## Friday, 21 April 2017

### Reg SHO: 2017 Q1 Fails-To-Deliver = $131.95 billion This is a continuing story that no-one has ever cared about. I continue to meander around the subject as I can't help but slow down and strain my neck as I pass the car crash that is NMS's Reg SHO and its Fails-To-Deliver (FTD). I've previously covered the topic in June 2016: "Billions of shares fail to settle - every month - you and I don't care." I know you don't care. I can hardly blame you as the old back office and settlement details rarely excite. Us HFT types though, we get by only if we pay attention to the details. For my own curiosity I thought I'd check out what has been happening since mid 2016. After all, it is after 5PM on a Friday here in Australia - so, why not? Firstly, things don't quite seem so bad as the FTD share count dropped in 2016 to only a wee forty five billion shares - hardly noticeable:  Fails-To-Deliver Share count 2014 56,809,415,216 2015 58,823,475,247 2016 45,046,004,655 Q1 2017 11,694,519,023 2017 seems to be heading in roughly the same direction as 2016. If you squint and hold your head on the correct angle, there may be a down trend emerging in the following monthly chart:  Reg SHO Monthly FTD - data sourced from SEC(click to enlarge) Still, having around 4.5B shares per month failing to deliver for the last couple of months is not the happiest of circumstances. Let's dig into the 2017 Q1 data and pretend we care for a moment or two. So, the Q1 FTD share count is 11,694,519,023 fails. Who would miss such a small number of shares? To provide some perspective, IEX handled around 2B shares of lit volume in the same period, so it is more then five times IEX's lit volume. You'd think someone would care, but you'd be wrong. What's the dollar value? There are a couple of hundred thousand lines of stock and if we multiply quantity by price in the SEC's SHO files, we arrive at a grand total of:$USD 131,947,723,560

Which is slightly wrong. The figure is actually marginally higher. When we dig into the data we see around 5,000 of the more than 280,000 daily stock lines have no value and the SEC file simply reports a "." for the stock price for the symbol.

Let's look at one such line:
20170223  922101100  VPOR  2,231,979  VAPOR GROUP, INC. COMMON STOCK  .
Here we see 2.2M shares of VPOR failed to deliver with a price of "." VPOR seems to have gone through a name change and has become Newgen Concepts Inc with the same OTC stock code:

 (click to enlarge)
The main feature of interest is that the share price was less than one cent and closed at $0.003. In the picture above, Feb 23 is the second spike in volume traded in that patch of heavier trading towards the end of the month. So, the FTD was worth around$6,858 but we record it as zero due to the file's pricing issue.

Those 5000+ lines represent quite a few low priced shares. Of the 11.7B total FTD shares, 5.373B are low priced and generally OTC. 6.321B share have prices above 0.01 and are thus are properly registered in the SEC's files.

That is, the $US 131.9B of FTD value represents 6.321B shares with an average price of$20.87.

Let's look at the stats on those 6.321B shares in terms of daily value per stock of FTD:

 Count 274,673 Average 480,381.12 Std Dev 5,527,080.22 IQR 74,255.04 Min 0.01 Max 1,352,098,833.75 Percentiles 0.0% 0.01 1.0% 14.10 2.5% 40.37 5.0% 108.67 10.0% 373.36 25.0% 2,171.52 50.0% 12,098.00 75.0% 76,426.56 90.0% 572,307.98 95.0% 1,642,421.79 97.5% 3,511,242.92 99.0% 7,573,721.26 100.0% 1,352,098,833.75

