Oh The Horror – Your Indices Are All Wrong!

I don’t dislike Joe Saluzzi and Sal Arnuk of Themis Trading – I just have a different perspective on market mechanics than they do.  I’ve actually had some intelligent and civil private email discussions with Sal which revealed some philosophical differences we have in our markets world view.

The good news about their latest white paper: “Phantom Indices,” is that it doesn’t contain any erroneous information.  The bad news is that the wider media is already starting to draw hysterical, fanatical conclusions from the paper, or at least hint that we should all be very very scared.  We need not be, and I’ll explain why.

First, a little history: Regulation NMS (Reg NMS), a handful of years ago, resulted in the fragmentation of the US equity markets.  This meant that instead of trading Proctor and Gamble ($PG), for example, almost exclusively on the New York Stock Exchange (NYSE), competing exchanges could quote markets in the stock (and other stocks, of course) and that traders must be protected with executions at the best price across all exchanges – the “top of the book” – the NBBO: National Best Bid/Offer.   Thus, if you send an order to buy 1000 PG, and the offer side of the “book” looks like this:

@ 63.41  1000  ARCA

@63.42  200  EDGX

@ 63.43 500 NDAQ

@63.44  10,000 NYSE

then your order will be routed to ARCA and you will buy the stock at 63.41.  The NYSE, although it is the primary exchange, doesn’t have the best offer, so your trade doesn’t go there.

Now, Themis, in their paper, points out a quirk of index calculations:  the index providers use only trades from the primary exchange to calculate the index values.  That means that for PG, our $63.41 trade would NOT be factored into the pricing for the S&P 500, the Dow Jones, the Russell 1000, and any other index that includes PG which only uses primary pricing.  Only PG trades done on the NYSE would impact index pricing.   To take another example: with CSCO, where the primary exchange is NASDAQ (NDAQ), trades done on EDGX or other marketplaces would not be incorporated.

Themis ran some numbers and presents a table, which you can see in their paper, which shows the % of shares traded on the primary exchange in each name.  In PG, for example, they conclude that roughly 30% of the shares are traded on the primary NYSE, and for CSCO they conclude that roughly 26.5% of the shares are traded on the primary NDAQ.  The next step, of course, is to insinuate that thus, since the index is “missing” between 2/3 and 3/4 of all of the trades, that the index values are thus “wrong.”  It didn’t take long for the Mainstream Media to pick this up and run with the question:

“It also makes you wonder who’s tracking (and profiting?) off the real intraday value of indices.”

That’s actually an easy question to answer, and the answer is: anyone who NEEDS the real intraday value of the indices tracks the “real” value of the indices.  But the good news is that it’s not “you,” the poor retail sap, who gets ripped off in this sequence.  You don’t need the zero-error index level, and frankly, you really don’t care.  Why don’t you care?  FT commenter “Slackful” summed it up simply:

“The fierce competition in the index and ETF arb markets results in prices that are generally extremely accurate, a wonderful service to those investors who don’t have the time to do it themselves” * (See footnote)

Ironically, this discussion becomes very similar to the discussion about high frequency trading and nano-second delays in data that Themis and I have written so much about previously.  Even more ironically, Themis’s prior papers have already, essentially, rendered this concern moot:  it doesn’t matter that your index data is a tiny fraction of a percent “inaccurate,” because it’s already “stale” anyway by the time you get it.

So, if I’m an index arbitrage trader who cares about every basis point, then I will have a spreadsheet which contains the components of the index that I’m trading, weighted by their index weights, which calculates the index value based on the prices at which I could buy or sell the underlying components – in other words, the NBBO inside market in each stock.  I would calculate the “index level” myself.

What readers need to understand is that the “index” level – what you see on the screen on CNBC, for example, next to “SPX (S&P 500 Index)” – isn’t something that anyone can trade.  There’s no such product as the “S&P 500 Index” that I can just say “Oh, I’ll buy that.”   The SPX is, of course, the weighted sum of the underlying components, which is what actually gets traded, and what gets calculated by the arbs.  So, if the SPX value on CNBC says “1270” and my calculated value of where I could buy the 500 stocks that make up the SPX comes out to 1265 in my spreadsheet,  well, I can’t magically go out and fleece poor mom and pop investors just because CNBC’s screen number is wrong.

But the news is even better than that – because the “mispricings” that Themis is talking about, although they didn’t calculate data on this, are miniscule – the VAST majority of the time.   In other words, it wouldn’t be an index level of 1270 vs 1265 – it would be more like 1270 vs 1270.05 (and no: I do not have the data on this.  But seriously – if you’re worried that you’re not getting complete index data, don’t complain to me about it – calculate it yourself.  Problem solved).  We’re talking about a matter basis points.

