Daily Decoupling Follows BP News

You know it, you love it, it's here. The daily EURJPY-ES decoupling appears like clockwork the second there is any BP news. ES takes off like a bottlerocket, even as E&Y (representing EURJPY, not an incompetent and allegedly criminal Repo 105 specializing auditor). Sell ES here, buy EY. Rinse. Repeat.

Comments

[ZeroPower]

Futures this morning were a joke too. Super bearish reports on housing followed by a 5 minute drop, then back up to gains and then some.

Weeeeeeeeeeeeeeeee goes the market!

EURUSD pumping it too.

[firstdivision]

Robos are fighting to keep the ES up today.  There is no way this will close up for the day (and by no way, I mean a true reason to be up).

[IBelieveInMagic]

Option expiry week -- there is no way they are going to risk not collecting their premiums.

[Bankster T Cubed]

completely phony market

controlled by a single computer program

must get long!!  before the big crash

the crash that will take all of 20 minutes

that's right, climb aboard the "rally"!

sick sick sick evil bankster farce

if all goes according to their plan, there will be no survivors

[LeBalance]

In 20 minutes all paper wealth will cease to have value.

And at the end of the 20 minutes, the bombs will fall in Iran, signalling the beginning of WWIII.

This may be touched off by a massive FF op in the US in which a major population center is sprayed by faux Islamo-fascists (Blackwater).

As Marc Faber said,"The cities will be gassed. You won't be able to get home. It is better to have a small place in the countryside, well away from everything."

As Hugh Hendry said,"I would panic."

These gentlemen are not crying wolf, it is all too apparent from history exactly what happens next.

Be calm and panic at the same time.

Namaste. Om namah Shivayah.

(caveat: I am a fool, don't listen to me.)

[plocequ1]

The machine is working. Repeat... The machine is now working. Over and out.

[jkruffin]

And good old Benny Boy takes down 18.4 Billion of the 25 billion in the 56-day auction this morning.

[Cursive]

I have seen nothing like this equity market.  Not even the Greedspan Y2K pump was this unbelievable.  This market snacks on kryptonite and shits (paper) gold.

[Boilermaker]

I know I'm probably naive...but how many times can you flip a double headed nickle before the crowd (sheep) realize what's going on?  I know MSM controls the vertical and the horizontal...but...even the most rubish of rubes eventually realizes there is no peanut under the walnut shell.

This is just laughably bad / sad now.

They should seriously just come out and admit (and spin as positive) that they are propping the market.  This isn't even feasible anymore.

[Cursive]

@Boilermaker

how many times can you flip a double headed nickle before the crowd (sheep) realize what's going on? 

Brilliant analogy.

They should seriously just come out and admit (and spin as positive) that they are propping the market.  This isn't even feasible anymore.

Exactly.  BOJ openly admits that it buys equities to prop up the market.

[Xibalba]

Rahm: Hey F... you Loyyd.
Loyyd: thanks. Hi to you to sir!
Rahm: rip the cord.
Loyyd: Yes sir!
Rahm: F... you.

[TheSettler]

Spell my name correctly, it is Lloyd!

[Xibalba]

shhhhh.

[Eric Cartman]

Sweeet!

[legerde]

Greetings,

I love this site, so please dont attack the messenger.   In fact I am sure I am confused somehow, but I was confused a bit about all this decoupling/recoupling talk so I started to look at it as close as I could and I am still confused.

What I have been doing is looking at all of the charts that get presented to understand how the trade could be accomplished and how I could watch the charts for myself.  What I have found is that the left side of the chart is scaled in relation to the right side of the chart.  No problems here, if the scale is 1:9 then it tells you how many "units" of each side to buy.   Put half the money on each side of the trade.  Like a bookey.. :)

But what I have found is that over the course of these posts, the ratio/scale has appeared somewhat arbitrary.   Here are my data:

