At the heart of our markets is the idea of fragmentation and speed. The faster and more fragmented our markets are, the more competitive they are. And competition is good ... right?
But, at some point, does a market become too fast, too fragmented and too competitive? Does the technology needed to trade in this market make that market too fragile? To answer this, we need to examine fragmentation.
Most think about fragmentation in terms of trading venues. And yes, the U.S. equity market has a number of trading venues. There are 13 exchanges, about 50 dark pools and many internalizing brokers that can match buyers and sellers. To ensure the best match, sophisticated order routers need to examine many trading venues. And without being in two places at once, the more venues checked, the greater the chance buyers and sellers miss each other.
While checking many venues complicates execution, markets also can fragment in other ways, including price and time. The more price points, the more missed trades. This is especially true as liquidity providers spread orders across more price points and venues. As fractions turn to pennies, and pennies to sub-pennies, through dark pools and the NYSE Retail Liquidity Program (RLP), trading volume becomes fragmented from 16 price points per dollar to 100, or 1,000.
The same problem exists with time. The quicker the quote timeframe, the more messages I need to produce and analyze. If I am quoting across 13 exchanges and 20 dark pools, then I need to manage my quotes across 33 venues. If my quoting frequency changes from seconds to tenths to tens of milliseconds to milliseconds, then my quoting rate increases from 33 to 330, 3,300, 33,000 quotes per second. Multiply that across 5,000 stocks trading at pennies or possibly sub-pennies and think about the massive technology and message infrastructure needed to trade.
Comments | Post a Comment
21 Comments to "Fragmentation Is Redlining the Markets":
Anonymous
10 August 2012
Like the point made here, does "reining" it in equate to turning the clock back and is that ever possible. The error of our ways are clearly for all to see, evidenced largely by the massive outflows from equties and the frequency of major market issues, black swans are supposed to be infrequent, seems we see one passing by almost weekly these days. The fear is that the regulators in their well intentioned endevours will create the same chaos in deriviatives trading. My own view is we need to put the handbrake on your McLaren, restrict the engine, slow the market down, what value is a microsecond fill to the average investor?
kurtkujawa
10 August 2012
I’ve said once and I’ll say it again, exchanges for profit have been the problem. Just as the argument by the HFT’s that they provided liquidity has been debunked; the argument that competition between exchanges provides better pricing and execution is also false. Its’ the buyers and sellers, regardless of where they execute, that create the prices, not the exchanges. So therefore, if all the buyers and sellers were in one place, there would be more competition amongst the buyers/sellers, which would create deeper and more transparent markets resulting in better prices for everyone involved, whether it’s the 100-200 retail investor or the 100k block institutional investor. As for the outflows from equities, no offense but that is cyclical, the money just moves around depending on risk appetite. Right now the 401k’s are in bond funds, when the market and economy change you will see money flow back into equities. It would be nice if we could get a CLOB in place for when the tsunami of cash returns.
Comments (171)
agrody
10 August 2012
I concur we have a problem and it is of our industry's own doing. What is fundamentally an engineering process to build a complex market infrastructure should be conducted not by regulatory committee's and comment letters but as a formal and disciplined system's definition through to system's implementation project. Imagine the NASA program run as we have run NMS1,NMS 2 and soon NMS3. Imagine this country's nuclear program run this way. A cooperative engineering process should be undertaken, not the committe-industry interaction in full public view, where posturing and politics rear its inevitable ugly head, as it has been to this point.
Comments (42)
csparrow
10 August 2012
Quantizing time would impose structure on the market. What is lacking today is co-ordination and synchronization across venues which leads to nefarious activities such as latency arb that erodes confidence, creates an explosion of data, and a race to minimize latency which benefits technology vendors, trading venues and HFTs but not natural investors looking to deploy capital.
If we allowed airplanes to land continuously, there would be crashes as they all race to land first. Imposing the structure of air traffic controllers that co-ordinate landings creates a safer system leading to travellers having confidence in the system. Allowing equities to trade continuously leads to flickering quotes, flash crashes and systemic risk leading to investors losing confidence in the system as a subset of traders game the system.
