Marketside chats #1: market vs limit orders

September 04, 2013
This is the first in a series of articles about general market topics. They will be simple to follow and include the kind of knowledge and insights that are hard to get from publications.

Some people consider market orders to be a much worse way to trade than limit orders, because you are paying the bid-ask spreadIt’s not quite that simple though:

Basic definitions
There is no well-defined single price for a traded security. There is a
  • bid price – the maximum price someone out there in the market is willing to buy at
  • ask price (a.k.a. offer) – the minimum price someone out there in the market is willing to sell at
One analogy is the used-car market: I can sell my car to a dealer for $8k, and he’ll sell it at $9k. Here the market is $8k – $9k.
Typically, as there are multiple participants in the market, one cares about the highest available bid and lowest available ask price. In the used-car example, if some other dealer will buy your car for $8.2 but sell for $9.2, it means the inside market is $8.2k – $9k. The best bid and best ask don’t need to be from the same person. In most securities markets, this “inside market” is also called NBBO (national best bid & offer).
Say the market for IBM is $184.50 – $184.53. Let’s define some terms in terms of a buy order:
  • market order is an order to buy at the current market, whatever that market is currently. In this case, the order would be executed/filled at $184.53 (the offer price).
  • A limit order actually specifies a price. Depending on the current market it can be:
    • marketable: its specified price is such that it would execute at the current market: e.g. a limit order to buy at $200 would get executed immediately at $184.53.
    • non-marketable: e.g. buy IBM at $183. This order can’t execute now, because nobody is desperate enough to sell to us at $183, so the order would be posted to some exchange. If the market later drops, our order would be filled. It could never get filled though if the market never drops.
(Sell order cases are similar.)
What is toxicity? 
The following are in descending order of toxicity“, i.e. the “informed-ness” of an order:
  • buying from a sophisticated high-frequency trading firm (HFT).
  • buying from a retired person selling stock to pay his rent.

As an extreme (unrealistic) example, the most toxic orders are those of a clairvoyant who can predict stock movements; you don’t want to buy from him because he knows it will drop in value right after he sells it to you. All else being equal, you’d rather not buy something that will instantly drop in value. Therefore, trading against a toxic order is bad. Random orders would be the least toxic.
A (rough) quantitative definition of the toxicity of an order O is the expected value of the adverse price move in the stock, conditional upon me trading against O, very soon after the trade occurs. “Adverse” means a stock dropping in value after I buy it, or rising after I sell it.

