A friend of mine held an oil ETF through a stretch where spot crude rose and his position still lost money, and he was sure the fund was broken. It was doing exactly what its documents said it would do, holding front-month futures and rolling them every month. The thing eating his returns was the shape of the futures curve. That shape is acting on every futures position in every market, all the time, and in my experience most people holding those positions have never once looked at it.
Two shapes, one piece of vocabulary
A futures curve is nothing exotic. Take one asset, list the futures price for each delivery month, and plot them left to right from nearest to furthest. If later months cost more than near months, the curve slopes upward and the market is in contango. If later months cost less, it slopes downward and the market is in backwardation. That covers the vocabulary, and the more useful question is why a curve slopes one way or the other, because the slope is telling you something real about the asset underneath.
The textbook frame is cost of carry. A futures price is roughly spot plus whatever it costs to hold the asset until delivery, meaning financing, plus storage if the thing is physical, minus any benefit you get from holding the real asset instead of a paper claim on it. That last piece is called convenience yield, and it sounds academic right up until you watch it move a market.
Oil, gold, and Bitcoin slope for different reasons
Oil is the classic case because storage is genuinely expensive. Tanks, insurance, financing the barrels while they sit there. In a calm, well-supplied market oil tends to sit in contango, with futures priced near spot plus carry, because anyone can buy physical crude, store it, and deliver it against a future, and that arbitrage keeps the gap honest. But refiners and airlines need actual barrels now, and when supply gets tight, a barrel today is worth more than a promise of a barrel in three months. When that convenience yield outweighs storage and financing, the curve flips into backwardation. The extreme version came in 2020, when storage nearly ran out and the front-month WTI contract briefly traded below zero, which was contango stretched to the point of absurdity.
Gold barely plays this game. Storage is cheap relative to its value, above-ground supply is enormous, and nobody has an urgent industrial reason to hold a bar this week rather than next quarter. So gold sits in mild contango almost permanently, and the slope is mostly interest rates in disguise. If you want a clean picture of what pure financing cost looks like on a futures curve, gold is the one to stare at.
Bitcoin has no storage cost and no convenience yield, so its curve is financing and sentiment with nothing else in the mix. In bull markets, leveraged longs pay up for dated exposure and the futures premium gets fat. The annualized basis has historically reached well into double digits during euphoric stretches, which is why the cash-and-carry trade exists. You buy spot, short the future, and collect the spread as the two converge. When the market gets scared the premium collapses, and the curve occasionally flips into backwardation, which has historically been a reasonable sign that leverage just got flushed out.
What the slope does to a position you are holding
Here is the mechanical part my friend with the ETF never saw. A futures contract has to converge to spot as expiry approaches. If you are long in a contango market and spot goes nowhere, your contract drifts down toward spot, and when you roll into the next month you sell the expiring contract and buy a more expensive one. Do that twelve times a year on a steep curve and you can bleed a meaningful chunk of the position without spot ever falling. Backwardation runs the same machine in reverse. Your contract drifts up toward spot, each roll has you selling dear and buying cheap, and you are effectively paid to hold, which is one reason trend followers have historically liked long positions in backwardated commodities.
The habit that protects you takes about two minutes.
- Before holding any futures position or futures-based ETF for more than a few weeks, pull up the curve and annualize the spread between the first two contracts. That number is roughly what standing still costs you, or pays you, per year.
- In steep contango, spot has to rise by roughly that annualized spread for a rolled long position to break even. Decide whether your thesis actually clears that bar.
- In backwardation you have a tailwind, but ask why it exists. It usually means tight physical supply, and tightness can resolve suddenly.
- Never buy a front-month futures ETF as a long-term proxy for spot without doing the first check. This is probably the most common retail failure mode in commodities.
The same logic runs perp funding
Perpetual swaps have no expiry, so there is no curve to plot, but the same economic pressure has to surface somewhere, and it surfaces as the funding rate. When the perp trades above spot, longs pay shorts a small amount at every funding interval, which drags the perp back toward spot. That is contango expressed as a cash flow instead of a shape. When the perp trades below spot, shorts pay longs, and you are looking at backwardation under another name. Annualize funding and compare it with the basis on dated futures for the same asset and the two tend to track each other, because arbitrage desks move between them whenever they drift apart.
The upshot for anyone holding a leveraged crypto long is the same as for my friend and his oil fund. Positive funding is a continuous roll cost, and when funding runs hot you are paying a crowded-trade premium at every interval whether price moves or not. I check the annualized number before holding anything levered for more than a day or two, and I read a sudden spike as information about how crowded the trade has become.
I do not use curve shape to predict direction, and I would be careful with anyone who claims they can. What it prices is the cost of holding, and that cost compounds whether or not the trade works. Check the slope before you enter, recheck it when it steepens against you, and be suspicious of any futures backtest that ignores roll costs, because in markets that spend most of their time in contango, the roll is often the biggest driver of long-run returns.