Having the market "tick" with all venues trading in a co-ordinated and synchronized fashion where information is disseminated equitably will restore confidence, reduce systemic risk and technology costs, eliminate latency arb and contain the explosion of market data.
We don't need to turn the clock back - instead we can borrow ideas from how a clock works. All clocks must "tick" because time itself is fundamentally quantized. Have all venues execute with the same tick and allow "equities controllers" (i.e. regulators) to control the pace of the ticks, stopping the market clock when necessary. Instead of reining in the market, regulators need to grab the reins, take control and impose order.
Comments (37)
kurtkujawa
10 August 2012
Csparrow, I would respectfully disagree. 1st understand the analogy; I just disagree with the premise. Markets should be based on price not time, so by regulating time, the fragmentation issue still has not been resolved. 2ndly, if you simplify things, CLOB, you won’t need more of the inept bureaucrats and regulators that got us into this mess. I for one can appreciate the beauty of simplicity, the fewer moving pieces, the fewer things that can break or go wrong
Comments (171)
csparrow
10 August 2012
I agree price is most important consideration. The problem is that with many venues all quoting and trading in continuous time, many problems arise. A CLOB makes a lot of sense but we would have to give up venue competition. IF (capitalization intended) we want comeptition amongst venues, then we should co-ordinate the quoting and trade matching activity.
Comments (37)
kurtkujawa
10 August 2012
That is my entire point; I do think we should get rid of all the venues. Although their argument that the competition would be good for price discovery sounded good to a neophyte, that fallacy has been exposed for what it really was, just another scam like the HFT’s providing liquidity. Once again, it’s not the competition amongst venues that creates price discovery. The bigger and deeper the market is, the better the price discovery. It is undeniable that auction markets are still the best and most transparent way to find clearing prices. So there fore if it’s too late to get rid of all the venues, we could at least force them all into 1 pipe that would display, let’s just say for arguments sake, every second. That way all orders would at least have to be posted and good for the amounted time. Personally I think you could do it every minute. I mean really, what investor needs to worry about every millisecond!
Comments (171)
kkhetan
10 August 2012
the issue is we have too many trading venues, too many price points and too many changes of orders in a second.
two simple solutions (are you listening Commissioner Schapiro?)
a) require a one second minimum display for every order. we will automatically reduce traffic by 90+%.
b) market makers can only use MPV equal to a quarter of what everyone does. so none of the .0001 nonsense.. you want to step in front then pay up .0025.
the HFTs abetted by the SEC have killed the golden goose in their greed for more risk-free profits making every trading venue toxic.
Comments (12)
ltabb
10 August 2012
I am not sure a one second minimum delay will stop this problem, it may only exacerbate it. Given that processors and technology only get faster, and clocks don’t synch exactly, a 1 second delay will only cause the HFTs or anyone else looking to cancel an order, to constantly ping the exchange with cancel my order messages. It will be like get me out, no I can’t, get me out, no I can’t, get me out… These messages will act as a denial of service attack on the exchange. That is unless the exchange manages the cancelations – in which case the exchange will need to take responsibility for getting folks out.
Think about the whole issue of quote stuffing, but coming not just from the HFTs but whoever is looking to cancel a quote within that second that is using an algo.
Don't disagree however about the quarters.
Comments (303)
anonymole
10 August 2012
Imaging an auto race track. On it we allow Indy500, NASCAR, home built hot rod and soccer mom mini-van racers all to complete together. This is the market. Could such a race ever be considered "fair"? Of course not. Logically each classification of competitor should be racing on their own track. But how could we create four different tracks each called "MSFT"? Separate them by maximum track speed perhaps? The soccer moms may only want to race/trade once a day or once a month. And the Indy500 HFT racers want the microsecond track no doubt. If tracks were separate by execution speed would there be a way to maintain balance between them? The Indy500'rs screaming by as fast (or faster) than the speed of light. The mini-van drivers trading on a paced market (continuous dutch auction) cycling at once a minute, or hour or day?
The bottom line is the different levels of racers here target different needs, but are forced to all race on the same track.
