"For those who believe, no explanation is necessary. For those who do not, none will suffice." - Joseph Dunninger.
I've resisted writing a piece about high frequency trading lately. Although I have a very good understanding of the subject, I see it as a kinda lost cause to try to educate people about - uninformed people want to believe that computers are front running them, stealing their money and manipulating the markets. People who understand equity trading know that this is just the culmination of technology advancement and competition for spreads, which has resulted in equity bid/ask spreads being narrowed to their lowest levels ever. There has never been a better time for the individual stock trader to execute orders: our bid ask spreads are narrower than ever. A high frequency trader may have a computer program trying to scalp a penny or a fraction of a penny from you - but this is better than it's ever been - better than the days of specialists scalping 1/8ths and more.
I left a comment stating as much on Floyd Norris's latest article "What Should be Done," only because he got it precisely backwards when he wrote, "In one sense, this is a return to the bad old days. Before reforms, the Nasdaq market kept the bid-asked spread for the brokers. The public had to buy at a higher price and sell at a lower one." On the contrary - high frequency trading has resulted in competition to capture those exact spreads, and has narrowed them drastically.
In my opinion, reaction to high frequency trading is all about the quote at the top of this post if you change the word "believe" to "understand." I will at least try to educate those who do not understand high frequency trading, so that they can form intelligent opinions - although I still fear that "no explanation will suffice."
Joe Saluzzi of Themis trading has gotten a ton of press based on a paper he published about the evils of high frequency trading. It's important to notice that Saluzzi is an execution trader, and that algorithms and high frequency traders make his life very difficult, because no human can do the job as well as an algorithm can. Saluzzi concludes his paper: "We also recommend that institutions use algo systems only for the most liquid of stocks. Anything less must be worked, the same as in the 'old days.' Institutions need to re-learn how to 'watch the tape' and take advantage of, or work around, high frequency traders." Or, institutions can embrace technology and use or develop better algorithms. Either way, Saluzzi's suggestion that it's kosher for him to execute the orders by "watching the tape" to gauge optimal timing, but that it's unfair for a computer program to do the same thing reeks of hypocrisy. The best comment I've read on the subject was this (emphasis mine):
Frontrunning is trading in front of a customer order. It is illegal. Collecting and analyzing publicly available information and trading based on your insights is legal. And guess what, someone will be the fastest and most accurate in doing this. The result of their actions is to translate meaningful information (strained from a stream of mostly noise, it's incredibly difficult to do) into price changes which make the price more accurate relative to what's knowable at the time. Unfortunately for people like Saluzzi, a 19th century "tape reader", a lot of the information that quicker, more talented, machine-using, more insightful players uncover is information about large size that he's trying to deceptively move without anyone knowing the truth about what he's doing. There's nothing wrong with what Saluzzi is trying to do. What's wrong is crying foul when the one-sided benefit you wanted to obtain is defeated by people who have made a bigger investment in what's important for trading. The big winners in all this are small traders who are not fleeced by large professional size deceptively getting them to buy or sell at insufficient prices. I can understand Saluzzi's campaign against technology and modern information processing, it's his ox that's getting gored. The thing that's truly hilarious is that what Saluzzi wants to be legal, his "tape reading", is just another example of where professionals have an edge over the little guy. But it would never occur to Saluzzi to outlaw what HE does, only what his competitors are doing better than him.
Let's get one thing straight: front running is when someone sees your order and executes ahead of it. It's illegal, and no one is advocating it. What many of these high frequency trading algorithms do, however, is not front running. High frequency trading is about using computers to do what human traders used to do - take advantage of available information (such as orders that are visible in the marketplace) and try to figure out where a stock is going. They do it better than humans can do it. They try to figure out what you want to do, and try to profit from what they think you want to do. They do it faster and more accurately. That's called trading - it's the nature of the beast.
If you put in a limit order to buy 100 shares of IBM at $80.50, and a computer or anyone else immediately bids $80.51 for IBM, that's NOT front running. That's someone willing to pay more for the stock than you are. It doesn't matter if that someone hopes to buy the stock and offer it back to you immediately like a high frequency rebate trader, or if they plan to hold it in their IRA for 50 years. Similarly, if, at the time you bid $80.50 for IBM, stock is offered at $80.51, and as soon as you put your bid in someone takes the offer, that's NOT front running. You had the opportunity to take that offer - you chose not to. This is the first key realization people need to make.