The distribution is fairly skewed with the average daily line for a stock being $480,381 but the median being$12,098. Here is a picture of the distribution on a log scale so you can see the tails more clearly:
 (click to enlarge)
You can see the top 10% of FTD lines are over $500k, 80% are over$1,000, and 50% are over $10k. The biggest daily line was for 1.352B of SPY and here are the top fifty lines by value:  S.DATE SYMBOL QTY (FAILS) DESCRIPTION PRICE VALUE 20170322 SPY 5,784,875 SPDR S&P 500 ETF TR 233.73 1,352,098,834 20170223 QQQ 4,895,964 POWERSHARES QQQ TR UNIT SER-1 130.50 638,923,302 20170321 QQQ 4,774,824 POWERSHARES QQQ TR UNIT SER-1 131.81 629,369,551 20170124 SPY 2,595,083 SPDR S&P 500 ETF TR 226.15 586,878,020 20170213 IWM 3,657,902 ISHARES RUSSELL 2000 ETF 137.94 504,571,002 20170125 SPY 2,145,917 SPDR S&P 500 ETF TR 227.60 488,410,709 20170323 SPY 1,746,066 SPDR S&P 500 ETF TR 234.28 409,068,342 20170331 SPY 1,612,840 SPDR S&P 500 ETF TR 236.29 381,097,964 20170206 SPY 1,584,108 SPDR S&P 500 ETF TR 229.34 363,299,329 20170209 XLP 6,789,190 CONSUMER STAPLES SECTOR SPDR 53.36 362,271,178 20170125 TLT 2,973,254 ISHARES BARCLAYS 20+ YEAR TREA 120.31 357,712,189 20170208 XLP 6,401,409 CONSUMER STAPLES SECTOR SPDR 53.14 340,170,874 20170322 QQQ 2,482,350 POWERSHARES QQQ TR UNIT SER-1 129.81 322,233,854 20170106 XLU 6,583,463 UTILITIES SECTOR SPDR 48.68 320,482,979 20170215 SPY 1,288,863 SPDR S&P 500 ETF TR 233.70 301,207,283 20170217 QQQ 2,132,456 POWERSHARES QQQ TR UNIT SER-1 129.25 275,619,938 20170125 QQQ 2,186,774 POWERSHARES QQQ TR UNIT SER-1 124.18 271,553,595 20170221 QQQ 2,091,016 POWERSHARES QQQ TR UNIT SER-1 129.81 271,434,787 20170210 XLP 4,985,542 CONSUMER STAPLES SECTOR SPDR 53.58 267,125,340 20170109 SPY 1,166,234 SPDR S&P 500 ETF TR 227.21 264,980,027 20170216 SPY 1,125,050 SPDR S&P 500 ETF TR 234.92 264,296,746 20170321 XLF 10,487,104 FINANCIAL SECTOR SPDR 24.25 254,312,272 20170112 TLT 2,018,438 ISHARES BARCLAYS 20+ YEAR TREA 122.16 246,572,386 20170315 SPY 1,016,746 SPDR S&P 500 ETF TR 236.90 240,867,127 20170213 XLP 4,434,265 CONSUMER STAPLES SECTOR SPDR 53.62 237,765,289 20170223 IWM 1,700,034 ISHARES RUSSELL 2000 ETF 139.58 237,290,746 20170310 HYG 2,612,713 ISHARES TR 86.33 225,555,513 20170313 BWXT 4,855,127 BWX TECHNOLOGIES INC COM STK ( 46.12 223,918,457 20170207 SPY 964,292 SPDR S&P 500 ETF TR 228.93 220,755,368 20170301 ENB 5,206,097 ENBRIDGE INC 41.85 217,875,159 20170112 SPY 944,340 SPDR S&P 500 ETF TR 227.10 214,459,614 20170321 KRE 3,737,397 SPDR S&P REGIONAL BANKING ETF 55.73 208,285,135 20170320 KRE 3,639,657 SPDR S&P REGIONAL BANKING ETF 56.54 205,786,207 20170227 IWM 1,461,923 ISHARES RUSSELL 2000 ETF 138.65 202,695,624 20170315 HYG 2,276,326 ISHARES TR 86.11 196,014,432 20170207 BIVV 4,406,429 BIOVERATIV INC OM 42.83 188,727,354 20170321 XLU 3,588,888 UTILITIES SECTOR SPDR 50.89 182,638,510 20170202 ACWI 2,935,053 ISHARES MSCI ACWI ETF 60.92 178,803,429 20170228 ENB 4,265,507 ENBRIDGE INC 41.59 177,402,436 20170109 QQQ 1,413,347 POWERSHARES QQQ TR UNIT SER-1 121.93 172,329,400 20170316 IWM 1,249,139 ISHARES RUSSELL 2000 ETF 137.73 172,043,914 20170117 CLNS 11,417,165 COLONY NORTHSTAR INC CL A COM 14.63 167,033,124 20170131 TLT 1,394,371 ISHARES BARCLAYS 20+ YEAR TREA 119.27 166,306,629 20170201 TLT 1,361,902 ISHARES BARCLAYS 20+ YEAR TREA 120.10 163,564,430 20170126 EWU 5,005,998 ISHS MSCI UTD KINGDOM ETF 31.77 It seems unusual to me that QQQ and SPY would be so prone to FTD. By any reasoning that is a lot of value in those stock lines that are failing proper settlement. However, the basic tenet of Reg SHO is that you and I don't care. What's a few hundred billion dollars a year between friends? T+2 can only help ;-) Nothing to see here. Move right along. Happy trading, --Matt. _______ The SEC's Reg SHO FAQ & keypoints may be reached from the footnotes of my previous meandering: "Billions of shares fail to settle - every month - you and I don't care." ## Tuesday, 11 April 2017 ### IEX Signal or IEX 2.0  IEX: Dark Fader - come to the dark side On April 10, IEX kicked off a publicity launch for a branding exercise they seem to call IEX 2.0. The launch relates to their new crumbling quote indicator, or auto-fader, they're branding IEX Signal. That is, this update concerns yet another update to the IEX Dark Fader's formulae. I've written about IEX's Fader's formulae previously with: IEX Discretionary Peg (DPEG) calculation and patent; and, IEX innovation killing innovation. You're best advised to spend your time on more productive pursuits than reading this. I have little