Comments (83)
timquast
10 August 2012
Larry, compelling perspective. Two observations:
Fragmentation in markets is not a natural event. "Markets" for all things tend to consolidate when they become commoditized and disaggregate during periods of creative disruption, if left to their own devices. Regulators have defined "competition" in the equity market, so fragmentation mutliplies in a commoditized market. It's an artificial market.
That leads to the second observation: Artificial markets that depend on unnatural behaviors and price controls (that's what mandated spreads, the NBBO, limit-up/limit-down, quote and trade rules, price-improvement, the LRB on and on it goes, are) will not be fixed through more controls.
If we don't want messy, risky, artificial markets, we should disconnect them from each other and remove the price controls. Why don't some poeple reading here test this? Set up a competing market and ask the SEC to exempt you from all the rules -- 605, 606, Reg NMS, mandatory spreads, and everything else. Make participation for securities and participants voluntary. Forget fairness -- just treat it like any other marketplace, be it a farmers' market or a shoe store.
Compare. See which thrives. We know which will thrive. So if we know, why do we persist in doing what doesn't work?
Comments (28)
timquast
10 August 2012
Sorry -- not LRB, RLP -- retail liquidity program. Typo.
Comments (28)
ltabb
10 August 2012
Tim,
I am not sure markets do want to consolidate. Markets want to consolidate for two reasons. One the cost to build out networks and bring folks together used to be high. I would argue that low cost networking, standards, and inexpensive matching technology allows the market to fragment.
Second, there is also the idea that a buyers market wants to consolidate while a sellers market wants to fragment. If you are a buyer you want all your products together so you can compare. If you are a seller however you want them to fragment cause you don't want your customers to compare. You want to offer a secondary value proposition other than price.
The more the market is controlled by sellers, The more fragemented. The more it is controlled by buyers, it would contract. Now sellers are not the sell side but the liquidity providers. The buyers would naturally be the price takers.
Bringing both concepts together - given shoppers get into a car and go to a place to shop they go to the mall (centralized exchange). Given they they don't they either go to the local retailer (where price is less important), or the web where who knows where the procuct is coming from.
Given this market, no doubt who has the control.
Comments (303)
cmackie
10 August 2012
I don't have answers, only observations:
1) who was hurt in the Knight debacle? Knight and Knight alone. It caused some unusual price and volume and this event should be studied so that corrective measures can be implemented that will make it harder for it to happen again. Yet, for all the hand-wringing, the Knight fiasco was software development 101: don't deploy code in to a live environment unless it is bullet proof.
2) kickbacks, whether you call them rebates, incentives, price improvements or market maker schemes, are an anethema to efficient markets. Take them away and many of the structural problems will disappear.
3) the concept that individual investors have lost confidence and abandoned the equity markets because price move too fast, the screen flickers, or they can't get MSFT at their bid price doesn't ring true to me. Investors have abandoned the market because the returns have been bad, we are in the midst of a global deflationary cycle, and they are moving on to new vehicles like ETFs. Heads up: they won't be back like they were in the past.
4) history strongly suggests that new technology always leads to improvementsand that markets left to their own will produce the most favorable long term outcomes. It seems to me that the difficult times that we have all experienced the last several years have caused us to turn on one another and, by extension, the markets. I believe that history will show that things aren't nearly as bad as they currently seem and that the prudent response is to take measured and thoughtful action in small steps.
That's my four cents.
Comments (26)
ltabb
10 August 2012
cmackie
probably a lot of truth in that - especially 3 and 4. Rebates are just about price fragmenation and sub pennies. If rebates were gone, we would just transition from an invisible fee (the rebate) to a more visible fee (the double sided fee).
Comments (303)
riati
10 August 2012
you are right on this, Chuck. this is more about poor imlpementation of technology than it is about market structure. Now there is a lot to be said for improving flaws in our market structure, but I think this issue turns out to be one where the party that caused the problem has paid considerably for its error. That is called appropriate consequence.
