Finora AI Search markets
Language: English
Open bot

Is negative funding bullish or bearish? What 90 days of Binance data showed

A negative funding rate means shorts are paying longs. Funding is the payment that keeps a perpetual future’s price close to spot, and it turns negative when the perpetual trades far enough below spot. In 90 days of Binance USDT-M perpetuals, extreme negative funding (−0.10 percent per 8 hours or deeper) was neither a bullish signal nor a reliable bearish one: price was higher 24 hours later in 43 of 100 cases, against 48 after normal funding. What changed was the size of the move, a median 24-hour range of 13.6 percent against 7.1.

On this page (7 sections)

What negative funding tells you

A perpetual future never expires, so the exchange keeps its price near spot with funding, a payment between longs and shorts at set times. When the perpetual trades far enough below the spot index to outweigh a small built-in interest component, the rate turns negative and shorts pay longs. The payment is the position’s value times the rate, and only positions open at the funding time take part.

The rate shows that sellers are pushing the perpetual under spot and paying to stay short, but not why: some bet on a fall, others hedge coins they hold elsewhere and lose little overall in a rally. Each contract also has a cap and floor per payment: on November 7 Binance listed 0.30 percent each way for BTC and ETH and 2 percent for PROM and LSK, so readings near −2 percent only show up on smaller coins with the wide band.

Many contracts don’t pay every 8 hours

At about 17:00 UTC on November 7, 92 Binance USDT-M coin perpetuals had traded more than 10 million USDT in 24 hours: 34 paid funding every 8 hours, 57 every 4 hours and one, HUMA, every hour. The biggest contracts were split. Of the 31 above 50 million USDT, 16 paid every 8 hours, BTC and ETH among them, and 15 every 4 hours. COTI was already paying hourly when our data starts, until September 1. All seven switches to hourly payments we found after that, on six coins, came right after a payment at the −2 percent floor, and LSK then paid every hour for almost 12 days.

The rate you see is per payment, so the interval sets the daily cost. On a 10,000 USDT short at −0.10 percent, each payment is 10 USDT: 30 USDT a day on an 8-hour contract, 60 on a 4-hour one and 240 on an hourly one. On September 14, LSK’s 24 hourly rates added up to −16.2 percent. To compare contracts we convert each rate to 8 hours, rate × 8 ÷ interval in hours, so −0.05 percent every 4 hours counts as −0.10 percent. The funding rates page shows each contract’s interval.

What followed extreme negative funding

The control group is the normal band, 0 to +0.01 percent per 8 hours, which most contracts pay while the perpetual trades close to spot.

Funding per 8 hoursEvents (coins)Higher after 24h, of 100Median 24h high-to-low rangeMoved 10 percent within 24h, of 100
−0.10 percent or deeper439 (49)4313.6 percent52
−0.03 to −0.10 percent545 (74)459.8 percent36
0 to +0.01 percent (normal)9,120 (120)487.1 percent19
Normal, same 49 coins3,321 (49)477.8 percent25
+0.10 percent or higher325 (33)4212.3 percent48

The table skips 0 to −0.03 percent. Every negative reading on BTC, ETH and SOL in these 90 days fell in that band, so the extreme results above come from smaller coins.

A day later price was higher after 43 of 100 extreme readings, against 48 after normal ones. The median change was −1.5 percent and the average +0.6 percent, lifted by a few huge squeezes. About half the events traded at least 10 percent away from the start within the day: three in ten reached 10 percent above it and three in ten 10 percent below, with 7 of 100 doing both. PROM’s payment at the −2 percent floor at 04:00 UTC on August 11 is one case. Price first fell about 15 percent, then traded 61 percent above the start and closed the day 26 percent higher.

The same 49 coins had a 7.8 percent median range on their normal-funding days, so jumpy coins alone don’t explain it. Extreme positive funding, with longs paying shorts, came out almost the same: 42 of 100 higher and a 12.3 percent median range.