new to meander around that doesn't rhyme, even if it is not exactly a verbatim self-flagellating repetition of thoughts I've previously covered. As part of the IEX 2.0 launch, IEX released a whitepaper on their slightly sketchy - sans sanity - Signal, "The Evolution of the Crumbling Quote Signal." It isn't going to win any awards for academic rigour, but it doesn't have to. IEX is all about marketing, not reality. On the same marketing front, IEX issued a blog post, which contained the "IEX 2.0" sub-header, and a press release as part of their newly offensive new offensive. Be careful though, "Fighting Math with Math" may make you too feel a little like throwing up. A generous interpretation: IEX now acknowledge that their innovation, designed to stop your innovation, was flawed. Now, with this new newly flawed innovation, they hope to continue the stifling of your innovation and further deteriorate the role of public price discovery with more parasitic behaviour, both literally and figuratively. Oh, please :-| What wrath has the SEC wrought? Lewis' fictional "Flash Boys" and other IEX marketing is the pestilence. The IEX exchange approval process was the fire. It seems IEX is hoping their parasitically dark capture and subversion of the role of public markets with their hall of mirrors will become their flood. Good common sense is suffering from acute famine. ### Details Back on March 10 IEX put forward yet another revision to their crumbling quote calculation:  SR-IEX-2017-06 34-80202 Mar. 10, 2017 Notice of Filing and Immediate Effectiveness of Proposed Rule Change to Amend Rule 11.190(g) to Modify the Quote Instability Coefficients and Quote Instability Threshold Included in the Quote Instability Calculation Specified in Subparagraph (g)(1) for Purposes of Determining Whether a Crumbling Quote Exists Comments due: 21 days after publication in the Federal Register Additional Materials: Exhibit 5 On March 13, the SEC approved, amongst other things, the use of non-displayed quote fading via the crumbling quote indicator for the Primary Peg order type on IEX:  SR-IEX-2016-18 34-80223 Mar. 13, 2017 Order Granting Approval to a Proposed Rule Change to: (i) Amend Rules 11.190(a)(3) and 11.190(b)(8) to Modify the Operation of the Primary Peg Order Type; (ii) Amend Rule 11.190(h)(3)(C)(ii) and (D)(ii) Regarding Price Sliding in Locked and Crossed Markets to Simplify the Price Sliding Process for Both Primary Peg Orders and Discretionary Peg Orders Resting on or Posting to the Order Book; and (iii) Make Minor Technical Changes to Conform Certain Terminology See also: Notice: Rel. No. 34-79883, Notice: Rel. No. 34-79502 On March 29, IEX corrected some typographical errors to their presented formulae from the March 10 filing:  SR-IEX-2017-08 34-80331 Mar. 29, 2017 Notice of Filing and Immediate Effectiveness of Proposed Rule Change to Correct Typographical Errors in SR-IEX-2017-06 Comments due: 21 days after publication in the Federal Register Additional Materials: Exhibit 5 On April 10, as mentioned and linked to above, IEX launched the promotion of these initiatives culminating to this new IEX Dark Fader 2.0. In the "puzzle masters" continuing quest to simplify ordering types (sic), the quote instability factor, or QIF, has been added to the Primary Peg (PP) as well as being updated for the Discretionary Peg (DPEG). The Primary Peg will be an HFT order type of choice at IEX. Its judicious use will best enable the capturing of the spread. The PP's automatic non-displayed quote fade, enabled by the QIF, will assist in stopping market makers being traded through. However, PP will leave those poor dumb bunnies using IEX's displayed quotes to be hit. Suckers. That is, the naive passive participants at best who have chosen to use displayed quotes will be unable to move out of the way in time. Displayed quotes don't have the time cheat enabled by dark orders at IEX. IEX's speed-bump's poor QIF innovation disables external innovation, quashing the ability of market participants, whether they be funds, brokers, or traders, to improve their service to their own constituents. Buggy whips for everyone. It's just weird. It is sad that the SEC appears to have been so influenced by the pressure and marketing around IEX that they couldn't see the obvious flaws. The saving grace is that being a dark and expensive market impedes IEX's trade growth and potential tape revenue. Not many people should think they will be fulfilling their customers' best execution obligations by sending an order to IEX. If they do think that, then either they