The conclusion first: the routine part takes very little time. The problem is that this activity has no closing hour. What actually consumes attention is not the recurring task list — it is the events that arrive unannounced with a window measured in hours.

Four separate costs, counted below.

One: opening the position

Choosing the asset, sizing the two legs, getting both filled. This is a one-off, but it takes longer than it looks, because the tradable set is far smaller than the leaderboard suggests.

Our database currently covers 534 symbols and 1,029,262 settlement records (snapshot as of 2026-10-08 06:00 +08:00). That sounds like plenty of choice, but only two of them — BTC and ETH — have more than two years of continuous history; the rest are overwhelmingly recent. Strip out the names with no spot market at all and what is left that can reliably support two legs is a small set.

Getting both legs down at the same time is not two mouse clicks either. Whichever fills first is naked exposure until the other lands, and however far price travels in those seconds is the starting error on the position.

Two: the routine, counted in settlement periods

Most people still assume funding settles three times a day. That default no longer holds.

Eight-hour contracts settle 3 times daily; the four-hour tier settles 6; names pushed to the one-hour tier settle 24. All three intervals coexist on the same exchange. Measured over the trailing 30 days in our own data: a typical four-hour symbol logged 180 periods, while a denser one such as LSKUSDT logged 392 in the same window — about 13 a day.

Every settlement does two things: moves cash according to that period’s rate, and shifts the relative value of the two legs. So in principle, after each one, three things want checking:

  • Has the sign flipped (negative means you are now paying)
  • Are the two legs still proportioned correctly
  • How much of the liquidation distance has been eaten

Nobody actually looks every period. In practice you set thresholds and look when they trip. But the thresholds themselves need maintenance, and rebalancing is about margin rather than delta — a judgement that does not reduce cleanly to one simple rule.

Diagram: three groups of vertical lines on a shared baseline with density increasing left to right until the rightmost group is nearly solid; above each group sits a small square of identical size while the line counts below differ several-fold

Three: the events that do not arrive on schedule

This is the real cost, and the hardest to estimate.

In the last three weeks alone: Binance removed five margin pairs on 17 September, freezing the spot leg for three hours; revised leverage and margin tiers on 16 perpetuals on 24 September, meaning every liquidation distance needed recalculating; and delisted three contracts on 5 October with four days between announcement and forced settlement — where the final hour runs on different liquidation rules.

None of these wait. Delisting in particular is hard-edged: settlement happens at the stated time whether or not you are at a screen. And announcements go out on the exchange’s clock, which may be three in the morning on yours.

Once you include this line, the shape of the cost changes. It is not a fixed slot each day; it is a pager. The real price is that you cannot be offline for long, not that you spend a few minutes daily.

Four: reconciliation

The cash that arrives rarely matches the funding statement, because several other things sit in between: the spread on entry and exit, rebalancing friction, fees and rebates. Pulling the detail and reconciling it periodically is the only way to know how much of this month’s return genuinely came from funding.

Skipping it does not just leave a number wrong. It leads you to treat a one-off spread gain as recurring funding income, and then to forecast next month off that.

So is handing it off free?

No. It converts the time cost into two other things, and that should be said plainly.

One is the fee. The other matters more: delegating execution means taking on the risk of whoever you delegated to. Whether the platform absconds, whether key permissions are scoped tightly enough, whether the reported returns can be verified — those become the new homework. We list the part that can be checked in how to verify the returns a platform reports to you, and we put our reason for not even requesting withdrawal permission in writing.

What our system does is essentially run costs two and three continuously on the user’s behalf — settlement cadence, contract lifecycle, liquidation distance. Whether that is worth it depends on whether your capital base absorbs the fee, and whether you want to be permanently reachable.

A final caveat: saving the time does not remove the risk. Rates turn negative, contracts get delisted, exchanges change rules — all of that happens regardless of who is at the keyboard. The worst year on our performance page is what it looks like when it does.