06/16/2010 - 111:1090 and 113:1108  (Giving ~ 1:9 ratio) - Daily Decoupling Follows BP News
06/15/2010 - 112:1093 and 113:1108  (Giving ~ 1:15 ratio) - "Wax On": Intraday Decoupling Accelerates As Computers Are Once Again On Their Own
06/15/2010 - 112:1093 and 113:1102  (Giving ~ 1:9 ratio) - "Wax On": Intraday Decoupling Accelerates As Computers Are Once Again On Their Own
06/13/2010 - 110.40:1072 and 111.60:1088 (Giving ~ 1:13 ratio) - Decoupling Spread Closed, Or 6/6/6.5
06/11/2010 - 110.40:1071.50 and 111.40:1084 (Giving ~ 1:13 ratio) - Here Comes 6/6/6
06/11/2010 - 110.40:1071.50 and 11.40:108 (Giving ~ 1:13 ratio) - Intraday Market Commentary
06/10/2010 - 109.8:1065 and 110.8:1086 (Giving ~ 1:21 ratio) - 5/5/5
06/10/2010 - 109.8:1066 and 110.8:1085 (Giving ~ 1:19 ratio) - It's That Decoupling Time Of Day Again
06/09/2010 - 108.5:1042.5 and 110.5:1076 (Giving ~ 1:17 ratio) - And Scene - 4 Out Of 4 in 4 Days
06/09/2010 - 108.5:1042.5 and 110.5:1076 (Giving ~ 1:17 ratio) - Decoupling Trade #4 Is Here: ES Now 10 Pts Rich TO EURJPY
06/08/2010 - 109:1050 and 110:1067 (Giving ~ 1:17 ratio) - FX-Risk Spread Recouples Like Clockwork
06/07/2010 - 109:1057 and 110:1068 (Giving ~ 1:9 ratio) - FX-Risk Decoupling Trade Is Back: Third In Three Days
06/04/2010 - 110.50:1083 and 113.5:1102 (Giving ~ 1:6 ratio) - Short-Term FX-Risk Decoupling Collapses As Expected
06/03/2010 - 112:1073 and 114:1103 (Giving ~ 1:15 ratio) - It Is Time For The Recoupling Arb

So, what gives?   The ratio is sometimes as low as 1:6 and sometimes as high as 1:21.   I have the feeling that if you can adjust the ratios each day, correlations can be shown among many things.   But as you can tell from my newbieness, maybe this is acceptable in quant land.   Each day is a different day, so holding the same ratio may not make sense.  But it seems to inject more risk back into this seemingly risk-free trade because the ratio is eyeballed and important. :)

Can anyone help me understand more?

Thanks all.  Love this site..  I learn so much.

[KidDynamite]

Tyler's charts show the existence of a divergence. if you want to evaluate the magnitude of the divergence, or figure out the proper hedge ratio, you have the problem you noticed: you need to have a consistent scale.   I think this is what Tyler was trying to show yesterday when he included an indexed chart below the first chart.

[Eric Cartman]

There's one thing I've learned since joining this site - Tyler Durden is always right. 

[firstdivision]

Here we go folks, looking like a possible late day reality check :)

[Borat]

If you could manipulate the market, wouldn't you do the same?
Just pump it as much as you can, going net short on ES.

[Joe Shmoe]

Overview of HFT 

Forgive me for the long post (it's 2 pages in Word).  I’ve been reading a lot about HFT and have been taking notes.  When TD posted that exceptional interview with Themis, I thought I’d organize my notes and post them.  Basically, here are some key points about HFT as I understand it.  This all may be old news to you (or arguable), but I nonetheless want to offer this to the mix.  I believe most investors, whether institutional or retail, don’t know much about HFT.

HFT Defined

Robert Iati, a partner at TABB Group, defines high frequency trading as “...fully automated trading strategies that seek to benefit from market liquidity imbalances or other short-term pricing inefficiencies. And that goes across asset classes, extending from equities and derivatives into currencies and a little into fixed income."

Basically, HFT is: using super powerful computers (i.e., on the order of milliseconds or less) and math PHd geeks to develop and implement equity trading strategies.  This approach grew out of the algorithm, quant, and program traders of the 1980s.  But it really come into its own in recent years.  The strategy began in exchanges such as NASDAQ and only recently, due to regulatory changes (read, unintended consequences of government intervention), has it come to NYSE.  HFT has found a fertile landscape in NYSE listed stocks, given the features present for most blue chip type equities (such as high number of shares available, liquidity, etc.  more on this later).  

HFT platforms are owned and operated by the large broker/dealers, independent HFT firms, and to a smaller extent by hedge funds.  There is also a great deal of cross-ownership between various exchange platforms, HFT firms, B/Ds and funds.  They use several different fields of math in order to build their models.  For example, linear regression analysis, game theory, pattern recognition, and neural network analysis are used. 