don't understand the obligation, or they don't quite understand IEX's real market structure. Regretfully the odour of the Order Protection Rule may force them to route to IEX despite it not being in their best ex interest. ### The New New Thing Anyway, here is the new newly stupid IEX QIF formulae which will now go by the schmarketing pseudonym of IEX Signal: Crumbling quote if QIF>{(0.39, if spread <=$0.01), (0.45, if $0.01 < spread <=$0.02), (0.51, if $0.02 < spread <=$0.03), (0.39, if $0.03 < spread) :} The variable definitions below are quoted from pages 33 & 34 of Exhibit 5 to the March 10 IEX SEC filing. Note that in this filing instead of including all the markets in the number of protected quotations IEX has chosen to incorporate only eight exchanges (XNYS, ARCX, XNGS, XBOS, BATS, BATY, EDGX, EDGA), thus N and F may range from 1 to 8. Three exchanges (XNGS, EDGX, BATS) still get a special mention, as per the last formulae's iteration, in the Delta definition. 1. N = the number of Protected Quotations on the near side of the market, i.e. Protected NBB for buy orders and Protected NBO for sell orders. 2. F = the number of Protected Quotations on the far side of the market, i.e. Protected NBO for buy orders and Protected NBB for sell orders. 3. NC = the number of Protected Quotations on the near side of the market minus the maximum number of Protected Quotations on the near side at any point since one (1) millisecond ago or the most recent PBBO change, whichever happened more recently 4. FC = the number of Protected Quotations on the far side of the market minus the minimum number of Protected Quotations on the far side at any point since one (1) millisecond ago or the most recent PBBO change, whichever happened more recently 5. EPos = a Boolean indicator that equals 1 if the most recent quotation update was a quotation of a protected market joining the near side of the market at the same price 6. ENeg = a Boolean indicator that equals 1 if the most recent quotation update was a quotation of a protected market moving away from the near side of market that was previously at the same price. 7. EPosPrev = a Boolean indicator that equals 1 if the second most recent quotation update was a quotation of a protected market joining the near side of the market at the same price AND the second most recent quotation update occurred since one (1) millisecond ago or the most recent PBBO change, whichever happened more recently. 8. ENegPrev = a Boolean indicator that equals 1 if the second most recent quotation update was a quotation of a protected market moving away from the near side of market that was previously at the same price AND the second most recent quotation update occurred since one (1) millisecond ago or the most recent PBBO change, whichever happened more recently. 9. Delta = 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 were at the same price at any point since one (1) millisecond ago or the most recent PBBO change, whichever happened more recently: XNGS, EDGX, BATS. From the whitepaper [p28] here is a description of IEX's results with this new false positive machine, "On our example day of December 15, 2016, our current formula resulted in about 1 million true positives and 975,000 false positives. This new candidate formula would have produced about 2 million true positives and 2.1 million false positives." Interpret that how you will. The IEX QIF remains a weak adverse selection criteria that tries to shoehorn in one set of formulae for all stocks. Many market participants could do much better if left to their own devices but they are prevented from innovating by the perversion that allows IEX to use future data in this way. I do feel we should all refuse to use the name IEX Signal and rather prefer IEX quote fading factor or just IEX's dark fader for short. If NYSE, BATS, or Nasdaq would like to send their data over, I'd be happy to provide them with a better performing quote fading factor. It's not too hard to do a lot better. Just sayin'... No news. Just more of the same. Enjoy your false positives and happy trading, --Matt. ## Tuesday, 4 April 2017 ### IEX trading IEX have long misled the buy-side about trading on IEX. They have always said they're inexpensive, simple, transparent, and you don't need co-lo thanks to the magic shoebox coil. The truth has always been that they're expensive, complex, dark, and you need co-lo plus more. I've written about this previously, most notably in IEX - the good, the bad, and the ugly. I'd just like to meander through a couple of updates to this story line. ### SIP