Is liquidity absolute or relative?

Here are two ways to get illiquid stocks out of your universe.

  1. Start with All Fundamentals, then cut out the bottom 30% or 40% of stocks in terms of liquidity by using FRank.

  2. Start with All Fundamentals, then cut out all stocks with a liquidity lower than, say, median daily dollar volume of $100,000 over the last three months.

Which method actually holds water in terms of trading difficulty (market impact, slippage costs, unfilled limit orders, etc.)? When dollar volume in the market as a whole is extremely low, as it was in 2002 or 2009, was it easier to trade a stock with $30,000 median daily dollar volume than it is now, when dollar volume of the average stock is so much higher? Or was it equally difficult?

Thanks,

  • Yuval

Hi Yuval,

I only started investing in 2012 or so, so I do not have first hand experience. I asked a similar question quite some time ago: https://www.portfolio123.com/mvnforum/viewthread_thread,8203#42321. No real conclusion, but it does seem like volume was fairly normal in 2008/2009.

Median volume always backtests worse for me when I try it. Averages are ‘distorted’ a lot by days with extremely high volume. But I think that the higher volume usually signals very good or very bad news. The 10 day volume divided by the 30 day volume is a very good factor for example (not counting slippage). What I’m trying to say is that using average instead of median volume helps to select ‘good’ or ‘bad’ stocks. The bad ones are hopefully weeded out with the ranking system.

On the other hand, averages can be very deceiving for microcaps. They often have a couple of high volume days, with practically nothing in between. They can be a nightmare to trade with reasonable slippage even now.

I always use method 2. Like you, I don’t really know what’s best though.

Regard,
Peter

Peter -

Since I posted this I did some testing using spread estimators based on high, low, and closing prices and found that bid-ask spreads correlate much better with absolute median daily dollar volume than with the rank of median daily dollar volume. So I’m also going to stick with method 2. Thanks for your feedback.

And regarding median vs average volume, I’m going to stick with median because of the microcap problem you point out. But maybe when on the rare occasion I do technical analysis based on volume I’ll use average going forward . . .

  • Yuval

Yuval, thanks for bringing this up. I have been using average daily total for years and did not even notice the median version. One of my ports selects some small caps and after re-analyzing it using median (I use MedianDailyTot(60)>1000000), it gave me better Sharpe after 2006. Before 2006, it was worse. But pre 2006 had some weird picks that in retrospect I would rather not have owned. They were ‘one day wonders’ for volume. I too feel more comfortable using median because it handles outliers better.