Comments (51)
mai
10 August 2012
I would like to suggest a more fundamental root cause for the issues that Larry has raised, namely, the failure of regulators to live up to their published mandates to ensure that markets, and the changes that they authorise to those markets, pass the dual tests of fairness and efficiency. This failure comes about because regulators have not adopted the most basic principle of evidenced policy making which requires one to measure ones progress against one’s goal…which requires them in turn to define and measure fairness (they have never done so) and efficiency (they have done some work in this area but it is inadequate) and build operational measures of the two so that those measures can be compared pre and post market design changes, such as, the introduction of regulations to facilitate market competition. How their political masters (and ultimately us) have allowed them to get away with this is beyond me other than to suggest that we have a tendency to miss the forest for the trees. As a result we are now changing market design at such a pace (in response to events like the Flash Crash) that the changes themselves although well intentioned (without the proper evidence of their effects on efficiency and integrity) may actually precipitate future problematic events. At the risk of being a little serving, can I direct interested readers to the following article which provides clues as to how we might begin to adopt the principles of evidenced based policy making which I accept is far from an easy thing to do. Evidenced-Based Policy Making for Financial Markets: A Fairness and Efficiency Framework for Assessing Market Quality, by Michael Aitken and Rick Harris, Journal of Trading, Vol 6, No 3, Summer 2011, pg.22-31.
Comments (11)
Martin Jermakyan
11 August 2012
Larry, this is an outstanding observation…
Most if not all market cataclysms of recent times have been associated with the stock market.
Indeed, we observe fragmentation (or segmentation) of stock markets more than of any other markets (except commodities).
With some periodic frequency, as earlier discussed and reported in various venues, we witness technological failures, or so do they classify them, driving markets into wide swings, and as most of the time, there is an army of people ready to blame HFT for these problems.
My initial thought was that there is a technological problem with the order routing applications, which explains why the problem is more apparent in stock market rather than futures.
Your article has led me to a further crystallization of an understanding that the technological problem is just a consequence of non-recognition of structural problems.
It appears that the issue is not much different from that of U.S. merchant power markets of late ‘90s and early ‘2000s. You might remember that the price of electricity at the Cinergy hub reached almost $7,000 per MW while in neighboring hubs it was priced at about 100 times less. Similar effects were observed also in California and elsewhere throughout the next half a decade. We had occasional such cataclysms in US natural gas markets as well.
When the same product, not meaning necessarily the same instrument, but perhaps meaning the same category of risk is traded in multiple venues, and many of the individual stocks might be viewed as belonging to the “same category” due to the very high correlation between them in our days – we live in a high beta economy – under such circumstances we deal with an additional risk, i.e. basis risk, which is a manifestation of “transmission” blockages between different trading venues, i.e. orders (aka distribution of risk). Orders do not flow freely between various venues to create equilibrium as not all of the market participants have equal access to all of these venues.
Such fragmentation of markets can lead to dislocation of risks and generation of risk packets, and make the markets susceptible to instabilities, whereby small perturbations make them enter into a regime of wild swings.
More than a decade ago, the unstable situation in power industry was generated by the violation of equilibrium between supply and demand of power due to the dislocated consumer and producer bases connected with each other with transmission lines which had limitations of transmission capacity.
Itlooks to me that current technologies are developed without the understanding of the inherent instabilities in order routing processes (just when we are talking about smart order routing) due to the underestimation of basis risk between different risk venues resulted from “transmission” limitations and blockages.
In this sense, the technological glitches we observe must probably be seen as signals of structural dislocation of risks between various venues.
Surely, this implies that the technological solutions for smart order routing should accommodate for basis risk and its premium when writing smart order routing applications. Who could expect 5 years ago that the stock market is going to replicate some of the problems that were observed only in energy and some other commodity markets in the past…
However, the fundamental solution of the problem will rest with the regulators – not in the sense of the creation of new barriers manifesting further blockages of order flow, but removal of these blockages. An efficient way off regulation of the market might be found through the proper pricing and trading off basis risk.
In this sense, the experience and understanding accumulated from the energy industry might be very useful for the understanding of what really is going on in modern stock markets.
sarnuk
14 August 2012
We have been outspoken against all of the artificial fragmentation introfucing artificial competition since Reg NMS was implemented. Addressing NYSE's RLP - ironically it does not introduce fragmentation. It increases it. And it increases it by segmenting customers within a trading venue. While this is done in nebulous ways in dark pools, it isn't done on public exchanges. This will of course change as the other exchanges ready their own "retail let me save you a tenth of a penny" programs.