Currently, according to a TABB Group study, HFT trading accounts for over 70% of US equity trading volume.  Who knows if this number is accurate.  For any of us who watch the markets, it is clear that the impact of HFT is a (if not the) dominant force driving equity markets in 2010.  

HFT Profitability Model

This sector is based on tiny margins from incremental asset gains/losses and rebates, so extremely large volume is required to make money real money.  Volume can be on the order of millions of shares traded, essentially immediately.  Also, high levels of leverage are required to amplify those margins.  10X to 30X leverage is not uncommon.  HFTs see themselves as not managing high levels of risk because the norm is that they finish flat each day (i.e., don’t carry positions overnight).  

Trading Strategies Used

Several math-based strategies are common among HFTs.  Here are some categories I’ve read about:

• Market Making:  They place, and rapidly edit, limit orders just outside the bid/ask spread of stocks in order to arb these disparities.  Remember they do these edits and adjustments many times per millisecond.
• Statistical Arbitrage: Many traders are familiar with convergence trades, where two securities deviate from their historical pattern and as the deviation corrects itself, money can be made by shorting the one that is likely to come down and going long the one that’s likely to revert up.  Statistical arbitrage is the same, except they do it with 4 or more variables, at the speed of light.
• Low-latency Trading: Provide bids/asks to their own algorithmic trading platforms much much faster than anyone else.  Kind of like a first-mover advantage, where milliseconds matter.
• Liquidity Detection: This one pisses a lot of people off, and may be regulated out.  They sniff out the potential for large blocks coming behind small trades.  Nothing new or wrong about that, except that there is evidence that the HFTs have access to superior data (and get it earlier) than is more broadly available.  So, by pinging immediate fill or cancel orders (often in dark pool trading exchanges that most people don’t have access to), they can identify when a big seller, say a mutual fund, is making a move.  Then they can immediately buy up all offered shares for that, and resell them at a higher price than would have been available if they weren’t there.  Some say that specialists have done this for years, but the fact that it’s now done at speed of light seems to make an unfair practice borderline criminal (sorry for editorializing...).
• Flash Orders: Apparently, (according to Themis) HFTs have superior access to data streams provided by exchanges.  Because of this, they can act accordingly, milliseconds before anyone else... which is all the time they need.  This is a hotly contested area, though regulators do not seem to understand or have appetite to address it.

Assumptions/Requirements

For these HFT models to operate, they assume that:

• There is enough liquidity in the market to meet their trading volume.
• Correlations must hold, or at least be predictable enough to be added to and updated in timely fashion.
• Data availability.  HFTs need to capture, cleanse, store, and analyze data on an intraday (real time?) basis.  Algo trading firms often use end of day data such as variable weighted average price (VWAP).
• Colocation is a big issue.  Because data transmission can get bogged down as it goes through servers and routers, they locate their servers as close as possible to exchanges.  
• Regulatory environment continues to provide advantage climate for operation (such as, rebates aren’t adjusted).
• Number of competing firms doesn’t get so large that available margins disappear. 

Issues/Questions

• Is there a bullish bias in these activities?  If so, what does that mean for strong bear markets (flash crashes)?
• Speaking of the “flash crash” of early May, was that an anomaly or a tremor along a fault line?
• It appears that all of this activity has in large part scared away both the retail and institutional investor.  Doesn’t this undermine one of the key fundamental assumptions of HFTs, namely availability of liquidity?  
• Isn’t there likely a high regulatory risk, given that so few investors, let alone politicians, understand HFT?
• What happens then the correlations cease to hold?  Or, when something exogenous (and unforeseen) becomes the primary variable driving things? 
• I assume (though I don’t know this) that HFT models are based more heavily on technical market analysis, than on fundamental analysis.  It seems to me that technical analysis offers the type of data that would fit perfectly into mathematical models.  Judgement, so common in fundamental analysis, can’t be programmed.

My Own Takeaway

When I think about Iati’s definition of HFT, it surely sounds accurate.  But the overall self-conception of the HFT industry seems to leave out a key consideration.  I look at this from a dynamic systems perspective.  HFT tools have been applied to an existing system, in order to “benefit from market liquidity imbalances or other short-term pricing inefficiencies... across [all tradable, my addition] asset classes...."  By all accounts, HFTs have been hugely successful at their task.