games First, the SIPs, both CTA and UTP, have sped up considerably since I wrote that previous article. This means that your IEX trading is going to be at a big disadvantage if you don't have infrastructure at Mahwah, Carteret, and Secausus - along with the necessary RF links. Why? Because of SIP gaming. The SIPs are considerably faster now. There is a good chance that other traders will know about your trades before you, thanks to the IEX 350 microsecond delay, unless you're also cleverly listening to the SIP feeds. Here are the SIP CTA and UTP latencies for February 2017: Yes, it's an unholy mess for the IEX trader. Here's a lame little diagram I sketched up to show you the approximate latencies involved:  IEX verus the SIPs - click to enlarge The unknown, without measuring, is the extra overhead in the hand-off from the IEX engine to the SIPs. I don't know if IEX's internal data is handed off from NY5, NJ2, Carteret, or Mahwah. If you know, let me know and I'll update this information here. Remember those SIP hand-offs are before the 350 microsecond delay all IEX customers have imposed on them by the speed-bump. It would make sense for NYSE and Nasdaq to pick up the data at NJ2 and transport it themselves. They really should offer that to IEX to ensure the SIPs perform at their best. To beat some poor sucker dutifully waiting near IEX for their IEX market data, or trade info, you just have to be faster than that 350 microsecond POP delay. You may do that by listening to the SIPs. Faster SIPs have improved your chances of gaming that 350 microseconds pop delay, plus possibly the additional 15-20 microsecond engine to pop latency - depending on how the SIP hand-off takes place. It's not quite the easy, "don't need no co-lo" experience, is it? So, if the IEX feeds to the SIPs are reasonable, then traders at Nasdaq & NYSE will have IEX quote and trade data at those locations before IEX customers near IEX. Also, if those customers use their RF links to send the data back to IEX, those traders may still have the data at IEX before the IEX originating customer has their own data. Now you can see why you not only need co-lo at IEX, you need co-lo at Nasdaq & NYSE plus the appropriate RF links to trade optimally at IEX. Ugly. ### Still dark and expensive The other short update is that IEX remains, like a restaurant, dark and expensive. In March, only 19.8% of handled volume at IEX was lit volume. More volume was routed away than actually traded as lit. The "puzzle masters" seem to have out foxed themselves with their complex orders that their own customers may struggle to understand in detail. Remember, non-displayed orders cost 9 mill a side; as well as some displayed orders if you don't meet their zero - virtual rebate - criteria. Yes, dark and expensive.  (click to enlarge) I truly believe Brad and his team had and continue to have good intentions. The reality is just quite far from their investor friendly messaging. I don't think I'm deluded. If I'm not deluded, IEX staffers must be. Who is right? I wouldn't like to be an investor in IEX. Fortunately IEX have cash, lots of marketing goodwill, and plenty of time to correct course, even if it would involve eating a good deal of humble pie. Ego or road kill? The clock is ticking. Caveat emptor and happy trading, --Matt. ## Saturday, 1 April 2017 ### Sub-pennies rule! I'm not a fan of payment for order flow (PFOF). Wholesalers pay for orders to be sent to them from Schwab, TD Ameritrade and their kin. Mainly the retail trade. After your trade executes, far from any public exchange, your retail share statement shows a price with a few extra decimal places even though you specified a round penny, like you have to. It is sometimes a puzzle as to what may be going on and how. It's pretty opaque. In the SEC we trust. Similarly, discretionary or mid-point pegs, and some other dark orders may execute in sub-penny increments. But, but, but, there is a rule against sub-penny pricing. What's going on? Let's meander a little bit through the twisty passage ways of the NMS maze. The NMS Rule 612 "Minimum Pricing Increment" disallows sub-penny pricing for stocks above a buck. This was justified at the time so "meaningless" price improvement would not be used to step in front of prices and gain priority. I'm not sure about you, but I can imagine a lot of times that someone may think a half a penny extra on their trades would be nice. Are sub-pennies really that meaningless? Here is a FAQ from the SEC on Rule 612 which goes through some particulars and exceptions. There are a few