A few coins carry much of the result

ACE and COTI produced 38 extreme events each and LSK 34, so three coins account for a quarter of the 439 and ten coins for close to six in ten. Fifty-six events happened while the contract was paying hourly, all on seven coins. To check whether they drive the result, we reran the numbers:

  • Without the three busiest coins: 41 of 100 higher, median range 13.7 percent.
  • Without the ten busiest: 41 of 100 higher, 13.6 percent.
  • Without the seven coins that paid hourly: 40 of 100 higher, 13.0 percent.
  • Without August 8 to 14, the ranking week: 44 of 100 higher, 12.8 percent.
  • With every coin weighted once: 39 of 100 higher, and an average change of −1.6 percent.

In every version the median range stayed near twice the normal 7.1 percent, while the gap in higher closes was less stable. We also redrew the sample 2,000 times, picking whole coins at random, and in 95 of 100 draws normal funding led by about 1 to 9 points. Without the ranking week that range ran from about 9 points one way to 2 the other, so one 90-day window gives no reason to treat extreme negative funding as a sell signal.

What a short squeeze needs

A squeeze starts when price rises far enough that short stops fill and shorts get liquidated, and those forced buy orders push price higher still. Negative funding shows that shorts are paying to stay in and that those without a hedge can be forced out by a rally, but price has to start rising first and the rate doesn’t say when.

On LSK, funding was negative at every payment from 04:00 UTC on September 11 to the top of a three-day run from about 0.10 to a 2.37 wick at 03:00 UTC on September 13. Its extreme readings at 12:00 UTC on September 10, 11 and 12 were followed a day later by gains of 18, 67 and 290 percent, the last the biggest 24-hour rise in the sample. Its payment at the −2 percent floor came at 12:00 on September 13, nine hours after the top, with price already back near 0.83. Over the next 24 hours LSK traded 34 percent higher and 21 percent lower before closing 13 percent down, then lost another 48 percent the following day while shorts kept paying every hour.

At the median extreme reading, −0.24 percent per 8 hours, a long collects about 0.7 percent a day, small next to a median 24-hour range of 13.6 percent. A long on such a coin needs the same structure as anywhere else: in a downtrend, a close above the last confirmed swing high, then the pullback after the first bullish break of structure, as in the BOS and CHoCH guide. With ranges twice as wide, the buffer past the invalidation level should come from that coin’s own recent wicks. A wider stop means a smaller position for the same loss and less usable leverage, since the liquidation price has to sit beyond the stop (see leverage and liquidation).

Where Finora shows funding

Finora’s chart reads of coins with a Binance perpetual carry a Flow line with open interest, the coin’s current funding rate and the taker buy/sell balance. The rate is per payment, and the Flow line doesn’t show the interval, so check it on the funding rates page, or in the exchange’s contract details for smaller coins, before turning the rate into a daily cost. For direction, use the trend-flip level in the read’s chart table, the close that would turn the trend on that timeframe.

How we measured

  • Window: funding payments from 00:00 UTC on August 8 to 00:00 UTC on November 6, 2026 (90 days), with 1h candles to November 7.
  • Universe: Binance USDT-M perpetuals on crypto coins without stablecoins, ranked by average daily volume over August 8 to 14 and cut at 120 (the last traded about 11 million USDT a day). That week is inside the window, so coins that got busy during it, PROM and LSK among them, are over-represented (the list above shows the result without that week’s events, though the coins it picked stay in). Delisted contracts still in Binance’s contract list were included (two made the cut), but those dropped from it are missing, a survivorship bias toward coins still trading.
  • Rates: each payment’s interval is the gap to the previous payment, and its rate is converted to 8 hours. A coin counts once per 24 hours in each band, the first payment winning.
  • Outcomes: the start is the open of the 1h candle at the funding time. Higher means the close 24 hours later is above it, the range is the 24-hour high minus low over the start, and a 10 percent move is a high or low that far from the start within those 24 candles.
  • Limits: one exchange and one window, no fees or slippage, and events on different coins overlap in time, so a market-wide day counts several times.
  • Reproduce: Binance’s public futures endpoints exchangeInfo, klines and fundingRate, plus fundingInfo and the 24-hour ticker for the November 7 count.

Today’s trend and key levels: Lisk (LSK) · Bitcoin (BTC) · Ethereum (ETH)