I have some doubts about perps.
Let me start with futures. In exchanges, futures are cash-settled, so they’re basically bets on the price of an underlying asset.
They have an expiration/settlement date, so there’s a proper way to compute the fair (i.e. no-arbitrage) forward price F\_t of a future at time t. In particular, if T is the expiration date, then F\_T is also the price of the asset. If this weren’t the case, there would be an arbitrage opportunity around time T.
Let t0 be the time when a position is opened. Long positions get F\_T – F\_{t0} (or pay, when negative), while short positions get F\_{t0} – F\_T (or pay, when negative). In exchanges, settlements are periodic, so the payments to long positions are F\_{t0+1}-F\_{t0}, F\_{t0+2}-F\_{t0+1}, …, F\_T-F\_{T-1}. The sum is still F\_T-F\_{t0} (it’s a telescopic series).
Now let’s talk about perps. Perps have no expiration date, so there’s no proper forward price F\_t (unless we assume a renewable T=t0+1?). Let’s ignore leverage, indexes, and interests/fees, for simplicity. If I understand things correctly, perp_price > asset_price when there are more longs than shorts, and perp_price < asset_price when there are more shorts than longs. The periodic cash settlements are called funding rates and are a strictly increasing function of perp_price – asset_price. When that difference is positive, the longs pay the shorts, and when it’s negative, the shorts pay the longs.
When there are more longs than shorts, perp_price > asset_price, which also means that the longs pay the shorts (since the difference is positive). When the longs are paying, the long positions depreciate and so start decreasing in number, so perp_price decreases and gets closer to asset_price. A similar argument holds for shorts. This seems all good and well, since it means that perp_price tracks asset_price, but now my question is
*What are long and short positions betting on, exactly?*
With futures, they’re betting on the price of the underlying asset. In particular, longs are betting on the asset price going up, which is equivalent to betting on the future price going up, since the future price is a fair price and tracks the asset price. In the case of perps, longs are paid when perp_price < asset_price, but that happens when there are fewer longs than shorts. So, *are longs betting on there being fewer longs than shorts?* That doesn’t make much sense. Shouldn’t they bet on the price of the asset? Maybe the funding rate is computed using the previous perp_price? I doubt that. What happens if almost all positions are long and the asset price goes up? Do the longs still pay the shorts because they’re the majority? I’m clearly missing something.
>What are long and short positions betting on, exactly?
They’re betting on the price of the perp contract in the future +/- funding payments.
As you said there is no theoretical guarantee that the perp price will converge with the spot price as there is no expiration. In practice, the funding payments are usually enough to keep perps and spot aligned, although they often diverge for as long as traders are willing to pay the funding rate.
You can’t directly correlate the number of longs and shorts, or the total long/short open interest with future price expectations because orderbook markets don’t have uniform liquidity at all price levels.