exceptions to the sub-penny rule. For example, you can use a performance target on an order, such as VWAP, with sub-penny "targeting." Question 13 in the FAQ handles the sub-penny price improvement scenario for market centres: "Question 13: May a market center provide sub-penny price improvement, compared to the NBBO, in an NMS stock for which sub-penny quotations are prohibited by the Rule? Answer: Yes, provided that the execution does not result from an order, quotation, or indication of interest that was itself priced in an impermissible sub-penny increment. <snip>" Sub-penny trades are OK. Do those pennies matter? Are spreads really that tight? I just took a snapshot of the active stocks trading on BATS today:  (click to enlarge) The spreads on all the stocks are a single penny wide. You may have to click to enlarge the image to see them properly. Penny spreads are not universal, but they are common as you can see. We've come a long way from quarters, eighths, and nickels. Now, think about venues or systems offering sub-penny pricing. You have to ask yourself is it fair that PFOF and dark orders can jump the priority queue with sub-penny increments when regular stuff can't? Aren't public exchanges' displayed orders being disadvantaged? ### PFOF Let's meander around PFOF a little. PFOF is an essential piece of the NMS and retail would be worse off if it was eliminated today. Orders are price improved and thus are better than the best that could be achieved by hitting the National Best Bid or Offer. Those sub-pennies give customers better prices. PFOF is a blight on the NMS as it represents an abrogation of a broker's duty to, you know, do their job: broke. A broker has an obligation to best execution on behalf of their clients. How is it such that they can sell your order to a third party, get a fill back, and pocket that little payment bonus? The obvious thought would be: what if the broker could pull their finger out and execute just as well as the third party, and provide you with the benefits of the fee they didn't have to pay, and give you the benefit that would otherwise accrue to the PFOF firm. Wholesalers are not a charity. Why should they make money off you? It would seem you should be able to do better by keeping it all in house, right? Perhaps not. Executing transactions well is complex and expensive. Smaller brokers are disadvantaged, especially in term of technology, and outsourcing is perhaps reasonable there. Also, the sub-penny availability for price improvement means that an internaliser has a great deal of opportunity to take advantage of two sided flow and share the benefits between the broker (fees), wholesaler (spread), and customer (sub-penny price improvement). If your order, God forbid, went to a transparent, fair, efficient, public market, you'd have to trade at a penny price and get a worse fill! That is, the ability to trade in sub-pennies discourages trading on public market. Transparency is harmed. Price discovery and efficiency is also, arguably, harmed. That is a real conundrum. The customer is better off, but the market structure is poorer for it. ### Dark, parasitic orders We get to the same conclusion if we think about dark orders, such as types of mid-point peg, in a similar way. They are parasitic and rely on the efficiency of the surrounding market. They don't contribute to price discovery. However, you may get a price improvement that beats crossing the spread. It seems wrong to me that a publicly posted price can't compete with such prices that dance around in the dark. One way of looking at sub-pennies is that they destroy markets. They discourage posting of public prices. Transparency and price discovery is harmed. A good example of this phenomenon can be found at IEX. Let's have a look at a BATS execution quality report for one of the stocks referred to above, AMD:  (click to enlarge) You'll see the "Effective Spread" is$0.00429 at IEX. It seems a lot better than the other exchanges. The reality is that this table is not comparing like with like. IEX is mainly a dark exchange with only around 20% of volume sent there actually being lit volume. A good deal of the volume in that table is dark for IEX. I checked that by adding up the "eligible shares" volume for enough symbols to get to a point where the total eligible share exceeded the IEX lit volume traded but it is not immediately clear what exact criteria is used for the eligible shares, but some of the table's IEX volume is dark. You must also not forget that IEX's dark fees are excessively high at 9 mills on both sides.

