Best Time to Trade Crypto and Perpetual Futures: Hours, Sessions and Days, Measured (2026)

Daylight saving shifts the US and European windows, not Asia's. Japan, Singapore and Hong Kong have no daylight saving. Europe's clocks go back on October 25, 2026 and the US clocks on November 1, 2026. From then until March, everything tied to New York happens one hour later in UTC: the 8:30 a.m. data releases at 13:30 UTC, the 9:30 a.m. stock open at 14:30 UTC (23:30 in Tokyo), and FOMC statements at 19:00 UTC (04:00 the next morning in Tokyo). If you trade from Asia, your "busy window" moves an hour later in winter.

4. When is the best time to trade crypto in my time zone?

The busy window is anchored to the New York morning, so it is easiest to remember in New York time: 9 a.m. to 1 p.m. ET, all year. Converted:

In the week between October 25 and November 1, 2026 Europe is already on winter time while the US is not, so London and Berlin see the window one hour earlier (1–5 p.m. and 2–6 p.m.). The winter column assumes the window keeps its New York anchor, as the US data releases and the stock open do.

5. What is the best day of the week to trade crypto? Weekends: volume, ranges, spreads and the Monday open

Short answer: Tuesday to Friday. Crypto trades on Saturday and Sunday, but far fewer people do. Counting only complete UTC days in our window, with the average open-to-close BTC move for each weekday:

Weekend days averaged $1.16B against $3.35B on weekdays, about 34%. ETH was less extreme at 53%. The busy US-morning hours lose the most: the median 13:00 UTC hour fell from $242M on weekdays to $41M at weekends, while the late-evening hours barely changed. Friday topped this sample, but two of its four days carried scheduled US data (the September 4 jobs report and the September 11 CPI release), so treat the ordering of weekdays as noise; the weekday/weekend gap is the robust result.

The price column is there because "best day to buy" is a common question, and it shows why a month of data can't answer it. Monday's +2.03% average comes mostly from one +6.7% day, and all four Tuesdays fell, but four observations per weekday is far too few to call a pattern. Studies of "day-of-week effects" in crypto disagree with each other for the same reason. Choose the day for liquidity, not for a direction.

What this means in practice: a weekend order book is thinner at every price level, so the same market order moves the price more, stop-losses are more likely to be hit by a single wick, and a liquidation cascade has less resting liquidity to absorb it. Funding keeps running at weekends too. Sunday evening UTC, when Asia's Monday begins, is when volume starts to return. The longer windows below show the same gap.

Weekends over 90 days and 52 weeks: quieter, but not for altcoins

The 30-day table is short, so we also pulled 2,160 hourly candles covering 90 complete UTC days (June 27 to September 24, 2026: 64 weekdays and 26 weekend days) and 364 daily candles covering 52 full weeks (September 27, 2025 to September 25, 2026). Over 90 days BTC perps on Hyperliquid averaged $1.08B a weekend day against $2.75B a weekday (39%), and the average daily high-low range was 1.58% against 3.40%. Over 52 weeks weekend volume was 47% of the weekday level, and only 3.8% of weekend days closed more than 3% from their open, against 16.5% of weekdays.

Hyperliquid API candleSnapshot (1h), fetched September 25–26, 2026. Volume = contracts × average of open and close. “Largest 1-hour move” is the largest open-to-close change of a single hourly candle.

The further you go from Bitcoin, the less the weekend calms things down: SOL kept 63% of its volume and nearly 80% of its range, and its largest hourly move came on Saturday, August 22, 2026, when it traded from 100.09 down to 87.69 within the 05:00 UTC hour. Thin weekends also make each dollar traded move the price more; the worst weekend day in 52 weeks, Saturday, January 31, 2026, took BTC down 6.54% open to close.

Direction: over 52 weeks BTC closed up on 52.9% of weekend days and 46.9% of weekdays, Sunday averaged +0.21%, and no day of the week passed a basic significance test. “Sunday dumps” did not show up in the data.

