I keep running into people who think of leverage as a single lever. Turn it up, take more risk, done. But if you want the same long exposure to something, say a fixed dollar amount of Bitcoin or of an index, you have at least three ways to finance it, and the cost of each one lives in a completely different place. A broker margin loan charges you an explicit interest rate. A futures contract charges you nothing you can see, because the cost is baked into the price you pay above spot. A perpetual swap charges you a stream of funding payments that can flip sign. Same exposure, three invoices, three hiding spots. The trick is converting all three onto one axis so you can actually compare them.
Get everything onto the same axis
The only number that lets you compare these fairly is the all-in annualized financing cost as a percent of the notional you are controlling. Not the fee, not the spread, the carrying cost of holding the position for a year expressed against the exposure. Once everything is a percent per year, the comparison is trivial and often surprising.
Start with the margin loan, because it is the honest one. Your broker quotes a rate, you post some equity, you borrow the rest to buy the asset outright. The cost is just the loan rate times the borrowed portion. If you buy a dollar of exposure with fifty cents of your own money and fifty cents borrowed at, say, a rate that tends to sit a few points over the risk free rate for retail accounts, your annualized financing cost on the full dollar of exposure is roughly half of that rate, because you are only borrowing half. That last point trips people up constantly. The quoted margin rate is on the borrowed slice, not on your total exposure, so leverage that uses more of your own cash carries less financing cost in absolute terms but also gives you less of it.
Now futures. There is no visible interest line, and that is exactly why futures fool people. A dated future trades at a price that already embeds the cost of carry until expiry. In normal conditions with positive interest rates the future trades above spot, and that premium is the financing cost. To annualize it, take the percentage gap between the futures price and spot, then scale it by how much of a year is left until the contract expires. A contract three months out sitting two percent over spot is costing you roughly eight percent annualized, because you are paying that two percent to hold for a quarter, four times a year. The invoice never shows up as interest. It shows up as the future converging down to spot as expiry approaches, quietly eroding your position even if the underlying never moves.
Perps are just the same cost, repriced constantly
Perpetual swaps do not expire, so instead of embedding carry into a term structure they use a funding rate to tether the perp price to spot. Longs pay shorts, or shorts pay longs, usually every eight hours. When the crowd is long and the perp trades above spot, funding is positive and longs pay. To annualize, take the periodic funding rate and multiply by the number of periods in a year. A funding rate that looks tiny per eight hours, something you would barely notice on the screen, becomes a serious number when you realize it is charged three times a day, over a thousand times a year. I have watched people hold a leveraged perp long for weeks through a crowded rally and lose more to funding than they would have paid in explicit interest on a margin loan for the same exposure, and they never once saw an interest charge.
Here is the workflow I actually use when I have to choose a route for the same trade:
- Fix the exposure and the holding horizon first. You cannot compare financing without pinning both, because the cheapest route flips depending on how long you hold.
- Margin loan: quoted rate times the fraction you are borrowing. Add any borrow fee or hard to borrow spread if you are short.
- Futures: percentage premium over spot, scaled up to a full year by dividing by the fraction of a year until expiry.
- Perps: current funding per period times periods per year, and check whether it has been persistently one sign or whipsawing.
- Add the round trip execution cost, spread plus fees, amortized over your horizon. On a one week hold this can dominate everything else.
Why the cheapest route keeps moving
The reason there is no permanent winner is that all three of these costs are driven by different forces, and those forces move on different clocks. Margin rates track the central bank rate cycle almost mechanically, so when short rates are near zero, borrowing from your broker is close to free and margin looks unbeatable. When short rates climb, that same margin loan gets expensive fast, and the basis in futures widens right alongside it, because the futures premium is largely just the cost of carry, which is largely just the risk free rate.
Perp funding, by contrast, is mostly a crowding tax. It tracks positioning far more than it tracks interest rates. In a euphoric market where everyone piles into leveraged longs, funding can blow out to levels that make perps by far the most expensive way to be long, even when margin rates are high too. In a quiet or fearful market where longs are scarce, funding can go flat or negative, and you can occasionally get paid to hold a leveraged long. So the ranking is genuinely regime dependent. High rate environment with a crowded long book, futures or a well priced margin loan usually win. Low rate environment with balanced positioning, perps are often cheapest and most flexible. There is no answer you can memorize once.
The failure mode I see most is people picking a route out of habit, usually perps because the interface is frictionless, and never once annualizing the funding to compare it against the boring margin loan sitting in their brokerage account. The screen shows a small number and the brain files it as small. Do the arithmetic and the small number is often the largest of the three. On Blockcircle we surface funding and basis side by side across exchanges partly so this comparison stops being something you have to reconstruct by hand every time, but you can do the whole thing on a napkin. Fix the exposure, fix the horizon, put every cost in percent per year, and pick the smallest one. The answer will not be the same one next quarter, and that is the point.