Thursday, April 24, 2014

Are There Really Good and Bad HFTs?

After writing a half dozen HFT posts all which remain unpublished in draft form, I will cut right to the proverbial chase in a last-ditch attempt to vomit my thoughts into the public domain: despite my predisposed feelings to the contrary, I reckon there is little practical difference between "good" and "bad" HFT and therefore market-structure should be leveled to reflect this.

Let me state at the outset: I am naturally sympathetic to bona-fide market-makers like Tom Petterfy & Timber Hill (one of the earliest and most thoughtful innovators using computerized technology for market-making purposes) chiefly because, in the main, they are attempting to a capture a portion of the bid/ask spread in exchange for risking capital by providing immediate liquidity. Likewise, I am naturally suspicious of predatory HFTers who have no natural raison d'ĂȘtre to transact and are, at best parasitic, and worst, criminal. These participants in the market ecosystem, like frigate birds, prey upon bastardized market structures,  ultimately disadvantaging, on-average, bona-fide buyers and sellers evidenced by the extraordinary profits derived from the undertaking.  In the increasingly vocal debate following Lewis' PR onslaught, and Schneiderman's attention, for the record, I think Dave Lauer (@dlauer) has it right, "@ModernMarkets" is no more informative than Radio Pyongyang, Craig Pirrong (@StreewiseProf) is thoughtful with a usefully wide-reaching analytical framework but off-the-mark in estimating the ratio of nefarious activity, while the NANEX/Themis (@nanexllc & @joesaluzzi) nexus has consistently been the most vociferously transparent shiners of light into the darkest most contentious corners of both structure and abusive practice - a position sadly not helped by their too-often overly-strident timbre and their adoption by the tin-foil hat brigade as anti-establishment marauders.  

On the basis of the preceding, one would have thought I would support advantages to bona-fide HFT market-makers and desire to impose restrictions upon predatory HFT. Both views suffer slippery slope arguments, and they are, despite what I consider to be divergent virtues, in essence, more the same than different.

Dr. Pirrong makes an important contribution highlighting the drive that electronic market-makers pursue to combat adverse selection resulting from informed trading.  Informed trading can range from immense orders, for quantities and a time-scale larger/longer than even the best capitalized bona-fide market-makers can supply, to private possession of material non-public information. Bona fide market-makers pursuing spread-capture strategies, typically have no view, and will always be on the wrong side of informed trading which is why Petterfy protests so loudly when there is obvious malignant activity before a high-profile takeover.  The depth and breadth of private material non-public information being acted upon in the market, remains underestimated. Whatever MMs capital and time-frames, their spread-capture, must exceed the drag of this adverse selection (being on the wrong side of liquidity supply that never comes back). The more informed the trading, the greater the necessity to avoid adverse selection by widening spreads, diminishing quote size, narrowing stops, or outright limiting or eliminating exposure where one estimates the presence of informed trading. And make no mistake: this is highest in the equity shares of public companies. In its simplest distillation then, bona-fide market-makers seek to deploy their capital, to the greatest extent possible, only to random or noise-like activity to capture some of the spread.

Predatory HFT (of which there are many flavors and time frames), in its simplest distillation,  seeks to avoid the deployment of capital in bona-fide market-making activities where they are exposed to adverse selection, focusing instead on strategies that commit as little capital as possible to pay lip-service to market-making for structural advantages (latency, rebate, order information) that allows either a risk-less arbitrage between venues, or high-probability predatory bets that for all intents appear indistinguishable from front-running.  On average, this intermediation increases the cost of trading, though likely moves the price more rapidly in the direction where its going in the presence of informed trading. Some term this movement greater efficiency, but as highlighted in Stiglitz' argument against HFT, it taxes information, transferring wealth from those with information (however appropriated) to the (free-riding) frigates who are adept at spotting others heavy with information.    

Bona-fide market-makers attempting to avoid adverse selection creates a slippery-slope. Where is the line and what is time frame? There appears no possible line of demarcation between limiting or avoiding adverse selection and using the same information for positive selection.  Just a smooth continuum. Why, one might ask, be content just limiting negative exposure? Why not, where one suspects the presence of PIN, warehouse or positively discriminate one's markets to get on the right side of it? Few, if any, in my two and one-half decades of experience - not specialists, not jobbers, not market-makers, not brokers, have shown themselves able to avoid using the possession of an advantaged position for private profit. So despite my distinction centered on intention and virtue, I see no reason to confer advantage to someone rather likely to abuse them.    

Predatory HFT are cynically trying to do the same thing using their conferred advantage within the markets as presently allowed and structured. There are many calls for all manner of outright restriction and regulation. The slippery slope centers on what, precisely, is "bad". Is it short-term speculation? If so, how short? Yes, certainly some things are illegal (quote  stuffing, spoofing etc.) But the essence of the most egregiously blatant abuse (with respect to the market and welfare) is the abuse of a structurally-privileged position - be it proximity, latency, or order information.  Some argue such abuse of privilege has existed for as long as markets have existed. And this is true, but so what. This is not setting the bar very high. We cannot, and with high likelihood, should not prevent speculation at almost any time frame given its slippery slope.  But we have the means to be better, and fairer and we certainly shouldn't positively promote and effectively subsidize the activity whether by selective designation and privilege, promotion of fragmentation or similar.

