Through the Liquidity Glass
Vol #108: November 30th, 2022
“We make it up on volume” - First Citiwide Change Bank
No matter what business you’re in, doing more of a thing that makes you money, makes you more money. MoreVolume x MoneyThing = MoreMoney. Call it the commutative property of volume.
For professional traders, this was usually summed up by contracts traded and PnL per contract. If you have a keen portfolio manager who can separate the good trades from the bad ones1, you want to get her in front of more trades and showing bigger size. The higher the PnL per contract the better you were doing; the mercurial combination of larger winners and fewer losers.
It’s a nice clean statistic, but it hides the real complexity of how to manage opportunity and compare trading talent. Doing more isn’t always easy. More might put you over your Personal Risk Tolerance. Or what if there’s some constraint on the upside, where doing more volume introduces declining or even negative marginal returns?
The easiest lay-ups in trading are as obvious as they are limited. Market makers love a predictable, price insensitive customer. When mandate constrained players come to the marketplace, it’s usually good for solid and repeatable edge. The JP Morgan Hedged Equity Fund is a great example. They roll a very large options position in the SPX every quarter, and the fund’s objective makes pretty clear which strikes they’re going to be trading.
That ticket is only so big however. It’s not so easy to grab more of that, as relationships and competitive dynamics mean there is only so much to go around. You’re probably not even on the call list. If you push yourself or get pushed to do more volume, it’s not coming from that plum opportunity.
Even less predictable orderflow is capacity constrained. For high touch transactions that are negotiated over IM, there are a lot of politics to get the look, let alone your desired size. Often the best way to get more of the good stuff is to be willing to take the marginal trades. That’s a complex calculus for your net PnL. (Wine lovers will recognize this; to get access to the grand cru you often need to purchase the village wine, or commit despite a bad year.)
With purchased orderflow at wholesalers, getting a bigger slot on the broker’s wheel is a game for sharp elbows. This randomizer that doles out retail orders according to a target allocation is a lifeblood for large shops. You need to stay just ahead of your competitors, but not impact your margins by stretching on payment levels or overpromising on execution quality.
In other situations, the only way to get more volume is to offer more competitive pricing. The dynamics of price improvement auctions reward the counterparty who brings the customer order to the exchange. Tie goes to the wholesaler. If you want to collect more edge, you’re going to have to do it at a reduced rate. Be careful though, markets are pretty good at washing out those who sell flood insurance too cheaply.
Scaling production is a delicate balance that changes constantly. With scarce resources - eyeballs, risk capital, or compute power - choices about where to deploy must be made. Liquidity providers make decisions about where to go to work all day long. To understand opportunities, you have to follow the liquidity.
The definition of liquidity is fluid. In the wishy-washy spirit, I like defining it as the ability to get in or out of a position when you need to. Liquid products don’t quit on you when the going gets tough.
(More on the image used above here2)
Securities that have good liquidity will have the most accurate pricing. This makes them useful products, and desirable to trade. There is a virtuous circle to market liquidity. If options are being used to manage risk and tailor outcomes, customers want them to be as fairly priced as possible. The more confidence a market maker has in their pricing, the more liquidity they will provide. This gives the customer a fill closest to the “true” price.
I’ve always taken a hybrid approach to evaluating this practically. When this was a big part of my day job, I used both quantitative data and qualitative context to answer the question for our trading firm. Where should we be focusing the trading attention, how should we structure our access to this opportunity, and what tools did we need to do that?
There’s the short game of “get us signed up to quote XYZ stat”, and the longer term game of “we need to build a coordination tool that breaks up electronic crosses”. It’s all about touching that good paper, and guessing where it will go next.
Particularly in lower volume securities or ones that are rapidly exploding in activity, there can be a lot of pitfalls in looking just at the numbers. Indicators point you in the right direction, but there is always a story that needs contextualizing. At the end of the day I spent a lot of time looking at the screen markets and the time and sales record of where things traded.
With a byzantine menu of ways to transact, for the inside players not all volume is necessarily the same. Are 500 contracts coming from 1 big trade or 50 little ones? Did the trade take place via open outcry in a crowded pit, or did it sneak through the FLEX mechanism with less competitive dynamics? Was the volume from a dividend play? Is this activity from a one off event like earnings, or is it likely to continue?