This is largely a result of another phenomenon that Themis has written much about: the furious pace of trading in our modern markets.  Think about it this way:  let’s say PG trades an average of about 400 shares every second.  Even though only 30% of those shares are on the NYSE, the NYSE trades occur frequently enough that they well represent the current price of the stock.  By the time the data travels up the pipe, onto your TV screen and into your eyes, it’s old anyway!  Alternatively, by the time you hit “refresh” on Yahoo! Finance or Google Finance, you’re not looking at the most current data anyway.  Which is all to say that if you’re now worried that your prices are 70% inaccurate because they are “missing” 70% of the trades, well, you’re barking up the wrong tree.

As I wrote on the FT’s comment section about the Themis piece:

“Oh Izzy… you breathlessly wonder “who’s tracking (and profiting?) off the real intraday value of indices” – while the answer is much ado about nothing: anyone who NEEDS to be accurate within a basis point is trading off of their own calculated index values. That means index arb guys, guys arbing SPY, etc…. It’s not last sale that matters, it’s bid/ask index value that matters to them.

what are the implications for the Common Man? almost none… Maybe the true SPX value is 1270.35 instead of 1270.40, but 99.9% of the time it will be 99.9% accurate. But the Common Man doesn’t trade SPX anyway – he trades SPY, which is kept in check by the guys in my first paragraph.

Of course, during times like the Flash Crash, everything goes out the window and the data is unreliable – regardless of if you’re using all exchanges or primary exchange – and the latency is a much bigger problem than the “missing” data”

Themis is correct when they mention that during periods of FUBAR pricing like the Flash Crash, the index pricing may have been severely wrong.  No doubt about that – but it was again a “secondary” or perhaps tertiary problem, because the data was late/outdated/stale and trading systems couldn’t be accessed anyway.

Saluzzi’s conclusion is that all trade data, not just primary exchange data,  should be used to calculate the index levels.  Sure – go for it – I applaud the mission – but let’s not panic in the meantime.  Even more complete data certainly wouldn’t hurt us, but it’s important to understand that the current data is not woefully inadequate.

If you need to know what’s in your CDO^2, you had better do the research on the underlying issues yourself instead of relying on an investment bank, CDO manager or ratings agency to bless it for you.  If you need to know if Panda Express is a fraudulent Chinese restaurant chain (please say it ain’t so!), then you need to do the research yourself instead of relying on what the company or other analysts tell you is fact.  If you need to know the value of the S&P 500 to within 1/100th of a percent, then you better calculate it yourself instead of relying on anyone else’s data feeds.  However, if, like the vast majority of all investors and traders, being within a handful of basis points is close enough for you, then you need not panic that non-primary exchange trades are being left out of the index calculations.


note: I have no reason to believe that Panda Express is a fraudulent Chinese restaurant chain

* ironically, again, while the implied fearful question was “who is arbing us out of our money?” the reality is that the arbs are now making it so that you don’t need to care about the data.  Let me explain:  25 years ago, when my old boss and his colleagues practically invented program trading, they used to do index arbitrage for upward of 3 index points – on an index that traded around 300!  HUGE, monstrous margins, because there was no competition, and very few people had the systems to enact the arbitrage.  It wasn’t even electronic back then – he tells me stories about how they had the indices broken down into groups, and each trader had a little stack of index cards.  When they wanted to arb the SPX, they’d call down to the exchange floor and say “buy 2 units of group 1!” and the brokers would run around to each post on the exchange floor and enter the orders.  Then, they’d run around and get all the executions and report them back. It might take 15 minutes.

Nowadays, the price difference being arbed is more like 15 index cents on an index level of 1300 – a SINGLE basis point: 1/100th of a percent – and it takes fractions of a second to send the orders, have them executed, and get the reports back electronically.  Competition amongst arbitrageurs keeps the prices in line, and makes it so that the retail investor can, 99% of the time, enter their SPY order at the NBBO without worrying about calculating the precise index value – because the arbs are doing it for us.

My former boss also mentioned that during the crash of 1987, the SPX futures were trading 30 points cheap to fair value – roughly ten percent!  I mention this merely as proof of two things: 1) markets have always crashed and will always will crash – regardless of market structure, and 2) price abnormalities have always happened and will always happen in times of stress, like the Crash of ’87 or the Flash Crash, and regardless of if the index data captures 30% of the trades or 100% of the trades (back then it was almost 100%)

Kid Dynamite is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.com. If you click on my Amazon.com links and buy anything, even something other than the product advertised, I earn a small commission, yet you don't pay any extra. Thank you for your support.

The information in this blog post represents my own opinions and does not contain a recommendation for any particular security or investment. I or my affiliates may hold positions or other interests in securities mentioned in the Blog, please see my Disclaimer page for my full disclaimer.

blog comments powered by Disqus
Kiddynamitesworld Blog