But, I believe we have entered a moment when the HFTs themselves are the exogenous factor that they didn’t build into their model.  They are such a large player in the markets that they are the ones driving the relationships between various securities.  

This seems unsustainable.  It also ties with the observations of many market watchers.  Most people seem to have a sense that this is unconnected to fundamentals and cannot go on forever.  HFT firms driving prices up higher because other HFT activities have caused prices to approach various technical thresholds, and so on.

Finally, I have to confess a personal bias against blackbox investing schemes.  I’ve tried here to be analytical rather than opinionated.

I hope this is a helpful read. 

Sources Used

Sorry I didn’t put these in Chicago Manual of Style bibliography format...

• Vimeo: “High Frequency Trading” (this is an excellent introductory video)
• Advanced Trading: “Breaking it Down; An Overview of HFT” (An excellent overview)
• Kate Welling’s interview with Themis (so in formative!  TD posted last night on ZH)
• Wikipedia: “High Frequency Trading” (there’s a ton on that site... )
• Automated Trader: “High Frequency Data Analysis (more techie focused, but helpful)
  

[Eric Cartman]

Nice. Thanks for posting. I certainly enjoyed it and with regard to blackbox investing schemes, I share your opinion.

[ZeroPower]

Regarding

Algo trading firms often use end of day data such as variable weighted average price (VWAP).

Vwap is value weighted average price. And its not an eod variable (though it can be). VWAP is applied to intraday trading by algos when doing stat arb. If you plot the intraday vwap on most stocks you'll see (along with the S/R levels) the values levitate around it on a day where there is no clear trend up or down (i.e. today). 

Number of competing firms doesn’t get so large that available margins disappear. 

Unfortunately/fortunately (whatever), the # of firms only in the business FOR hft or shops moving towards it, is increasing. Squeezes the margins again, but it doesn't matter as the margins rely on the rebates from the ECNs which don't change often (some changes beg of year but nothing influential). Do a search for 'quant trader' or 'quant developer' and see how many hits you get for jobs (hint: systematic is a synonym). 

It hasn't scared away institutions (but youre right about retail), and it definitely hasn't reduced liquidity. A retail joe6p with a $5k etrade account makes up literally zero liquidity of the market. Add a million joe6ps and youre still at a big zero that doesn't move anything.

An algo has nothing to do with a companys fundamentals. All technical, and dealing with divergence from indices, sub-indices, currencies etc.

As to whether or not its unsustainable. Time will tell. Time, and lots of $$$ in between.

[tradergod]

Disclaimer:  Trade idea influenced almost entirely by Goldman taking the buy side today.  I enjoy playing the high flyers on the way down as they tend to give huge returns in fewer than 3 months.

Goldman upgrades PCLN and up it goes 7%.   This after it's down 100 points the past 60 days, or about 30%.

PCLN short interest is 1/3rd what it was one year ago.  I don't believe Goldman Sachs cares about more than this current spike they created for short term gain.

Without the usual army of short covering this stock has enjoyed for 5 years, the joy ride seems over.

Fundamentally, besides being a crappy travel portal, the BP disaster should screw up the earnings forecasts for PCLN.  This will give the appearance of growth turning to negative, which we have all seen before with every internet stock ever IPO'd.

Good luck to you and me both... I always enjoy it when it lands squarely on my face.

[youngandhealthy]

I like your summary....more of this...HFT is evil

[firstdivision]

Short squeeze!!

[homersimpson]

More like taxpayer-funded market manipulation.

[firstdivision]

Bencopter3000 just landed.

So going off the chart TD posted along with the current EUR/JPN spot of 112.46(ish) then the ES should be sitting pretty much about 1105 range.

[tradergod]

As of 3:13 EST, they are still decoupled.  In fact, slightly more decoupled than earlier.

[tradergod]

It's like watching a magic show.  ESU ducked a bit, just waiting for EUR/JPY bounce to re-couple the lovers.

[tradergod]

Re-coupling occurred at 3:30 PM.

[BeerGoggles]

not quite, it was a few pips/pts off but has now spread apart again as the PPT pumped up the ES at EOD.

[BeerGoggles]

Well, that's another 30pips today.

This is the latest from the guy over at the "Essentials of Trading" site:

http://www.theessentialsoftrading.com/Blog/index.php/2010/06/08/reaction...