Mermaids, Fireflies, and the Bid-Ask Spread

On ETF and option market-making

A reader asks:

Do you have any insight into the activities of market makers when they act as authorized participants in the ETF market?

If you ask the internet how market makers earn profits, the typical response is something like “by capturing the (bid-ask) spread.” I have always wondered how “capturing the (bid-ask) spread” can be so enormously profitable and consistent.

I have been looking at deviations of ETF prices from net asset value and am now under the impression that “spread” means something totally different to the market maker (in particular, authorized participants).

Am I way off here?


My answer below, plus some story-telling on ETF options and option market-making generally:

ETF market-maker life

I was an ETF market maker for SIG for a bit and we were APs.

It makes more sense when you remember that liquid names in anything don’t really need market makers because there’s enough organic liquidity.

[Aside: This is a broadly useful insight — it’s an important part of the meta of when and where to shift gears back-and-forth between position trading and bookmaking]

To answer the question:

  1. there is a very long tail of ETFs that might have plenty of edge in them but just don’t trade lots of volume.
  2. the cost to create/redeem is very low and large mm’s have economies of scale (good funding rates, low commissions, fast execution) so they can make money with 1/2 cent of edge on a trade.
  3. published intra-day NAVs are often wrong especially when you get into more complex ETFs (the spreadsheet to price HYG or FEZ is a giant book with many assumptions because there’s a staleness in the prices of the underlying components — so you need a model for guessing what the fair current price of the components are — think of a European ADR price after the local market has closed but the US is still open — you might “beta” the stock to the SPX from when the local market closed while adjusting for the currency-cross tick by tick)

There are several firms that have built large businesses on the back of trading ETFs across time zones and internationally as the markets open from Asia to the US. Billions of dollars have probably been made on this in the past 2 decades. [Redacted list] of under the radar places all made this their bread and butter. In fact, they were way more focused on this than options.

US equities were super competitive back in 2004-2005 when I was doing this — and that’s even after adjusting for SPX futs prem/discount.

To explain that more clearly:

  1. we’d compute the current price of SP500 basket using direct data from the exchange feeds
  2. add the fair value of the EFP (cash-futures swap) to create the fair value of the synthetic future
  3. then compute the premium-discount by subtracting this synthetic from the actual SP500 futures price. If that number was positive we’d say the futures were trading “over” by say 10 bps or whatever that premium represented.
  4. we’d then beta weight the ETFs by that premium to predict the true fair value of the NAV (basically this all comes down to the fact that futures lead the cash)

This was standard practice 20 years ago if not more.

A few more tedious details:

  1. We were computing NAVs based on the files sent to us by the actual ETF sponsors every morning including how much cash on hand the (this was called a “cash plug” in our lingo) sponsor had in the ETF portfolio based on the prior day’s create/redeem activity.
  2. Computing the fair value also requires knowing the true borrow rate for every name because that is an input (although that can sort of be aggregated from knowing the EFP market since it should be incorporated into the cash/futs arb).

The general tradeoff you’ll face

More liquid names will have no edge but you can execute.

Less liquid names might have flow-driven mispricings that market makers are not closing (because they might know a premium is going to become more premium for example) but execution is harder.

Also, there’s massive adverse selection in this — if you are getting filled easily there’s a problem. When you get into the nitty-gritty of ETF arbitrage you are in the realm of understanding the prospectus — market makers trying to pick off other market makers based on some small legal gotcha or upcoming reconstitution of an index (I’m vague on the details but I remember the Holders ie TTH, SMH going thru some change that a few market makers recognized before the rest and picked them off by dumping long dated option premium on them that was about to evaporate. That one might have ended up in court.)

Shifting to options now…

Here’s one to ponder:

Option quotes are streamed in accordance to where a market maker thinks they can get their hedge off in the underlying. Typically the options will be priced off the mid of the stock’s bid-ask.

But if a market-maker is quoting ETF options and believe the NAV is different than the midpoint what does that mean if the option customer gets filled?

Option market maker life

Capturing the bid-ask is more cloak-and-dagger than the simple spreads you see on the screen. A professional market maker can see the screens, but their job is constantly update “what’s the true bid and true ask?”

A friend who was the lead mm in a semi-liquid options market used to describe his job as “throwing dust in the air so nobody sees where it really is”.

An options market with sparse or wide screens might trade way tighter via voice (especially in ETF options). The screens can be framed however in anticipation of an order or to entice someone to buy or sell.

Trying to figure out fair value in a name that trades by appointment is like trying to catch fireflies that blink every so often. The 3-month 25d put just traded near the offer…is vol higher or is put skew higher? If the straddle was offered by another party at the same time then the skew is probably just higher now. Maybe the straddle seller is making a statement about vol since they are, well, trading a straddle but the put buyer is betting on direction. In that case, maybe I pass on the straddles, sell the puts and hedge their delta and wait for the straddle seller to step down even cheaper. Then I’m synthetically legging ATM/25d put spread for a cheap price. Sizzler’s on me tonight.

But what are the chances the straddle seller is going to step down and lower the price? Have I seen the signature of this order before? Can that help me handicap their tendency?

What about the put buyer…do I think I’ve seen them before?* Do they tend to be smart on direction? In that case I want to hedge the put sale on a “heavy” delta.

All of a sudden a call buyer shows up in a deferred month. Do I think the term structure is shifting? Which one of these orders is the most aggressive?

If this sounds like fun it actually is. You are playing Sherlock Holmes in real-time with quick feedback. Especially if there’s so much action that there is 2-way flow simultaneously. If any of you came to the mock-trading sessions we did for StockSlam/Pitbulls you know the feeling of needing to hold a bunch of info in your head at the same time, maintain your ability to listen, and quickly react before the right trade is obvious to everyone else. Which is why experience and feel are crucial — you need to maximize the ratio of “good decision” to “how much info I need”.

Slow and stupid are indistinguishable in the market-making game. It’s not good enough to be “faster than average” in a winner-take-all ecosystem.

*People wonder how you can know “who” it is. You can’t know for sure, but you are pattern-matching. Have I seen an order of similar size, aggressiveness, time of month, moneyness, thru this particular broker, and so on. And if you roll, you’ve shown me the Monty Hall donkey.

Look, if you’re an account trading options in the voice market you are probably a repeat player. Maybe you have a periodic hedging program. Maybe you keep coming back because you’re winning. Oh, I remember trading against Mr. “This-Is-The-End-Of-The-Order”.

And brokers themselves have niches. When I hear a French accent I already know what class of customer is on the other side of the phone.

“Allo, dis is ah bee-yair”

Hi Pierre. I assume you will be joining the screen bid?

“ah-ah, yoo noh meh too well”

I’m going to level with you mon frere. It’s a rev/con. I’ll still be bid there myself next Monday.

“vat ah-boot for hahf koh-mee-see-on”

See, Pierre, the Royal Pierre who stands in for the French national pastime of clinically computing hedge ratios to 14 decimal places, is more price sensitive than extreme couponers because his client is protecting the margin on a structured note sold to private wealth account and the sales team get 99% of the p/l attribution, while the exo trader shakes a tin can on Wall Street for someone to pair off risk at fair.

Now the regional broker based out of Florida — he I’ll pay double comms to fund his cargo-short wearing client’s appreciation of Mermaid’s dancers.

So sure, the market maker doesn’t know the client’s blood type. But it’s hard to fool people twice.