This game is not about competition to bring the best prices to investors. It is about competition to insert the most intermediaries between buyer and sellers.
cmackie made four points in his arguments above. We don't agree with them all, but boy is number two sage and wise. If the SEC were to eliminate payment for order flow at all levels (selling and buying retail orders) and eliminating bizarre pricing models such as maker-taker, the market will heal on its own.
Think about it. Addressing those two issues is elegant and simple. The SEC doesn't need to get trapped into legislating small, and in manners which render its work obsolete. The SEC does not need to mandate spreads, mandate speed, mandate minimum order lives, or even mandate code certification if it eliminates maker taker and payment for order flow.
Maker-taker pricing is a pyramid marketing ploy. Even its inventor Island's Josh Levine, now criticizes it. I can't imagine the architrect of any free market system (whether for trading horses, grain, stocks, or carbon credits) devising this scheme and thinking it natural and efficient. It creates classes of partcipants that have no care or value for the qualities of the insturments in the market. With maker taker we have:
1) Participants flipping hard fast and furious to collect rebates. They don't buy because they feel a stock will go up, or sell because they anticipate price depreciation. At their level of volume and market share of total volumes this "market maker" activity by itself is price distortive. Ask James Angel - he wrote about it in one of his works.
2) It is this rebate game that mandates that first in line equals THE win. This rebate game is what has increased the speed logarithmically in the market, and made them unsafe. Without this game we would have never even thought about words such as microsecond and nanosecond in this industry. This game drives the algo and speed wars. Without this scheme, you would not have had the Flash Crash, the BATS IPO failure, the Facebook IPO glitch, or the Knight error.
With regard to payment for order flow... that this is even a discussion amazes me. How can a broker "sell" and order - pardon me - direct an order... to anyone? It matters not whether an equivalent execution can be obtained from the buyer of the order. It is a business model built on a huge conflict of interest for best execution. Eliminate this conflict of interest and the market will decide just how many trading venues are natural. A regulator's mandate will be unneeded!
Will online retail brokers start charging a higher fee to retail investors? Will it jump from $7 a trade with 100 free trades in the first year to perhaps $20? Maybe. But it will be $20 bucks with no conflicts buried in rule 606 reports, and it will lead to a natural level of fragmentation and innovation in the markets, and not a distorted one. For a price increase equal to the value of a NYC pack of cigs ($13), retail investors will have a cleaner, naturally innovative, and appropriately fragmented market!
Two simple regulations. Eliminate payment for order. Eliminate maker-taker. Folks who have business models around selling speed will not like it, but the owners of the market will.
Comments (160)
ltabb
14 August 2012
Sal - don't really disagree but rebates are just another form of sub-pennies and queue numping. If you don't have a minimum tick and you ban rebates - you will just see folks subpenny each other to jockey to be in the front of the queue. It will just be more transparent
Comments (303)
sarnuk
14 August 2012
We currently do have a minimum tick - 1 cent. We do not have sub pennies for buying items in the grocery store, the mall, nor should we in the stock market (price per share). It just needs to be enforced.
I have asked dark pool providers how is it that some folks in the dark pools get reports that are $0.00, $0.10 or $0.25 better than the NBBO? I always get an answer about how "our proprietary algorithm optimizes liquidity level and price improvement", or similar. And the HFT market makers and "Alternative Facility Liquidity Providers" always seem to get the bestest prints, and mechanisms to see order flow in the dark.
Isn't the rationale for the existence of dark pools the trading of large orders in a blind, price-unimpacting way? I am pretty sure the reasoning is not to trade 50 shares or 150 shares.
In my opinion, one should never see a sub-penny print on an exchange, and one should never see one in a dark pool unless it is the midpoint. But I still think that eliminating maker taker and payment for order flow will naturally take care of this.
Queue-hopping is a rebate-arb issue more than anything else, no? No rebate - No race to get in front of the line.
OK. Coffee now.
Comments (160)