Do spreads widen at weekends? Measured on Hyperliquid

We took Hyperliquid order-book snapshots on Saturday evening, September 26, 2026 (22:55–23:01 UTC, one of the quietest windows of the week) and compared them with the same measurement on Friday, September 25. “$1M market buy” is the average fill of a simulated $1M market buy against the book, compared with the mid price.

Hyperliquid API l2Book; Saturday figures are medians of three to six snapshots. 1 bp = 0.01%. Single snapshots move a lot from minute to minute.

• On crypto majors the spread did not widen. BTC sat at one tick on both days (0.119 bps), with about $23–29M resting within 0.1% of mid on each side on Saturday evening. What changes is flow: fewer orders pass through the book, so a large order or a liquidation meets fewer counterparties.
• On stock and commodity perps it did. With their home markets closed, a $1M buy of xyz:SP500 cost about 2.7 times Friday’s figure and xyz:GOLD about 1.7 times.
• Stops: Hyperliquid triggers TP/SL on the mark price, which references external exchanges, so one wick in a thin book is less likely to fire a crypto stop. A market TP/SL accepts up to 10% slippage; a stop-limit set at the trigger can be skipped in a fast move. Place weekend stops outside a normal weekend range (BTC averaged 2.38% a weekend day over 52 weeks).

The Monday open and the CME gap

Crypto itself does not gap. BTC trades continuously, so the “gap” is just the weekend’s move. From Friday 21:00 UTC to Sunday 22:00 UTC over 12 weekends, BTC moved 0.52% on average (at most 1.64%), ETH 1.33% (at most 3.52%) and SOL 2.13% (at most 4.84%). Weekend activity picks up from Sunday 20:00 UTC (hourly BTC volume roughly doubled versus the same Saturday hours) as futures-linked markets reopen at 22:00 UTC and Monday starts in Asia. The CME gap is fading: CME now lists its crypto futures and options on Globex 24/7 except Saturday 2:00–4:00 a.m. CT and a 4:00–4:02 p.m. CT pause Monday to Friday, so new weekend gaps of the old kind should no longer form.

The real Monday event on a perp DEX is on stock and commodity perps such as trade.xyz’s xyz:NVDA or xyz:SP500. While their home markets are closed they trade on an internal price inside discovery bounds, then snap back to the external price at the reopen: NVDA’s Monday 00:00 UTC reopen hour moved 0.49% on average (at most 1.25%) over 13 weekends, the S&P 500 perp 0.14% (at most 0.32%). Keep leverage low on anything whose home market is shut. Details in our trade.xyz guide.

Funding does not take the weekend off: Hyperliquid charges it every hour, and over 90 days the BTC rate averaged about the same on weekends as on weekdays. If you trade small weekend moves, fees matter more than usual: Hyperliquid with code HOLYGRAIL cuts 4% off your first $25M of volume, and Lighter’s Standard account charges 0% taker (open Lighter).

6. Is the best time different for BTC, ETH and altcoins?

We ran the same 30-day window for ETH, SOL and HYPE (Hyperliquid's own token). Every one of them peaked in the same hour:

Source: api.hyperliquid.xyz candleSnapshot, 1h, August 26 – September 25, 2026 (SOL and HYPE fetched September 27, 2026). Our calculation.

Two differences matter. Altcoins are less concentrated in the US morning: SOL did 33.4% of its volume in 12:00–17:00 UTC against BTC's 38.6%, with relatively more in the Asian hours. And altcoin weekends are less dead: ETH, SOL and HYPE kept about half their weekday volume, BTC only a third. But altcoin books are much thinner to begin with: SOL traded about $0.32B a day on Hyperliquid against $2.76B for BTC. A size that is invisible in BTC at 04:00 UTC can move an altcoin. So the timing rule matters more for altcoins, not less. Trade them in the overlap, and use limit orders outside it. For Solana-native venues, see best Solana perp DEXs; for current volume and funding on every Hyperliquid market, see our Hyperliquid markets data.