The answers to these questions were the difference between scaling up or letting it pass. You didn't want chunky trades from one off events, but rather random scattered orderflow actively speculating on both sides. And if that came in a volatile issue with rapidly changing inputs - chef’s kiss.
A close cousin to volume is open interest - how many contracts are currently outstanding for that specific call/put or the underlying more generally. Since for every buyer there’s a seller, we can map how many contracts have been traded on that specific strike and expiration date. The magnitude of the OI indicates the degree to which that line is in play, well priced, or likely to see future activity.
Trading activity, and its representative data points of volume and open interest, give a strong indication about the quality of pricing in that issue. How much money has been betting on this price so far?
The rubber hits the road when a customer comes to the marketplace, and they are met with a best offer price and a best bid price. History be damned, the time to trade is now. The width between these two is the representation of how confident the market is with its pricing. Spread is the error bar that participants are collectively putting around their agreement of fair value.
There is a strong correlation between the amount of volume and open interest and the market width. Dealers give wider markets when there’s been no prior activity. Additional risk has to be priced in, because whatever the other side is doing, it’s a new trade. They’re not closing or unwinding an existing position.
Just as volume and open interest have their nuances, so does market width. Options prices are quoted in dollars and cents, but not all pennies are the same.
Much of option liquidity is ultimately dependent on the characteristics of the underlying stock. Starting simply with the underlying price, you’d expect options that have higher dollar prices to have wider markets in simple dollar terms. When SPX ($3963) options are .20 wide, that’s a lot different than when BAC ($36.86) has .20 wide markets.
Taking this as a percentage of the underlying helps, but it will obscure another important detail, the volatility of the underlying. Even with two stocks of the same dollar handle, .20 wide markets are very different levels of liquidity if that stock is moving 1% a day versus 5% a day.
Denominating market widths in volatility terms helps to normalize. 27 vol bid at 28 vol offered, gives a lot of information about the width of the markets. It’s also how practitioners tend to think about their levels in an aggregate sense - “I got short 1000 vega at 28 vol.”
Ultimately by combining the information we get from market widths with that of open interest and volume, we start to have a more complete picture of liquidity. Should I focus on trading 10,000 SPY contracts for a penny of edge, or 1000 of PEP for .10 in edge? The answer is complicated.
As we move out the spectrum of less and less liquid markets, strange things can happen. While I know that there were traders (not at my firm) who would take advantage of pushing around thinly traded equities, most public markets are too big for any one player to consistently and profitably swamp. Anyone who thinks they can would be wise to read about the Hunt brothers.
Where that might not be true is on the other side of the fringe in DeFi. Partly because they are unregulated, but also because they are tiny, there can be some difficult liquidity challenges. A code is law implementation means that without the guardrails of traditional markets, it is possible to exploit loopholes and liquidity air pockets.
Automatically liquidating collateral sounds like a reasonable risk management technique for a loan protocol, until the liquidation is so big that it distorts the underlying market. The same trading group has taken advantage of this twice in the last two months, first on Mango Markets, and more recently on Aave.
Defining liquidity is amorphous but important. I’ll bet AAVE token holders never thought they’d have to care about the depth of market in CRV. There are dozens of facets of nuance that we have to consider in deriving a true barometer. But combine a few stats with a gut check (scan) and you’re well on your way to better understanding the trading opportunity landscape.
While liquidity providers are, of course, obsessed with liquidity, it’s important for options customers to understand also. It can nick you trade by trade if your taste is for the more thinly traded names, or it can dramatically reduce the size of your winner as you try to close a three legged ITM butterfly on expiration Friday. A trade thesis is only as good as the liquidity that can support it.
Ever since I saw this video earlier this summer, I can’t get it out of my head. As a market maker you’re spammed with “opportunities” all day. Most of the time they mean your pricing should be slightly tweaked, but sometimes there’s a berry. This tomato sorting machine perfectly visualizes the exercise of separating the wheat from the chaff - though with significantly more precision than any honest trader can claim.
Below is the same prompt, but created with DALL-E that uses GPT-3 technology, whereas Midjourney uses GPT-4. While the above image feels more realistic, there is a certain “Pink Floyd Album Cover” charm to seeing the liquidity driven language with older technology.
While I was skeptical about “real art” being produced this way only a few months ago, I’m getting an amazing appreciation for how powerful this tool is. I specified the soil type (caillou) and pruning method (guyot), and it yields a remarkably accurate yet impressionistic image.