The next realization is that no algorithm can force you to pay more than you want to for stock. One thing the algo's do is "psych you out" - when you bid $80.50 and you see the $80.51 and $80.52 offers get lifted, you might get nervous and panic. That's your problem. If you put your limit order in and go to the kitchen to make yourself a sandwich, you'll probably find that your order is filled when you get back. Any algo that expected you to panic and lift stock from them lost on their gamble. You can defeat the psychological game by refusing to play it (place your order and don't stare at the screen) - or by lifting the offer in the first place, paying the narrowest spread you've ever paid due to the massive competition by various high frequency trading algorithms to maintain quotes on the inside market.
Let's talk about Chuck Schumer's proposed ban on "flash" orders. Flash orders are a little bit trickier. From the NY Times article:
When buy or sell orders are submitted to marketplaces like Nasdaq, they are sometimes flashed to a collection of high-frequency traders for just 30 milliseconds — 0.03 seconds — before they are routed to everyone else. In that half-second, (sic) fast-moving computer software can gain valuable insights regarding growing or declining demand in certain stocks and can trade ahead of other market participants, pushing prices up or down.
Although anyone can gain access to flash orders by paying a fee, they are useful only to traders who have computers powerful enough to act on the data within milliseconds.
First, note that ANYONE can gain access to flash orders. Similarly, anyone can gain access to co-located servers at the NYSE to have super fast execution speeds. These are not just available to a select chosen few - they are available to anyone who wants to make the investment in capital and technology. Now, the intent of flash orders is to allow participants in a given market center the opportunity to improve the current bid or offer so that an order doesn't need to be routed away to another market center.
An example: let's say GE is trading $11.45-$11.50 at DirectEdge, but that there is an $11.46 bid on ISLD (an ECN). If you submit an order to sell stock at $11.46 on DirectEdge, they flash this order to select market participants to offer them the opportunity to fill your order - otherwise the order gets routed out to ISLD and you (the seller) have to pay an extra fraction of a penny for the routing. As I tried to explain on some other posts regarding flash trading, this is basically a hyper-speed modernized version of how the NYSE specialists used to verbally quote orders to offer people in the crowd the opportunity for price improvement: "If GE was 11.25-11.27 50k up, and you walked in to sell 50,000 shares, the specialist would say out loud '25c bid 50,000, 50,000 at 26c, SOLD.' Anyone could say 'TAKE or BUY'EM' before the specialist said 'SOLD' which would result in the seller getting price improvement to $11.26 and if no one interrupted him, the trade was done at $11.25. Markets have NEVER been setup such that every participant has the same opportunity to trade on every quote."
The problem is if systems receiving flash orders can then turn around and hit that bid in front of the seller. That is blatant front running, and needs to be stopped. Here's a quote from a user of flash quotes explaining why he doesn't want them banned - a key point is that no one forces you to place flash orders:
responding to: "there are different forms of HFT. Flash is clearly frontrunning and anyone who says otherwise is delusional. its designed to cheat "
It is designed to keep orders from being routed-out under Reg NMS - which saves the person placing the trade money.
I am a small fry (sole employee of my small stat arb company) placing small to modest sized limit orders directly on ECNs. If I want to avoid paying a route-out fee when my order would otherwise become marketable on another exchange or ECN, the best execution + cost structure that I can get is to first have an order flashed locally to see if I can have a fill on the local ECN.
If a HF trader on the ECN gets the flash and 30ms later fills my order, I avoid paying a route out fee. If no HF trader responds to the flash, or the flash never existed because it got outlawed, then I end up paying a route out fee.
That is the intent behind flash orders. It keeps orders that become marketable more frequently on the local ECN rather than having them route-out....
I keep saying the same thing over and over again... but anyone who cares about orders flashing and doesn't want them to flash should just send the order through a different venue. It isn't rocket science. If you don't like the facilities provided by Direct Edge to execute your market moving large institutional order, then place it on Island.