7. Is there a best time of day to buy crypto?

This is where most guides make claims they can't support. We averaged the one-hour BTC return (close ÷ open − 1) for each hour of the day over the 30 days. The three best and three worst hours:

It looks like a pattern, but it isn't one you can bank on. BTC rose 6.5% over the window, so the average hour was slightly positive anyway. Each hour has only 30 observations, and none of them, not even 13:00 UTC, passed a basic significance test (a t-statistic of 2 or more). We also checked when the day's low was printed: the 12:00–17:00 UTC block held the low on 8 of 29 days and the high on 9, only slightly more than the 6 you would expect by chance for five hours of the day. The volatile hours make both the lows and the highs.

So the honest answer to "best time to buy crypto", "cheapest time to buy" or "should I buy at night" is: there is no reliably cheap hour. What you can control is execution cost. Buy with limit orders during the deep hours, split larger buys into pieces over time (a TWAP order does this automatically), and don't send market orders into a thin weekend or late-night book.

8. When do CPI, jobs reports and FOMC move crypto?

The biggest hours of the month are not random. Two of the four busiest BTC hours in our window line up exactly with scheduled US releases, and the jobs report stands out in ETH:

• September 11, 12:00 UTC — the August CPI report (released 8:30 a.m. ET = 12:30 UTC): $1.00B of BTC in one hour, range 2.75%. ETH's following hour traded $549M with a 6.2% range.
• September 16, 18:00 UTC — the FOMC statement (2:00 p.m. EDT), at which the Fed raised its target range by a quarter point: $931M, range 2.0%.
• September 4, 12:00 UTC — the August jobs report (8:30 a.m. ET): ETH $380M, range 3.9%.

A normal hour in the same window averaged $115M. The upcoming dates, from the BLS and Federal Reserve calendars:

Sources: bls.gov release schedules (CPI, Employment Situation, 8:30 AM ET) and federalreserve.gov FOMC calendar, checked September 25, 2026. Times after November 1 reflect US standard time. Schedules can change; check the official calendar the week before.

Around these releases spreads can widen for a few seconds, funding can jump, and leveraged positions on both sides get liquidated. If you don't trade the news itself, either be flat or cut leverage before the release and add stops that allow for a 2–3% hourly range.

9. When are funding payments made?

Funding is the other clock that matters for perps. It is paid between longs and shorts, not to the exchange, and when it is paid depends on the venue:

Sources: Hyperliquid and Lighter funding docs, Aster funding-rate doc, Aster and Binance fundingRate API history for BTCUSDT (September 23–25, 2026); edgeX public metadata API and Pacifica funding docs (checked September 27, 2026).

Why the interval matters for timing. On an 8-hour venue, a position open at the settlement instant pays or receives the whole interval, even if it was opened seconds earlier. Aster's docs even warn of a 15-second tolerance: open at 16:00:05 UTC and you may still be charged. On hourly venues the same mistake costs one-eighth as much, and funding accrues smoothly over the day instead of in three lumps.

What funding actually cost in September. Hyperliquid's baseline is 0.00125% per hour (0.01% per 8 hours). Over our 30-day window, 72% of BTC's 720 hourly rates sat exactly at that baseline, 93% were positive (longs paid shorts), and the highest was 0.009% in one hour. The 30-day total was 0.77%: a $10,000 BTC long held for the whole month paid about $77 of funding, more than 8 round trips of taker fees ($9.00 each at 0.045%). For ETH the total was 0.83%. Time-of-day matters less than holding time: every hour you hold is an hour of funding. Details in Hyperliquid funding explained and crypto funding rates explained.

