[mosermint]

Many market makers profit off speed and information advantages while providing liquidity. There are alternative market structures like frequent batch auctions that would allow better trading, lower spreads and negate the HFT speed arms race. This paper is a good overview:

https://www.aeaweb.org/articles?id=10.1257/aer.104.5.418

[jallen_dot_dev]

And the reason their orders execute in front of others' is because they offer the best price. If they didn't exist I would have to pay a little bit more to buy, or sell for a little bit less.

Is the full paper available without a login? Alternative matching schemes to price-time priority suffer their own drawbacks. Either there's no guarantee your whole order will fill (pro rata) or trade at all, and there can still be a speed arms race (there's an incentive to get your order in at the last possible moment before the batch to benefit from maximum information).

[darawk]

The alternative schemes he's referring to are batch auctions, which don't eliminate price-time priority per se. What they do is bucket time priority into discrete chunks, which eliminate a certain class of high frequency strategy that probably isn't particularly economically productive.

The problem with batch auctions relative to continuous time trading is that that discreteness forces market makers to charge larger spreads. That's the primary trade-off. Volume would likely be dramatically reduced while achieving comparably efficient asset allocation, but at slightly higher average transaction costs. Those higher average transaction costs however would likely go along with better tail behavior of spreads in unusual market conditions, and maybe better human interpretability under unusual conditions as well.

What it comes down to is a question of how much those non-monetary benefits are worth to your economy. The longer you force market makers to hold inventory, the more they have to charge for that risk, all else equal. However, when market volatility spikes, they're also going to be less able to play certain types of high frequency games that erode liquidity when it's most needed. It's kind of a robustness/efficiency trade-off, like many things.

[jallen_dot_dev]

Thanks for the insight, would appreciate any recommendation on where I can read more about this.

I'm curious what would attract market makers to this sort of exchange if it exposes them to more risk. Unless they can charge a premium for taking on that risk. But then why would traders want to pay more when they can get better prices (from tighter spreads) on today's more popular exchanges?

[darawk]

> Thanks for the insight, would appreciate any recommendation on where I can read more about this.

https://www.amazon.com/Trading-Exchanges-Market-Microstructu...

This is sort of the standard textbook on how exchanges and markets work, and has a nice discussion of the tradeoffs inherent in call auctions vs continuous time trading.

If you really want to get deep and are comfortable with math, this paper:

https://www.math.nyu.edu/~avellane/HighFrequencyTrading.pdf

explains the general framework market makers are using to make decisions.

> I'm curious what would attract market makers to this sort of exchange if it exposes them to more risk. Unless they can charge a premium for taking on that risk. Ultimately, market makers are service providers. They are selling liquidity to people, and they get compensated for that. What that means is that they will go wherever their clients go. And they wouldn't necessarily make less money in a higher spread regime, larger spreads mean bigger margins for the MMs. I think they probably would overall make less, because volume would decline so much, but it's just a little more complex than more risk == bad.

> But then why would traders want to pay more when they can get better prices (from tighter spreads) on today's more popular exchanges?

It's a little complicated. I mean, the sort of facile answer is: they wouldn't, and don't. We have a free market, if anyone wanted to they could start one of these up right now.

A slightly better answer is that some clients would prefer it and some wouldn't, and there's a reasonable argument to be made that the ones who would prefer it are more important to healthy market functioning. Specifically, it'd be much better for larger asset managers who research stocks and make careful picks based on rigorous analysis.

Right now, those asset managers, who are the people that really make the markets efficient, are being statistically frontrun by HFTs, both market makers and others. That is, some MM notices someone buying large blocks in a pattern consistent with one of these smart asset managers, and they label this "toxic flow". Backing up slightly, market makers want to sell to dumb money.

Consider a world where the only trades that occur are by "smart money". Someone only trades if they know the stock is going to go up or down, and there is no other trading activity aside from this. In this world, the MMs would make zero money, because they'd be constantly getting "adverse selected". MMs make profits roughly in proportion to the overall quantity of dumb/smart money. They make money when lots of trading happens, but the price doesn't go anywhere.

Getting back to the main thread, MMs are constantly trying to identify "smart money", and when they find that smart money, they either re-center their spread in the direction the smart money is trading, widen their spreads, or stop providing liquidity entirely. Continuous time trading makes it considerably easier to separate smart and dumb money than a batch auction would, and this means that overall these smart managers are making lower profits than they otherwise "should".

The flipside of this though is that these tactics actually improve the price that retail traders get. This is why Citadel buys Robinhood's order flow, and then offers them better prices than the open market does. They can do this because they "know" that all of the Robinhood flow is dumb money, and consequently they can safely tighten their spreads for those users.