Nanex Research

Nanex ~ 10-Dec-2013 ~ Adding Liquidity


It's rare to read a paper or article about High Frequency Trading (HFT) without coming across the claim that HFT provides liquidity to markets. The debate about whether or not this is true usually centers around the observation that HFT provides liquidity only when it suits them; as Andrew Haldane from the Bank of England put it: HFT provides liquidity during a monsoon, but withdraws it during a drought.

This paper isn't going to discuss the fleeting nature of HFT liquidity, but rather the actual definition of liquidity, and what it means to provide liquidity to the market. If this concept is foreign to you, I suggest reading this excellent, short description from Interactive Brokers. Basically, liquidity is a measure of how many buy or sell orders a market can absorb without a price impact. If an investor wants to buy a million shares of a stock, and there aren't a million shares offered for sale at the current offer price, the price is going to rise. If a market maker has a million shares for sale at the offer price, then the investor's million share buy order can be absorbed without impacting the price. We can say the market maker provided liquidity to the market.

More simply, the act of passively adding limit orders to an exchange order book adds liquidity, while aggressively executing against orders in the order book removes liquidity. This is a concept that every paper about market liquidity assumes the reader already knows.

Imagine my surprise, when a well known advocate for HFT claimed that "providing liquidity" means something entirely different, and may have the opposite meaning depending on who is adding or removing orders! Make no mistake, this wasn't a misunderstanding of words, or the result of things said in the heat of a debate. Read the exchange for yourself below.



On December 9, 2013, I had a discussion on twitter with Remco Lenterman that started off about Quote Stuffing and Credit Suisse's paper, and ended on the topic of liquidity. Specifically, what it means when HFT says it provides liquidity. I asked him:



His response took me by surprise, as I was pretty sure everyone in the industry knew what the terms adding (providing) and taking (removing) liquidity mean. It's especially shocking when you take into account that this is from the Chairman of the FIA European Principal Traders Association.

The terms adding/removing liquidity are not soft terms - they have clear definitions. One does not have to argue whether executing against orders resting in the book is adding or removing liquidity: this action by definition, always removes liquidity.

It's important to have a firm grasp of the basics - so when the claim is made "HFT provides liquidity", everyone understands what providing liquidity means. I had to be sure Lenterman understood the question:



Lenterman clearly understands the question, and is sticking to his original, wrong notion of what providing liquidity means.

A quick google search brings up many hits that define, explain and detail what the terms adding and removing liquidity mean. I sent him the top 2, one from Interactive Brokers and one from DayTraderGuru, and asked his opinion:



His answer: another surprise..



.. because now there is a conflict between his original notion and the industry definition of what providing liquidity means. I asked for clarification:



And that was the end of the conversation.

Would you believe I had a similar discussion when talking to an academic who advised the regulator on what happened in the flash crash? You'll just have to read the email correspondence for yourself.


Update: 12-Dec-2013 ~ Responding to Lenterman's Blog

In an attempt to muddy the waters and confuse the casual reader, Lenterman follows up on our discussion (see above - scroll up) with a blog post. The only thing he manages to clarify is that his definition of "adding liquidity" is not the same as the definition found on any one of the hundreds of financial websites on the internet.

Emphasizing his title, Lenterman tries to lend legitimacy to an outright lie:
by Remco Lenterman, Chairman, FIA European Principal Traders Association
The definitions of both liquidity and liquidity provider are always an issue of much debate and often confusion.

If the definitions of these simple, often used terms were debatable, then every academic paper using the term would specify their interpretation. Otherwise, a paper that claims HFT provides liquidity might actually mean HFT steals liquidity.

Lenterman writes:
When someone enters a limit order on the exchange that doesn't immediately execute (often called a 'passive order' or a 'non-marketable limit order') this action clearly 'adds' to liquidity. It is therefore often called a 'liquidity adding' order. When someone enters an order on the exchange that does immediately execute than that action will remove liquidity.

This is how I explained liquidity at the beginning of this paper. Lenterman gets it right and should stop right there.

Lenterman continues:
Where it gets truly confusing, even to some experts, as this twitter exchange shows , is when we use the term 'liquidity provider'. This is because a liquidity provider will both add to and also remove liquidity, sometimes evenly spread across their activities.
When a liquidity provider removes liquidity, they are no longer a liquidity provider: they are removing liquidity and competing with other investors who thought the liquidity providers would be providing liquidity.

Lenterman continues with more discussion of his prestige:
As an 'upstairs trader' with Goldman Sachs in New York, trading NYSE and Nasdaq stocks in the mid-nineties, I would make markets to clients as well as on the exchange. If a client traded against my price, I would ultimately liquidate any resulting positions by selling against passive orders on the exchange, or any other bids I could identify. Now in the strictest terms, that latter action removed liquidity. Yet, my activities back then were called 'liquidity providing' (or 'market making').
Lenterman was a liquidity provider to his clients, and a liquidity taker on the exchange - it is that simple. The bit about making a market on the exchange is not only immaterial (a client selling stock to him while he was making a market on the exchange, is simply a client selling stock on an exchange), it only serves to muddy the waters.

Lenterman writes:
When I was head of European shares trading 8 years ago in London for the same firm, we were pricing blocks of shares for institutional and corporate clients and thereby assuming the risk, an activity sometimes also called facilitation or again 'liquidity providing'. We would often unwind these resulting positions by 'removing' liquidity. That is what a liquidity provider does, he transfers risk from those who cannot bear it to those who want to assume it. Call me old-fashioned, but I think it is rather obvious that liquidity providers need to unwind their positions as well. Otherwise these positions become unmanageable.

Again, this is immaterial. Let's say that, to ensure profitability, the liquidity provider needs to unwind and sell a lot of stock. If during this unwinding an investor also wants to sell stock, what happens? Does the liquidity provider step in to buy from the investor, or do they compete against the investor and sell anyway? Since the "liquidity provider" has much faster access to the market, who do you think will be able to sell the fastest and get the best prices?

When liquidity providers and investors both rush to sell at the same time, flash crashes get started.

Lenterman writes:
Electronic market makers enter passive orders onto the exchange and by doing this add liquidity. Sometimes, they will liquidate any resulting positions and by doing this remove liquidity.
Yes! It's the act itself that's important, not the actor. It doesn't have to be an electronic market maker: it can be anyone from Warren Buffet to your grandmother.

Lenterman continues:
Sometimes they may value a security at a higher price than the best offer in the market and the liquidity provider's best bid may again remove liquidity. These actions create pricing efficiencies and I would still characterize them as 'liquidity providing', because they operate in exactly the same manner that traditional liquidity providers have always operated in.
Lenterman has conveniently omitted other participants who also value the security at a higher price, and wish to also buy stock and therefore remove liquidity (execute against sell orders). When the "liquidity provider" is buying at the same time as an investor, they are in competition. And we know who is going to win that race.

Just in case he hasn't yet thoroughly confused the reader, he borrows from Bill Clinton's famous "that depends on what the meaning of 'is', is" play.

Lenterman concludes:
Perhaps there is some logic to this as the word 'adding' is quite different from the word 'providing'. Purely theoretically the word 'provide' (furnish, accommodate) could also imply accommodating liquidity that was already there, i.e. removing it. But that is just semantics.
I simply have no words to respond to that. If you, the reader, are still confused, rest assured you are in good company: the Chairman of the FIA European Principal Traders Association.



Nanex Research

Inquiries: pr@nanex.net