Thursday, 8 December 2016

IEX innovation killing innovation

IEX gets far too much attention as it stands, but here is another reluctant meandering.

IEX is a small, expensive dark pool that was improperly licensed as a public exchange by an SEC influenced by the fan boys and girls of Michael Lewis' false and improper "Flash Boys" narrative which was written to support his friends who had invested in Spread Networks and the InvestorSexChange.

Phew, now that I've got that off my chest, most of the problem is not IEX but Reg NMS. The big mistake made by the SEC was giving protected quote status to IEX's delayed quotes. To me, the other main errors in the approval of IEX were not rule breaks but improper facets within Reg NMS. IIROC, the Canadian regulator, has handled things better than the SEC by not giving protected quote status to speed-bumped markets there.

Now there are two main pieces of somewhat old news to cover here:
  1. IEX's filing for its new Primary Peg (PP) order type to join its Discretionary Peg (DPEG) order type in fading, and
  2. an older filing changing the calculation mechanism for IEX's crumbling quote indicator.
Another argument, that I feel is important to understand, is how IEX's innovation, particularly PP and its DPEG, actually prevents innovation.

Primary Peg


You can read the PP IEX submission to the SEC here.

Basically with PP, you sit on the NMS bid or ask, NBBO, unless IEX uses its crumbling quote indicator to look into the future 350 microseconds and decides that you might be at risk of being traded through, or adversely selected, so it moves you one tick to safety. When the crumbling quote indicator says its safe to go back, IEX will put your order back to the NBBO. It is an automatic fade based on IEX's formula and their 350 microsecond look into the future. Remember the order is non-displayed, unlike CHX's LTAD, and thus will not interfere with the SIP feed.

It's a good order type for an HFT market maker. It gives you some adverse selection protection and lets you sit on the NBBO to graft out a living from the spread. Unfortunately for the market maker, your order will be prioritised behind displayed orders, but you don't have to worry too much about that as most of the activity, as you can see in the following chart, is not lit.

(click to enlarge)
IEX has some cunning bullsh*t in their application as it talks about reaching up to the BBO when the market is stable rather than thinking of it as a non-displayed quote fade. Schmarketing. Fading is fading. Go on. Admit it.

Marketing doesn't really matter as the effect is not a lot different to the DPEG in terms of NMS interaction. Thus, I can't see this order type not being approved. The SEC has already fumbled the ball. It's no worse than what already exists at IEX. Do you also think that it remains hypocritical of the "Puzzle Masters" to introduce yet another complex order type despite the "Flash Boys" pledge to only have simple order types? If I was forced to trade on this exchange, I would certainly be using such a PP if their hideously expensive transaction costs do not rule out such an application.

Innovation killing innovation


That brings me to my main problem with the PP. It is a good bit of innovation that kills innovation.

Normally a trader or broker would build their own algorithmic way of avoiding adverse selection. This is the usual activity you see in a market with little micro-structural avalanches of cancellations as traders avoid being the dumb bunny being traded through at best. It might be something as simple as saying, if there is only 10 left on the bid, bale. It could be what I'm used to doing, which is applying a machine learnt algorithm to the security or contract to determine your adverse selection criteria.

The logistic regression formula IEX uses for all stocks cannot possibly be perfect because all stocks don't behave in the same way. One size does not fit all. So their innovation is flawed from the outset. Perhaps good enough for some market participants, but not for others. The IEX speed-bump prevents you innovating yourself as you can't look into the future as IEX does. You're frozen out. You are forced to use their innovation which kills your pathway for any innovation you might have in mind. This usurping of brokers and traders is not healthy. No innovation for you. Like it or lump it. This is what I mean by IEX's innovation killing innovation.

It is kind of ironic that the SEC has asked for innovation and it is getting innovation that kills further innovation. It is the wrong multiplier to harness. You should be seeking to harness the multiplier of innovation from the many brokers and traders. Don't impede widespread innovation.

Parasitic darkness


So be it. I expect this order type will likely act to further darken the trading at IEX. Public price discovery will be increasingly impeded.

At some stage, the SEC needs to consider how much dark and parasitic activity should be allowed at a public exchange. I would argue that zero would be the appropriate number. However, parasitic trading is not necessary bad, as I have pointed out many times. Index funds are likewise parasitic, yet helpful to many. Perhaps a limited number, such as ten or twenty five percent, would be an acceptable threshold. That said, my gut feel is that if three quarters of the flow was parasitic the markets may remain functioning OK. Perhaps even ninety percent. My gut feel is not something I'd like to rely on. I do believe the role of public price discovery is too important to subvert in this manner. Zero dark, not thirty, should be the benchmark from a public policy viewpoint.

If you want dark, go to the dark corner: the ATS corner. Why be public if you don't have public pricing? We have to remember the beneficial role public markets play. IEX is not fulfilling that role.

Crumbling quotes


Back on August 9, 2016 IEX also filed an update to their quote instability factor (QIF) process that was used for DPEG. That QIF process will also apply to the PP. Now I've gone through this calculation in its old guise previously and this change to the formula just adds two more variables and changes the timing. Instead of the quote instability being triggered for ten milliseconds, it was changed to two milliseconds. The two new variables, are:

  1. a Boolean indicator that equals 1 if the last two quotation updates have been quotations of protected markets moving away from the near side of the market on the same side of the market and at the same price; and,
  2. the number of these three (3) venues that moved away from the near side of the market on the same side of the market and at the same price in the prior one (1) millisecond: XNGS, EDGX, BATS. 
The quote stability threshold changed from 0.32 to 0.6.

It's a reasonable formula they have extracted from their logistic regression but I suspect most of us could come up with a better one in our sleep. The biggest problem with this approach is that it treats all trading instruments the same. We all know that big caps and small caps trade quite differently. There are many other reasons to differentiate adverse selection criteria beyond IEX's approach. Why not a random forest or deep learning parameter set for each instrument? Tens of thousands of pages of appendices for the numeric weights in SEC filing appendices would be a fun way to send a perverse message to the regulator on the wrongheadedness of this innovation killing innovation.

Anyway, let's spell out the newish formula:

It's interesting IEX picks on XNGS, EDGX, and BATS in their formula. Make of that what you will.

Now that BATS has been purchased, there seems to be a vacancy in the market for a real exchange that supports improving price discovery and efficiency. IEX hampers both price discovery and efficiency. It is time the SEC thought long and hard about this issue. The SEC needs to revise their methodology for approving public exchanges.


SIP games


I wrote previously about the SIP games that were on the table at IEX:
"IEX - the good, the bad, and the ugly",   Thursday, 16 June 2016.
Particularly in the comments of that piece. The SIP times are improving dramatically and you may be gamed if you don't plan to have multiple sites beyond the necessary IEX co-location. Yes co-location is required at IEX despite the rubbish you may have been told by IEX staff. SIP gaming is yet another wrinkle to keep in mind when you co-locate to trade at IEX.


Bigger fish


Let's not get too carried away though by all the hypocrisy from IEX. There are much bigger fish to fry. The equity markets work pretty well, despite the need for modification. It is much worse outside equity markets. Why are we still paying outrageous spreads in foreign currency when we do foreign transfers? Some markets still live in the dark ages. We should pause and be thankful that equities are at least a little enlightened.

--Matt.

No comments:

Post a Comment