“Hidden gains” of market ordersFor simplicity, let’s say Acme retail trading firm only accepts market orders from its clients, or marketable limit orders – similar to market orders in this example. For further simplicity, let’s ignore trading commissions.
A first-order approximation for the “fair value” of a security is the average of its bid and ask price (1). Let’s say security ABC is trading at $50.00 – $50.02 (bid – ask); its fair price would then be $50.01. If Anna, a retail trader, sends a market order to buy, she’ll be buying at $50.02. This means she’s willing to pay 1 penny ($50.02 minus $50.01, also known as the “half-spread”) for the certainty of executing right now, thereby avoiding the risk of getting a worse price later.
Incidentally, in practice she may also not want the hassle of monitoring the market and reentering a new limit order at a different price later, had her original limit order not executed. Also, many retail brokerages give 20-min delayed quotes (much cheaper) (2). This makes it hard for Anna to send a limit order to buy at the (unknown to her) bid price of $50.00.
You can think of the half-spread as insurance; you get to trade right now, without the risk of trading at a worse price later. Of course, the price later could move both ways – but if you are sending a limit order to buy at $50.00 (the current bid price) and it doesn’t execute, the stock price is either the same or higher. If it were lower, your order would have been filled! So it’s now more expensive to buy. Therefore, that insurance is worth something. Many professional trading firms are competing to make a portion of that half-spread, and it’s not as easy as it sounds.
What is retail order internalization?
One way of thinking about toxicity is the expected % adverse price move conditional on having traded. In the $50.00 – $50.02 example, if I’m trading against some insider who is dumping his stock, my order to buy at $50.00 will get executed, but the stock is going to be dropping in price afterwards (perhaps because there are some bad news that only the insider knows about).
A more innocuous example is some pension fund selling a large amount of stock to pay pensions for this month. The fund may not have an opinion per se that the stock will go down, but their mere action of selling a large amount of stock will exhaust buyers (and out-compete other sellers) and drive the stock down. This is called market impact. In short, that half-spread isn’t such a low-hanging fruit; I can’t just be buying at the bid and selling at the ask all day and expect to keep that half-spread on average, because there are a lot of toxic orders out there that I could be trading against.
The aggregated set of orders that are sent by individual investors, i.e. not hedge funds, institutional investors, etc.) is called retail flow. It is (usually) neither very informed, nor large enough to move the market and cause market impact. Professional securities trading firms would therefore love to trade against retail: they are much more likely to keep that half-spread. Therefore, ACME will usually send its “order flow” to >=1 aggregators such as trading desks of large investment banks, which will “internalize” some of that flow (i.e. take the other side of some orders, without those orders ever showing up on a public exchange).
  • Retail clients win because the aggregator gives them a better price than a public exchange. There’s still the issue of adverse selection, as the aggregator is more likely to sell to you if they think the stock will go down (3), but that issue exists at a public exchange as well. In fact, if aggregators get first dibs on non-toxic orders, then it follows that it’s mostly toxic orders that go to exchanges. In other words, if you send a limit order to buy at the bid at an exchange, you are even more likely to trade against an informed counterparty than you might think. Therefore, anyone trading at a public exchange against you will assume that your order is somewhat toxic (note that trading is anonymous), and demand a higher price to sell to you (and lower to buy from).
  • Aggregators win because they get to keep some of that half-spread – though not all, as they compete on who can give the most price improvement. I have seen my retail market orders receive up to 50% of the half-spread as price improvement; e.g. when trying to buy at $50.02, getting an execution at $50.015, which is quite close to the fair value of $50.01). (4)
  • Exchanges lose because less trading activity happens there (5).
Conclusions:
  • market orders will get a better price than you think b/c they are likely to receive some price improvement.
  • limit orders are a worse idea than you think b/c of adverse selection.
  • for most liquid securities, the bid/ask spreads are small, so this makes even less difference.
  1. This can get complex; e.g. if there are 100 shares on the bid at $50.00 and 80,000 shares on the ask at $50.02, there should be more selling pressure, so the fair value is probably below $50.01. 
  2. Some websites display the realtime best bid & offer (BBO) of the BATS exchange, which is free. That’s not the “national BBO” (NBBO) tough; e.g. BATS shows $49.95 – $50.02 and NYSE shows $50.00 – $50.10, the NBBO is $50.00 – $50.02. So, in practice, you can get a realtime quote which is useful in most cases (e.g. liquid stocks, where the BATS BBO is the same or close to the NBBO), but not all.
  3. Professional trading firms will not trade with you only when they have an opinion on the stock, but also to reduce their risk (e.g. they ended up buying too much ABC). Also, they may be right 55% of the time (vs a random 50%); it’s definitely not 100%. So it’s not always all that bad.
  4. US stocks above $1 (most stocks you would encounter) have a minimum quoting increment of 1 penny, but can trade at prices of 1/100 penny. This means that I can’t tell the world I’m selling at $50.0199 (or send a limit order at that price), but an aggregator may execute an incoming retail order with a price improvement of at least $0.0001. Due to competition, price improvement is usually at least 1/10 penny.
  5. One could argue that large institutions lose as they aren’t given the chance to trade against such non-toxic flow; by extension the average investor loses, as some of his net worth (401(k), pension, etc.) may be managed indirectly by such an institution.

DISCLOSURE
The information provided here is for educational purposes only.  Nothing in this article should be construed as a solicitation or offer, or recommendation, to buy or sell any security.  Financial advisory services are only provided to investors who become Wealthfront clients.  Past performance is no guarantee of future results.