Why is that so friggin hard?
No one is forced to place any order types that they don't like, or use any ECN that they think disadvantages them.
If your broker forces you to place trades on venues that flash orders and you have no control over it, then you need to get a better broker.
Outlawing an order type that some people actually like to use under certain circumstances is ridiculous.
More constructively - make flash default to off, so that a trader needs to explicitly ask for a flash, and make all brokers who don't pass on fine grain order control to their customers set the default to off...."
I'm somewhat indifferent when it comes to flash orders - on the one hand I don't care if they get banned if they are being prolifically front run, but on the other hand I like the suggestion of defaulting the flash order status to "off" and allowing traders to still flash their orders to try to get price improvement. If traders are getting worse executions on their flash orders because they are constantly being front run, then they will stop using them.
Let me take a brief detour to negate one blatantly erroneous insinuation made in today's Financial Times article:
Themis also suggests reintroducing the New York Stock Exchange's curb on program trading that would apply whenever the market was up or down by more than 2 per cent for a day. This constraint was removed in October 2007, which is - maybe not coincidentally - when world stocks peaked.
The NYSE never had curbs that prevented program trading. The NYSE used to have curbs that imposed restrictions on index arbitrage orders, a specific subset of program trading, after market moves of a certain magnitude. The curbs ensured that index arbitrage buy orders had to be executed on a downtick, and that index arbitrage sell orders had to be executed on an uptick. It's basically like the uptick rule for short selling, only it applied to buy orders as well. This had little or nothing to do with high frequency trading, and is completely irrelevant to the discussion, except for the ludicrous assertion that the removal of the curbs somehow may have led to the depression of the markets. If Themis thinks that high frequency traders are manipulating the market higher, they need to realize that the reinstitution of NYSE index arb trading curbs would have no effect.
While I'm on the topic, let me debunk a few more of Themis's blog posts. Saluzzi's assertion that high frequency traders were manipulating the price of CIT higher so that it would be eligible for a rebate was proved wrong before he even published it, when the price failed to maintain the $1 level. CIT was trading massive volume because it was in the process of basically declaring bankruptcy - not because HFT guys were manipulating the price. Second, Saluzzi's post titled "The Three HFT Horsemen" is perplexing. He alleges:
The three HFT horsemen are C, BAC and CIT. These three stocks traded 860 million shares today which is 10% of all US Equity volume. Think about that – 3 stocks in a universe of over 5000 U.S. stocks represented 10% of the volume. How could this be? Look at the intraday chart of all three of these stocks and you will see a something in common: an early morning move followed by a flatline with a very tight range (around .05). Meanwhile, while these stocks were flatlining the market was heading higher. The S&P 500 gained around 10 points in the afternoon (or 1%) but these 3 stocks did not move. There was a constant bid to these stocks yet anytime they wanted to lift there seemed to be a constant offer just a few pennies higher.
Never mind the fact that C and CIT are low priced stocks that should theoretically trade in tight ranges: what is the problem with this? As a trader, Saluzzi should be sending thank you notes to the high frequency traders for making sure the price of the stock didn't move much at all - this makes his job easy, regardless of if his client is a buyer or a seller of stock. There was ample liquidity all day, and the prices barely moved. This is a good thing.
I am a capitalist. I like competition in markets. Our competitive markets have resulted in evolution to the point where traders have written computer algorithms that do the job traders used to try to do by hand - profit from stock movements and from what they feel net supply and demand for a given stock is. This is the main reason I'm against most attacks on high frequency trading. SOMEONE will always be the best - the fastest - the closest - regardless of if you ban co-located servers, or install mandatory latency in order execution pipes. Would there be anything really wrong with making sure that all orders placed are valid for a 1/2 second or a full second? No, I don't think there would be - and that might be the kind of compromise that we move toward - but we need to recognize that the playing field will never be level. It never has been level and it never will be level - there is always someone smarter than you, and it would be a shame to try to legislate that edge away.
Full disclosure: No agenda here: I do not run a high frequency trading system, and a ban on HFT would not have much, if any, impact on my life. Market position: short the market.