There are undoubtedly people who understand the minutiae SIPs and regs better than I. Like Dave Lauer.  Fundamentally, I am certain technology is good for markets. It is both democratizing and empowering. More importantly, I believe that bona-fide activity and integrity are good for markets; that if you enter the market to buy are sell a quantity, you should be willing to trade at the price and in that quantity, whatever the next quote-change or print. I am in favor of punting at whatever timeframe - whatever its utility. Without conferred structural advantage, guessing games can be minimized by execution games (as they have for time immemorial).  I think fragmentation is bad for markets. It may be good for certain brokerage participants, sponsors and exchange shareholders, but  I have yet to see a convincing argument how this fragmented structure can possibly deliver utility, fairer, more equitably, more efficiently, with more integrity than a single, well-run venue where everyone is equal before the rules, and the rules are laid down to best promote trade in as frictionless, cheap, and transparent mode as possible. Yes, perhaps back then, it was easy to do something better and fairer than the NYSE or NASDAQ market-making club, when neither exchange, nor their members, nor captured regulatory or oversight body had any concern for something one might term the Public's Interest.

There... its out. This was more for me, and now I feel better. It really should't be contentious. But sadly, most participants seem to put their private agenda's before the public interest. Nothing new about that, in America at least…..


Macro Man said...

Cassie, apologies for not touching base offline but I've been surprisingly busy. I tend to agree with what you've written, and would make one further point: much of the commentary defending HFT uses 'tighter spreads' as a clear benefit of the practice. However, the vast, vast bulk of the spread compression was the result of decimalization, which certainly does not require machines to narrow spreads to much tighter than 12.5 cents wide.

Mercury said...

1. HFT isn’t really good or bad so much as it is the natural and obvious product of how the market is currently structured. In other words, it’s a symptom, not the disease itself and the cause of the disease is unequivocally Reg NMS. Reg NMS is a federal agency regulation which has been blessed by multiple and various other government entities charged with protecting and promoting The Public Interest so I suppose taking issue with Reg NMS is “anti-establishment” but…so what?...that’s what the picture looks like when you connect all the dots.

First of all:

“I have yet to see a convincing argument how this fragmented structure can possibly deliver utility, fairer, more equitably, more efficiently, with more integrity than a single, well-run venue where everyone is equal before the rules, and the rules are laid down to best promote trade in as frictionless, cheap, and transparent mode as possible.”

Amen. At some fundamental level market fragmentation is antithetical to what a market actually is: a centralized hub that aggregates buying and selling interest and thereby facilitates price discovery and liquidity (my definition anyway). Market fragmentation diminishes this by definition even if it also produces other desirable results or is initiated with good intentions.

“Fundamentally, I am certain technology is good for markets. It is both democratizing and empowering.”

Technology had been doing wonders for increasing market data dissemination and decreasing transaction costs before Reg NMS which was ostensibly implemented to further reduce costs for (especially retail) market participants by promoting competition among a greater number of (fragmented) liquidity hubs. This was totally unnecessary as now evidenced by the fact that the current gigantic collection of exchanges and ATSs offer little to no differentiation beyond an assortment of ways for premium paying clients to gain privileged positions within the market microstructure. It doesn’t matter if this was the result of an honest mistake, a fuck-up or a conspiracy but the current results were perfectly predictable. Yes, Mom & Pop can buy 1-1000 shares of XYZ usually at or a penny above the last print for a $9.95 commission but that was pretty close to the case ten years ago too. Reg NMS has accomplished nothing in this regard that technology advances wouldn’t have accomplished by themselves. But if Mom & Pop have their money in some pooled investment vehicle instead, they’re probably paying a sizable vig to HFT.

Technology has also brought us to the point where information asymmetry is effectively zero. The last remaining barrier to zero latency is the speed of light and therefore, the physical distance between point A and point B. With or without Reg NMS and even without the “high frequency” part, market participants whose hardware is located in the same building as an exchange’s matching engines are able to react faster to market developments than other market participants who are hundreds or thousands of miles away – although many HFT strategies are built around positioning within the market structure so as to always being first in line regardless of speed. I think it’s possible to level the playing field in this regard by say, replacing continuous trading with regular crosses every 1 second…or whatever time frame. The gate opens up and bids & offers flood in, prioritized by price and ten by *randomized* time stamp. The gate closes, marketable orders are matched, next cross. While we’re at it let’s keep increments to a penny minimum, get rid of those hundreds of funky order types, fast/slow quote feeds and…well you probably don’t even have to worry about quote stuffing at that point. Something like that: simple, transparent, highly functional and robust, the way a public utility ought to be.

Mercury said...

2. The old, NYSE style boys club was an outdated model that had outlived its (once enormous) usefulness but what we have now is bullshit and even if it isn’t the general perception is that it’s bullshit and it’s certainly far too complex to adequately demonstrate otherwise to the broad investing public – which amounts to the same thing. I’m not sure the role of market maker is even necessary these days as technology has advanced to the point where all market participants can pretty much have the same info at the same time anywhere in the world. Mere volume is not the same thing as Liquidity and as these episodic HFT hiccups have shown, things can and do dry up in a hurry with the system we have now.

In conclusion, it’s all about the needlessly complex, fragmented market structure that Reg NMS hath wrought, not HFT per se. But it’s fixable.

RCJ said...

Great overview - I am interested in your thoughts on the role Exchanges are playing in this new ecology. It seems that they got off a bit free in Lewis'book. My impression is that once they went private (and stop being mutual organizations) that their incentives greatly changed and that they have aided the nefarious HFTS by offering very strange trading rules and order types.