Do stock perps predict Monday's open? What 24/7 trading on Tesla, Nvidia and other stocks says about the next opening price

The question

Wall Street is open for about 33 hours a week in the regular session. Stock perps on crypto venues trade for all 168. In the previous study we showed that Hyperliquid stock perps track the real stock price within a few hundredths of a percent while the stock market is open. But what about when it is closed?

If the perp price after hours and over the weekend reflects real news and real demand, it should predict the next official opening price. If it is mostly noise from a thin market, it should not. That is useful to know for anyone watching a perp on a Sunday and wondering what Monday will bring.

How we tested it

For eight large stocks – Tesla, Nvidia, Apple, Microsoft, Amazon, Meta, Alphabet and Coinbase – we took every trading day from 9 April to 2 November 2026: 1,144 openings, of which 208 were Monday openings after a normal weekend.

For each opening we measured the gap: the official opening price compared with the previous official close. Then, at each hour of the night or weekend, we looked at where the perp was trading relative to that previous close, and asked two questions:

1. Did the perp get the direction right? (We count only gaps larger than 0.25%, because smaller ones are noise.)
2. Did it reduce the error? We compare the perp's implied gap with the simplest possible guess – "the stock will open where it closed" – and measure how much smaller the average error is.

Weeknights: the opening gap is learned step by step

On weekday nights the perp price becomes more informative hour by hour.

• At 4 pm, when the regular session closes, the perp has no extra information: it matches the closing price.
• By 8 pm, the end of after-hours trading, it gets the direction right 66% of the time and cuts the error by 18%. This is when earnings and other after-hours news are absorbed.
• Through the night (8 pm – 4 am), when the US market is closed and the perp trades alongside Asian markets, its accuracy improves slowly: 79% direction at 4 am, with a 32% smaller error.
• From 4 am, when pre-market trading starts, it quickly converges: by 9 am the direction is right 94% of the time and the error is two-thirds smaller.

Weekends: mostly noise until Sunday evening

Weekends look completely different. From Friday's close until Sunday afternoon, the perp price had almost no relationship with Monday's opening. Its direction was right only 50–60% of the time, and using it would not have reduced the forecast error at all – at times it made the error slightly worse than simply assuming "no change".

The turning point is Sunday at 6 pm New York time, when US stock index futures reopen on CME and markets in Asia begin their Monday. From then on the perp quickly catches up: by Sunday 8 pm its direction is right 69% of the time, by Monday 4 am 83%, and by Monday 8 am 92%.

There is a simple reason. With the stock market and the major futures markets closed, weekend perp trading has no reference prices to arbitrage against and relatively little news to price. As we showed in the tracking study, weekends account for a third of all hours but only about 6% of stock perps' price movement. The moves that do happen are small and do not reliably carry over to Monday.

Big gaps are where perps help most

When stocks opened far from their previous close – a gap of more than 2%, which happened on 162 of the 1,144 openings, usually because of earnings or major news – the perp was especially useful. At 4 am New York time it had the direction right 91% of the time and the typical error was about half the size of the "no change" guess (1.5 points instead of 2.7). Most of these big gaps follow after-hours news, which the perps price in that same evening.

What this means

1. On weeknights, stock perps are a useful early indicator of the next opening price, and become more accurate through the night and pre-market.
2. On weekends, perp moves on Saturday and Sunday afternoon have not predicted Monday's open. Treat them with caution.
3. The information returns on Sunday evening, when US futures and Asian markets reopen.
4. Perps are most informative exactly when it matters most: after big news and before large opening gaps.

The study covers about seven months of data for eight very liquid stocks. Results may differ for smaller stocks, in a crash, or as stock perps attract more weekend trading. For live levels on these markets, see our Nasdaq-100, Tesla and Nvidia pages.