10. Do trading costs change with the time of day?

Fees are fixed per trade: a $10,000 taker round trip on Hyperliquid costs $9.00 at the base tier (0.09%), or $8.64 with a referral code, whatever the hour. Lighter's Standard account charges nothing; its Premium account 0.028% taker. Only the implicit costs move with the clock. For BTC on Hyperliquid they are small for retail size even off-peak: in an order-book snapshot at 07:59 UTC on Friday, September 25 (one of the quieter hours), the spread was $1 (about 0.1 basis point) and the 20 price levels on each side, within about $20 of the mid, held roughly $5.1M of asks and $4.7M of bids. A $10,000 BTC order fills within a few dollars of the quote at almost any hour. Timing starts to matter for orders in the hundreds of thousands, for altcoins with thinner books, and at weekends, when the whole book has less behind it. Two practical rules follow: in thin hours, switch to limit orders (which also pay Hyperliquid's 0.015% maker fee instead of 0.045%), and size down. Fee details: Hyperliquid fees, Lighter fees, lowest-fee DEXs.

11. What is the best time for scalping, swing and news trading?

Scalping and day trading

Scalpers live on spread and depth, so trade 12:00–17:00 UTC on weekdays and avoid the minutes around 12:30 UTC data releases unless that is your strategy. Fees decide whether scalping works at all: a 0.1% target barely clears a 0.09% taker round trip on Hyperliquid, while on Lighter Standard it is all profit before slippage. Hyperliquid's advantage is depth when you size up. See best DEX for scalping.

Swing trading

Holding for days makes the hour of entry less important and funding more important. Enter when the book is deep and you have the day's information (the European morning or the US session), avoid adding size into the Saturday lull, and check the funding rate before holding a crowded position over a weekend. Crypto daily candles are usually drawn in UTC, so the daily close is 00:00 UTC (09:00 JST, 8 p.m. New York in summer).

News trading

The spikes in section 8 are the opportunity and the risk: a 2–6% hourly range within minutes, liquidations on both sides and funding that can jump. Keep leverage low, use limit orders or small market orders, and never place a stop so close that the release candle alone triggers it.

Large orders and long-term buying

If you need to move size, do it in the US morning and slice it. Hyperliquid has a built-in TWAP order that splits a parent order over time (see our TWAP and iceberg orders guide). Long-term buyers gain little from hunting the "right hour" (section 7); regular buys with limit orders in liquid hours keep execution costs down.

12. When should you trade stock perps?

Stock and index perps on DEXs trade around the clock, but the underlying market does not. Aster's documentation is explicit: its US stock perps trade normally during the regular session (9:30 a.m.–4:00 p.m. New York time); in pre-market, post-market and overnight hours, limit orders more than 5% more aggressive than the mark price are rejected and market orders fill only up to 5% from the mark. Overnight, at weekends and on US public holidays, an EWMA-smoothed mark price applies. For stock perps, the "best time" is simply the US cash session, 13:30–20:00 UTC (22:30–05:00 JST) until November 1 and 14:30–21:00 UTC afterwards. On Hyperliquid, stock perps come from builders under HIP-3 (for example trade.xyz) and follow the rules of each deployer.

13. How do you plan a trading session?

• Know your window. 9 a.m.–1 p.m. New York time: 13:00–17:00 UTC now, 14:00–18:00 UTC from November 1.
• Check the calendar every Monday. CPI, the jobs report and FOMC days are the busiest and most volatile hours of the month.
• Read the funding rate and interval before you enter. On 8-hour venues, avoid opening on the paying side just before 00:00, 08:00 or 16:00 UTC.
• Thin hours: limit orders, smaller size. 03:00–07:00 UTC and weekends had a fraction of the volume, and altcoins suffer most.
• Weekend positions: less liquidity to absorb a liquidation cascade. Use isolated margin and lower leverage (see risk management and position sizing).
• Log your trades with timestamps. After 50 trades, check which hours your strategy actually makes money in. Hyperliquid lets you export your trade history as CSV.

FAQ

Disclosure: HOLYGRAIL (Hyperliquid), 718610TD (Lighter) and DTLpLc (Aster) are our referral codes; we earn a share of referred accounts' fees at no extra cost to you. Volume, return and funding figures are our own calculation from Hyperliquid's public API for the dated window; past patterns do not guarantee future liquidity or prices. Not investment advice.