Comment by nickles | Hacker News Reader

Comment by nickles | Hacker News Reader

Parent Full thread

Suppose a farmer wants to lock in the price of milk produced the farmer's dairy. The farmer wants to sell futures and sees a market of 18.30 / 18.50. These prices are probably being shown by a market maker. The farmer sells the futures to the market maker at 18.30. A few minutes later, Starbucks would like to buy the same contracts, to lock in its costs for the next month. Starbucks buys at 18.50.

These two participants have paid the market maker, in total, 0.20 in spread. Here, the spread is the fee the market maker charges for facilitating liquidity. At this point, the net sum between the three participants is still 0. However, we also need to factor in the fees charged by the exchange, any taxes that may be charged on the transaction (for example the SEC fee in equities), clearing fees, and funding costs.

On the whole, the transaction between the three participants was negative sum. However, the market maker is running a business by reflecting those costs, and the risk premium, in the spread. Even though this transaction is negative sum, it, presumably, still provides economic value to the farmer and Starbucks.

Isn't that example nearly zero-sum (ignoring the spread)? If the spot price for milk goes down later, Starbucks loses a bit of money. If it goes up, the farmer has effectively "paid" for an advance.

My point is that futures are by construction zero-sum (ignoring spreads) or negative sum (including them). Thus, any analysis that averages across market participants' P&L will conclude that futures trading is unprofitable[0], except for the market makers. This is true regardless of the savvy of market participants.

[0] Obviously, it doesn't capture non-cash considerations. Starbucks and the farmer both might be thrilled with reduced risk. But paper doesn't consider those factors either.

> Isn't that example nearly zero-sum (ignoring the spread)?

But ignoring the spread is incorrect, and nearly zero sum is not the same as zero-sum. The spread is the fee that market makers earn for a service they provide. You're also not considering other participants. The price changes you're talking about could be driven entirely by end users, or they could be driven by speculators, arbitrageurs, and others who are providing separate services (price discovery, liquidity, balance sheet, arbitrage, etc.) and are also capturing some of the value.

> futures trading is unprofitable[0], except for the market makers

If you are explicitly excluding participants who profit by providing financial services to markets, then you are begging the question.

> Obviously, it doesn't capture non-cash considerations. Starbucks and the farmer both might be thrilled with reduced risk.

Transferring risk can definitely provide economic value (insurance policies exist for this reason), but it should be noted that there are other advantages as well. These futures transactions can help Starbucks manage inventory without needing to maintain large stockpiles of perishable or reduce the volatility of prices charged to consumers.

Are you saying that the "misconception" is that it's actually slightly negative-sum, rather than exactly-zero sum? If so, great, I agree. It doesn't change my point at all; if anything it makes it stronger.

I am not excluding anyone; I'm trying to explain why I think the result here is not particularly enlightening.

My understanding of your comments is influenced strongly by this statement:

> Thus, any analysis that averages across market participants' P&L will conclude that futures trading is unprofitable[0], except for the market makers. This is true regardless of the savvy of market participants.

By virtue of offering various services to market participants, savvy market operators can consistently generate profit. In the context of this article, I took your statement to mean that any given participant cannot consistently do so.

At the same time, the end users entering into these transactions understand they will be paying these fees, just as a firm expects to pay a fee to borrow money from a bank. If we describe this process as a zero-sum game, it gives the impression that trading is a speculative casino, reallocating money to participants at random (the misconception). If we describe it instead as negative-sum, where participants are paying/paid for services, it better reflects the economic value of the transactions.

> ignoring/averaging across trades with non-daytraders

I didn't see this bit for some reason when originally responding to your comment. Had I, my comment would have been stated differently.

That's not what zero-sum means. The definition of a zero sum "game" is simply one where the participants' payoffs (here, profits) sum to zero.

Nothing in the definition prevents some participants from being better players than others. As you noted above, some entities might also prefer to "buy" stability for their main operation instead of trying to minimize their spending on some of its inputs. Nevertheless, for every dollar made on futures, someone has lost (at least) that much. Except for the market-maker, the entire thing is a closed system. Thus, you'd expect the average return, across everyone trading the contract, to be zero, which is exactly what this paper shows.

The paper does show 47 people who turned a profit. Their analysis can't distinguish between rubes who haven't (yet) reverted to the mean and savvy traders with some kind of effective edge. It would be interesting to see what the loss distribution looks like.

Reply on news.ycombinator.com