Symmetry and Fairness — a Few Thoughts on Exchange Fees and Announcing Ours

Maker / taker fee structures were one of the issues that really bothered me back when I was a statistical arb trader at an investment bank.

In theory, it shouldn’t matter how they are set, but in practice, it often benefits market makers and exchanges at the expense of the average customer. Let me explain my point of view on this.

Asymmetric maker / taker fees are a distortion to market prices. If you are a customer looking at the top of book of two exchanges, one where the market is 3999 @ 4001, and the other where the market is 3995 @ 4005, it should be obvious where to trade, right? Well, if the first exchange has a 20bps taker fee and 0bps maker fee, then the true cost to lift the offer there is more like 4009. If the 20bp fee were split evenly between maker and taker, then the true cost to lift an offer on exchange 2 is actually around 4009 as well.

But guess which exchange the customer will go to?

Exchanges work on optics — who has the tightest, deepest markets? And to make the optics look good, giving market makers reduced, zero, or positive fees (rebates) will always make the bid/ask spread look tighter than it actually is. They are counting on customers being happy with the executed price while hoping they are not looking at that fee column at the end of the month.

In institutional trading groups, there are plenty of strategies that lose trading profits but make it back, and more, in maker rebates. That means as a trading manager, you are OK’ing a strategy that loses money every year but the rebate column (which is essentially a fee with the wrong sign) is positive. If that seems bizarre to you, I guarantee you it is even more opaque to the average customer.

It might come as a surprise to some, but professional market makers are takers, too. We do that a lot when the right, ephemeral opportunity to transact against the market at favorable prices warrants the taker fee. So when a professional looks at the screen and sees 3999 @ 4001, we are modeling the real market as 3991 @ 4009, which is a big, big difference and changes the calculus as to whether an opportunity actually exists.

If you do this for a living, it’s no issue because you simply model it correctly and take it into account. The real issue is the customer who looks at a fancy interface and thinks that inside quote is pretty close to what they’re getting. It’s not.

To make things worse, the exchange industry perpetuates this myth that you have to give market makers reduced fees, zero fees, or worse, rebates, to incentivize them to provide liquidity. I suppose market makers never made money before the advent of the rebate.

That’s ridiculous, of course. Market makers are incentivized to make money and they can make money no matter the presence or absence of rebates — they simply account for that in their models for the prices that they stream. In the end, professional traders are interested in trading edge — fee structure is simply something they take into account. What never factors into their models? Transparency on the screen to the average customer.

There is a better way. With symmetrical, low fees, even if a customer is not willing or able to model the cost structure, you can at least rest assured that you’re meeting in the middle and that it is truly a level playing field. You don’t have to worry that one side of the trade is tricking you with a screen price while knowing the real price, since both of you are paying symmetrically. That’s how fair, transparent markets should look — not these byzantine, asymmetric fee schedules that take up entire pages and obscure the true cost of trading.

Myself and a few others in the high frequency space really disliked this sort of practice and are surprised it still exists. I once jokingly said that if in the unlikely event I ever started an exchange, I would never do asymmetric maker / taker — well, for better or worse that day has come and I’ll be true to my word.

LedgerX’s Day-Ahead Swap fees: $0.99 for makers, $0.99 for takers. Sure saves us an extra column in the fee schedule.

Paul