# Measuring liquidity

While liquidity is one of the key figure of financial markets, It seems to be very difficult to measure. Volume is sometime used as a proxy but can sometimes be completly irrelevant.

Could you point to relevant research on what data to use and how to compute the measure?

-

Volume merely indicates how much buy-side interest exists in a stock. For liquidity, the sell-side interest is more relevant, which implies the quote characteristics (the limit-order book).

In addition to the bid-ask spread, I look at the top-of-book quote size. Here's an example from BATS:

sym | bid    ask    bidsize asksize
----| -----------------------------
AAPL| 325.12 325.21  100     100
MSFT|  24.70  24.71 3900    5900


I can only buy \$32,521 worth of Apple without impacting price, as opposed to \$145,789 of Microsoft. So the slippage is smaller.

There are more sophisticated measurements for order book entries. I could look at the full book ("level II data") to see the depth of the order chain. I could look across multiple exchanges, which is what a smart order router must do anyway. I could even look at related asset classes if the investor's goal is merely to gain exposure to general risk.

To be really swanky, I could investigate dark pools, though that's harder since the quotes aren't displayed. For this, a quant would need historical data regarding how much has been executed in the past. That's one reason why the big banks have a competitive advantage in dark-pool aggregator algorithms: they have enough client flow to record execution patterns.

-
 AAPL's price is 13 times higher than MSFT's, but the spread is only 9x wider. So crossing the bid/offer spread in AAPL costs me less as a percentage of the fill than in MSFT. – Ted Graham Jun 22 '11 at 20:49 @Ted AAPL's lower basis-point spread only exists because sub-penny quotes aren't currently allowed. Also, my answer was about quote size; AAPL's unnatural "advantage" disappears once the client places larger orders. – chrisaycock♦ Jun 22 '11 at 22:13

For my master thesis, I used the bid-ask spread as a liquidity measure. Intuitively, it is the price to the have the liquidity (or even the price of liquidity); the bigger the bid-ask spread, the lower the liquidity.

I know that Carlo Acerbi of MSCI is also looking into liquidity risk management and has a very interesting model for liquidity which is explained in this presentation. Maybe it can give you some ideas.

-
 One problem with the bid-ask spread is that it doesn't necessarily reflect the "market". For instance, I was looking at Greek CDS bid-ask prices around the vote yesterday, and the spread was surprisingly tight. But just because quotes existed, doesn't mean that anyone would transact at that level. – Shane Jun 23 '11 at 1:49 Well it depends how you want to understand liquidity. To me it's more a measure of "possibility to trade" than a volume measure. – SRKX♦ Jun 23 '11 at 7:32 @shane. If you hit the quote, there is a trade, no? Maybe there is no depth if you are able to trade only one unit, but I think that if there is a quote for both bid and ask and the spread is small it means that the market is there – RockScience Jul 11 '11 at 10:41

You can have a look at what the guys at Nanex.

Here is an example of what they look at. The chart is colour coded for market depth (the colder the colour the less depth)

-
 Do you know any details of making of that chart od Nanex - is each colour a percentage of depth for given price level ? or mayby it's depth measured by quantity of assets ? – Qbik Feb 12 '12 at 19:02

You may also be interested in a series of papers by Easley, de Prado, and O'Hara (2011), Flow Toxicity and Volatility in a High Frequency World. This paper follows up on a measure of the effect of trades on prices developed by two of the authors in 1987. They show that the new measure, which takes volume and concurrent price movements into account, can predict rapid changes in liquidity such as the "flash crash". From the abstract (published in JPM Winter 2011):

The ‘flash crash’ of May 6th 2010 was the second largest point swing (1,010.14 points) and the biggest one-day point decline (998.5 points) in the history of the Dow Jones Industrial Average. For a few minutes, \$1 trillion in market value vanished. In this paper, we argue that the ‘flash crash’ is the result of the new dynamics at play in the current market structure. We highlight the role played by order toxicity in affecting liquidity provision, and we show that a measure of this toxicity, the Volume-Synchronized Probability of Informed Trading (VPIN)*, captures the increasing toxicity of the order flow in the hours and days prior to collapse.

-
 I'd be cautious about trusting VPIN. There is nothing like a consensus on its validity or robustness. – Ryogi Nov 7 '12 at 16:45 The authors are highly respected and the paper certainly makes for interesting reading. Also the JPM doesn't just publish anything, they generally have a decently high standard. I'm just putting it out there so people are aware. – Tal Fishman Nov 12 '12 at 17:46 It sure makes for an interesting read. As to referring to highly respected (which they are) to hint at correct and valuable, you lost me at the 'hi'(gly). Here is a differing viewpoint. – Ryogi Nov 12 '12 at 18:39 Yes, I read that. I think the original authors disputed that the measure they were calling VPIN is not their VPIN. In short, a half-baked attempt at replication doesn't suffice (still useful to know, as quants don't always have the time or resources to devote to a full-blown replication). – Tal Fishman Nov 12 '12 at 18:53 Exactly, lack of consensus. – Ryogi Nov 12 '12 at 18:56