Two years after the Flash Crash – Spreads are Narrower and Liquidity Deeper than Ever Before? Guess Again.
I bought 500 shares of IBM on the opening print today. You would think 500 shares would be pretty easy to liquidate within a penny or two considering how our market scholars claim that the market has deeper liquidity and tighter spreads than ever before. Anyways, a minute after the open, this is the quote that I was trying to liquidate that 500 shares into:
*charts courtesy of Neovest
Yes that’s right you are reading the quote correctly. The spread was 37 cents. 100 shares x 100 shares. Looks pretty thick and deep to me. And why does the quote change so many times in the one second increment that I have shown? HFT liquidity providers change their quotes continuously in that 1 second increment, usually trying to aggressively penny each other, but then cancelling their orders simultaneously.
So if I wanted to sell my 500 shares at the market, I could be filled on one hundred shares at 203.10, and pay that 37 cent spread. And on the last 400 shares, who knows where I would be filled.
But this is just an isolated quote isn’t it? Guess again. Here is the quote two minutes later. It’s actually a penny wider, 38 cents.
But it does gets better, and the spread does tighten up. 11 minutes into the trading day the spread is now a tight 27 cents.
Well IBM is a $200 stock so of course the spread is going to be wider. This isn’t the case with other DOW components is it? No they are indeed tighter.
Proctor & Gamble (PG) had a nice tight spread of 21 cents, 19 seconds after the open.
Coke (KO) had an 11 cent spread, in the first 30 seconds of trade.
Then tightened to a 10 cent spread, 100 shares x 100 shares 2 minutes after the open.
MMM had a cool 25 cents spread after the open.
These are supposed to be some of the most liquid stocks on our exchanges. Where is all this liquidity that our academic scholars claim the market has?
Well if you analyze spreads in the middle of the day, you will get an entirely different picture.
If you bring up the IBM quote in the middle of the day, you will typically find the spread to be only a few cents. Coke and P&G will usually have a 1 cent spread Why is this? HFT liquidity providers do not like risk, and when there is so much uncertainty of where the price is going (like when the stocks open, or right before market moving events), they simply do not quote aggressively. Many HFT programs don’t quote at all.
Nanex showed images of the E-mini liquidity around a 10:00 economic number release on April 19th. http://www.nanex.net/aqck/3233.html They call it a liquidity vacuum, as HFT participants cancel their orders ahead of the number.
Was this lack of liquidity an isolated event today? Hardly. This is a daily occurrence. I would challenge all our academic scholars who analyze spreads to take a good look at liquidity a few minutes after the open, or around any type of market impact event such as a 10:00 economic news release. You’ll find liquidity to be very thin and very dispersed. Indeed, we have plenty of liquidity in the market, just not during periods of market uncertainty.
Update: We’d like to thank Nanex for taking the time to post liquidity charts of the stocks mentioned in this article:
Dennis,
Very interesting posting, thanks.
Joe Saluzzi and Sal Arnuk of Themis Trading had a good blog piece today about ‘tighter spreads’;
http://blog.themistrading.com/their-2-cents-aint-even-worth-2-cents/comment-page-1/#comment-1454
Although it’s not quite up to date Nanex showed this fiction of tighter spreads in one of its Research pieces.
http://www.nanex.net/aqck/2805.HTML
DavidC
Thanks David. Joe and Sal do some great work at Themis. We are fans.
Thanks for your reply Dennis. I thought you might know them.I got linked to your blog piece from one of Sal’s Tweets! I’ll be following your site from today.
All the bes,
David
Sorry, should have read All the best…
David
Why do you assume liquidity is a right of any market participant? Liquidity-risk did not invent itself.
JSW,
I don’t believe that was the point of Dennis’s piece. HFT proponents say that they provide liquidity and reduce spreads (amongst other things) – this shows they don’t. If you read the links in my previous comment you’ll see further evidence.
DavidC
David,
Having studied global market microstructure for the past several years, the opening liquidity of ANY market, whether in India, Tokyo, the US, or Europe, is weak at best and non-existent at worst – but that is the nature of the beast. Any market is subject to liquidity. If the bid/ask spread is a measure of “risk” (as most academia has shown) inherent in the market, then you can tell the “most” risky times of day are obviously the open, the close, and any potential announcements.
To anyone active in any market for a reasonable period of time, this is nothing new and hardly surprising. In fact, it’s just an element of markets as a whole that must be dealt with.
I do not give any credit to the research of Themis so your links are lost on me unfortunately.
Hi JSW,
I agree with you that liquidity has always been less at the open and around market moving announcements. However, I have been a prop trader for 12 years and have made millions of physical trades, and I can tell you that the liquidity at the open is much thinner than it was even a few years ago. Not blaming HFT here, but I think we are lacking traditional participants as they have moved much of their trading to the dark. The fragmented structure only complicates things further as well.
What is also taking place, under the high letancy radar is low letancy order flow by hyper active co-located HFT robots. These HFT robots see the orders before most of us and execute flash orders before we even know they sent the order. So all we see is time and sales, never seeing the robot orders at all. It’s like an invisible robot market able see and react before the data ever reaches our computer screen. This practice also front runs the large visible orders that you’ve displayed the depth of. The robots search out large orders like this, so there’s no question they were all over FB. Almost 80% of volume today is automated (robot generated) bringing the average hold time of a stock today below 22 seconds.Nice PostJohn
Actually this blog I did a while back, might help to explain why much of our displayed liquidity has moved to the dark. http://premarketinfo.wordpress.com/2012/02/23/dark-secrets-where-does-your-retail-order-go/
JSW,
I agree with your comment about liquidity at the ‘risk’ times of the day and it is something I’m fully aware of (I’ve followed the markets since the early 90s having worked for a financial information vendor although I only started trading in 2007), but it still doesn’t negate my previous points (or, indeed, Dennis’s).