For what it's worth, here is how the BATS do the "Effective Spread" calculation in their table:

Here is the strange way that BATS counts eligible shares in them there numbery things:

Yes, you read that right, trades above the NBBO mid-point are considered buys. Put that in your price improvement formula and smoke it think about the corresponding seller who didn't just get a price improvement.

The other exchanges are mainly lit, so the comparison is not really an accurate one. It's a bit silly that a bid filled at best is penalised by such mid-point effective spread formulae. The price improvement is also kind of problematic. IEX's statistics are especially fraught when you consider the timing of fills and the likelihood of IEX's quote instability factor preventing trades from actually occurring at that most important time to trade, when the price is changing! "Lies, Damned Lies, and Statistics."

I'm not fond of parasitic dark trading like that which IEX expensively provides. I feel that even though it is certainly useful, it should be consigned to the dark pools where it belongs. Public exchanges should be about transparency, efficiency, and price discovery which are all things IEX's performance, sadly, does not encourage. Hopefully the SEC will see the mistake it has made and fix NMS one day. Until then, the SEC will likely have to allow NYSE's speed-bump, and perhaps others, to extend the fiasco that is the hall of mirrors. At least NYSE's mid point speed-bumped clone is likely to put IEX's current approach out of business due to a more sensible fee structure.

### Sub-penny pricing needs to be encouraged

I think the answer is obvious. If you want better executions in a transparent market place that encourages efficiency, you have to support some degree of sub-penny pricing. Just how to do that is an interesting and big question to ponder.

I'm not fond of PFOF, but it is delivering results despite it being a weird perversion of best execution obligations. Whilst the public may want to kick out at those HFT wholesaling folk, they really are providing a beneficial service to Jane Doe on Main Street. The service goes beyond what the public market can provide thanks to the sub-penny rule limitations. That is, if you stopped PFOF immediately, the customers would be worse off, and it is quite likely the uneasily negotiated PFOF efficiency and benefit could not be regained under current NMS rules. Awkward, huh?

Also, I'm not fond of dark orders in public exchanges, however, just like those parasitic index funds, parasitic dark order types, such as mid-point pegs, do have some utility. I just happen to think that such orders have no place in public price discovery and should be relegated to the ATS / dark pool space. The SEC should be encouraging open, fair, efficient, and transparent price discovery in public markets.

Is it just me, or do you also find it weird that much of the efficiency of the market comes from working out how to go around the rules to avoid open and transparent pricing?

Despite the obvious profit motive, giving the market better prices is a critical but much criticised chore. Soul matters. Perhaps if you see someone from an HFT wholesaler in the street, such as Citadel or KCG, you might think about thanking them with a hug.

Just don't get arrested.