the language filter is turned on.
• Coin tags: Major coins mentioned in a story are tagged, which makes it easier to match the story with price events in the same coin.

How should the feed be read?

Consider a hypothetical sequence:

This sequence may suggest that the drop followed the Fed story and was accelerated by liquidations. Even so, three points need attention:

• Happening at the same time is not causing. The story and the drop may have occurred in the same minutes, but the drop may have had a different cause.
• Liquidation totals are a lower bound. Binance sends at most one order per symbol per second in this stream, so the real total can be higher than what is shown, and Sezga says so explicitly.
• The feed describes the past; it does not predict the future. The event memory summary of "what the price did after similar events in the past" is a statistic, not a buy or sell signal. For events where no price could be found, the result is recorded as "could not be measured".

Frequently asked questions

Which events appear in Sezga's event feed?

News, Binance and Upbit announcements, liquidations of $1 million or more, liquidation waves, sudden price moves, whale positions of $2 million or more on Hyperliquid, and large changes in open interest and stablecoin supply. Numeric events enter the feed once they cross their own thresholds.

Why doesn't every liquidation appear in the event feed?

Liquidations of $50,000 or more are recorded individually, but only those of $1 million or more enter the feed. The long/short liquidation total for the last hour is shown separately.

Are a news story and a price move that appear together connected?

They may be, but appearing together does not prove it. The single timeline shows the order; to understand the cause, you need to look at the content of the event, other data and similar cases in event memory.

Key takeaways

• Sezga's event feed puts news, announcements, liquidations, sudden moves and whale positions on a single timeline.
• Numeric events have their own thresholds for entering the feed, for example $1 million for liquidations and $2 million for whale positions.
• Repeat suppression, news clustering, severity levels and the language filter keep the feed readable.
• Publication time and the time Sezga saw an event are kept separately, and the price is measured 5 minutes, 1 hour and 24 hours after each event.
• Do not assume a cause-and-effect link between events that appear at the same time.