Thursday, May 27, 2010

Machine Trading Good?

At a minimum, not bad.

The problem is that many people think stuff they don't understand must be bad. Of course, they still watch their plasma tv. Somehow, not understanding THAT must be okay.

Does Algorithmic Trading Improve Liquidity?

Terrence Hendershott, Charles Jones & Albert Menkveld
Journal of Finance, forthcoming

Abstract: Algorithmic trading has sharply increased over the past decade. Does it improve market quality, and should it be encouraged? We provide the first analysis of this question. The NYSE automated quote dissemination in 2003, and we use this change in market structure that increases algorithmic trading as an exogenous instrument to measure the causal effect of algorithmic trading on liquidity. For large stocks in particular, algorithmic trading narrows spreads, reduces adverse selection, and reduces trade-related price discovery. The findings indicate that algorithmic trading improves liquidity and enhances the informativeness of quotes.

(Nod to Kevin L)


Unknown said...
This comment has been removed by the author.
Unknown said...

Machine trading tends to provide liquidity when it is least needed and withdraw it when it is most needed, though. Also, liquidity is not always a good thing. Bourses halt trading in distressed counters and governments declare bank holidays for a reason after all. Machine trading is not off the hook.

aub said...

I'm not sure I understand what you mean, kaikaun. 'Most needed' by whom?

Firionel said...

There is one thing I consistently don't get: What the devil is liquidity?

I mean, I know what liquidity is in the everyday sense, but that is hardly a scientific definition.

So, what exactly is liquidity? Any ideas anybody?

spencer said...

The most basic concept of stock market liquidity is the ability to trade large volumes of a stock without significantly impacting its price.

spencer said...

Quantitative stock managers frequently talk about trades that their systems identified but when they actually tried to implement the trade the stock price moved so much that they could not execute the trade.

Anonymous said...

Liquidity is the speed at which an asset can be converted to cash.

With Respect to X said...
This comment has been removed by the author.
Firionel said...

@Spencer, BR:

I hear you, but am not entirely sure you are getting the depth of my uncertainty.

My point is precisely this: you just have shifted the ambiguity of the word 'liquidity' to another word. What does it mean to 'significantly impact a price', and what is a 'large volume'? (A similar problem rears its head as regards the speed of conversion, as that surely depends on the price I'm willing to accept.)

The troublespot I'm seeing here is, that what looked to be a one-dimensional entity ('liquidity') now all of a sudden is two-dimensional ('price change' and 'volume'; or 'price' and 'timeframe'), and still not very well identified.

Lest anyone think this is nitpicking, I believe it to be highly relevant to the article in question, as now it is perfectly possible to add 'liquidity' (as in volume) to the market, while my actual ability to sell a large volume at little price change ('liquidity' as well?) remains unchanged (or in fact, vice versa).

That's what I meant when I said I can see the everyday sense, but was looking for something more substantial.

Anonymous said...

Hold everything else constant (volume, price, etc) and measure the time it takes to sell a certain asset under circumstance A vs the time it takes to sell it under circumstance not A. That will tell you the impact A has on the liquidity of your asset.

The value of liquidity is in the opportunity cost (what else you could do with the cash) and the risk that the value of the asset you want to sell might decline.

Algorithmic Trading aids liquidity because it automates the time it would take humans to dig through piles of data to implement the same algorithm - the seller doesn't have to wait for a buyer to happen upon the deal you're offering - which might be stale by the time the non-automated buyer gets to it.

Firionel said...

Hi BR!

I don't mean to be obnoxious, but I think your response is begging the question to a certain degree.

I totally agree that holding price and volume constant you can then measure the time until a transaction goes through, and might call that liquidity if you wish. But now 'liquidity' is a feature of a particular trade, not of the market per se. Note too that it is propably massively discontinuous (at least along the price axis, I'm not so sure about the volume aspect), so this is not a trivial issue.

More precisely: lowering my asking price (very little) might (substantially) shorten the time until the trade goes through, thereby lowering opportunity cost resulting in a net gain. Had I held the price constant in my definition of liquidity I would have failed to capture that effect.

Since I'm not intending to make this a never ending discussion (and I am probably starting to sound like crank anyways) I will concede the following:
i) Very close to the market price (in a market where that is a meaningful concept) my above concern might turn out to be a nonissue. (I will insist on the conditional though.)
ii) In a number of situations the common non-definition ("everyone knows what liquidity is") actually works quite well.