Since the advent of Reg NMS and HFTs, the purported benefits of HFT have been shown to be false. The Nanex research piece(s) SHOW(S) this.
I’m intrigued by your not giving any credit to Themis’s work. One of the very reasons I started following them was because of their knowledge and the fact that they are market practitioners – they know what they are talking about, so I’m surprised you dismiss them.
DavidC
, the orders are being made plbiuc before others have a chance, that’s one thing. The controversy as I understand it about HFT is that the orders are NOT always made plbiuc but the super computers the large firms are using can discover price and volume of orders faster than others. Here is how it works: I put in an order to buy 100,000 shares of Google at up to $508.50/share but let’s assume that this order is not revealed to the market. Still, the order is sitting there on my server. Goldman’s supercomputer puts out an order to sell 100 shares at $508.25 (current market price as of this writing) and my server eats it.Goldman’s computer says, “Gee, that was easy” and puts in another order to sell another 100 shares at $508.26 and my server again takes it.Then Goldman’s computer tries $508.27 and is successful again.Goldman’s computer keeps firing out orders to sell 100 shares incrementally higher until it tries $508.51 and my server does not accept it. So now, Goldman knows my order is at $508.50 and starts firing off sell orders continually until the remainder of my 100,000 share order is filled. Goldman knows that it can buy these shares at the current market of $508.25 and make the spread of $0.25 per share which might not seem like a lot, but x 100,000, that’s $25,000 and their computers are doing this for hundreds or thousands of stocks all day. And remember this all occurs in a fraction of a second, which makes it nice work if you can get it
The only thing Goldman has to worry about is if Morgan’s computers which are trying to do the same thing are able to do the trade faster. This is why these companies are spending so much money on progamming whizzes and the fastest IT resources they can get. This is a technology cold war on whose computers are fastest. One millisecond can make all the difference. This is why companies like Goldman must locate their computing power right AT the exchange in NY and the exchange leases server space for this purpose. They also pay $3,500/mo. for data which is a lot for us, but nothing for them. The reason is that to locate a trading server even 100 miles from Wall St. means 1-2 milliseconds of delay due to simply the speed of light and data transfer, resistance of wires, etc. That can mean the difference between them scooping trades or someone else scooping them. Congress is reluctant to try to stop this because, technically, there is nothing illegal about what they are doing, but they do have a clear advantage over retail customers. We can never compete with HFT, but we can still make money in longer-term (in this case, >1 second) trading.===I look forward to reading any feedback to this.
Dark pool, light pool, OTC pinksheets, or private placement, it doesn’t make a difference. Trading, whether making markets, speculating, or hedging a book, isn’t charity. Liquidity isn’t a right, as this blog post suggests.
JSW,
I won’t respond any more so you may reply as you wish.
All I will reiterate is that that is not the point being made in this blog post (at least, I didn’t read it that way). The point being made, as the title suggests, is that the purported benefits of HFT are a fiction, NOT that liquidity is a right. I urge you to look at the Nanex research.
http://www.nanex.net/aqck/2805.HTML
http://www.nanex.net/aqck/2818.html
http://www.nanex.net/aqck/3247.html
Another thing that intrigues me – do you actually trade yourself?
DavidC
David, you suggest the benefits of HFT have been shown to be false, yet admit liquidity outside of “risk events” is still prevalent and available. This argument is akin to blaming diets because obese people still exist.
JSW,
Errm, no.
Levels of obesity are going up. If more people are dieting then diets aren’t working.
The use of HFTs is going up. If spreads are narrower (they’re not – see Nanex’s research) and liquidity is deeper (it’s not, given the point of Dennis’s piece) the HFTs aren’t working.
I’m not talking about the existence of varying liquidity through the day, I AM talking about the purported effect(s) of HFT.
No more from me so you may respond as you wish.
DavidC
Sorry, second paragraph should read;
The use of HFTs is going up. If spreads are narrower and liquidity is deeper the HFTs are working. As spreads AREN’T narrower (they’re not – see Nanex’s research) and liquidity ISN’T deeper (it’s not, given the point of Dennis’s piece), HFTs aren’t working.
DavidC
Dennis,
We posted a few more charts that go with your excellent article.
http://www.nanex.net/aqck/3018.html
Wow Eric. This is great. Thanks for the charts. I’ll add a link to this in the article as well.
David, the underlying tone of the article was enough to give away the general belief of entitlement in regards to liquidity premiums. The Nanex/Themis research is more of the same – watching market makers as they balance supply and demand in their various issues.
From the perspective of a market maker, Tom, what is the benefit to Tom in providing tight spreads on the open of a stock where he, by definition, has no clue which direction the market will go? Is Tom required to provide this service to the retail trader to the detriment of his own book, his own firm, and his own livelihood? Tom needs time to properly gauge supply and demand in order to properly firm up the book and eventually provide his full level of liquidity.
JSW,
OK, my final word. The article is NOT talking about ENTITLEMENT, it’s talking about supposed benefits of HFTs.
HFTs are NOT the same as a market maker – read Nanex’s research and buy “Broken Markets” when it comes out at the beginning of June.
No more from me.
DavidC
Ok, I realize you’ve had enough. But for what it’s worth, any market maker in today’s microstructure who is not by default HFT is not worth their weight in salt.
And yes, David, I do trade.