Tagged: Algorithms


“It’s like the market has become some uncontrollable machine lurching around,” said friend Pat at the NIRI Rocky Mountain chapter meeting last week.

I spoke there about Exchange Traded Funds and how they contribute to market instability through the slow adjustment of collateral for creating ETF shares and the very rapid setting of prices for stocks and ETFs. If you want the presentation, send me a note and I’ll share it in pdf.

But first, breaking news from Steamboat Springs: Ski conditions are superb and you can snowshoe in deep champagne powder atop Rabbit Ears between the Yampa Valley and Grand County. We took in WinterWonderGrass too, mass musical ode to banjo, mandolin and upright bass. Blurring genres and maybe the event’s best was Fruition. They might be giants.

Back to Pat’s point that the market seems like a giant robot with the humans inside desperately fiddling with buttons and levers as it swings around destroying furniture, wouldn’t that suggest something awry in the mechanics of the market rather than human irrationality?

While you ponder, the blame may be going unfairly to new SEC chair Jay Powell. Whatever he said to Congress yesterday, the markets had from 8:30a ET to read the text.

Maybe he blew the Q&A. He used the word “strong.” Durable goods were terrible and he thinks the economy can absorb more rate-hikes. GDP may come in lighter still for Q4 in today’s revision.

We’re casting about for rational explanations where there may be none. What you never seem to hear is an investor saying “the reason I sold Powell’s testimony is because of” blah blah blah. What you hear are investors trying to explain why other investors are selling.

What if it’s the machines?

In a terrific interview with the New Yorker last December, Jim Simons said, “I looked at the price charts and analyzed them, and they didn’t look random to me.” Simons, a mathematician, founded hedge fund Renaissance Technologies and is today worth $20 billion.

The firm’s flagship Medallion Fund has generated 80% annually before fees for almost thirty years. It’s size-constrained because big money can’t be moved quickly, and longer-term trading makes algorithms less useful. “It’s like the weather,” says Simons. The nearer in, the higher the certainty.

So think about that. Humans and their fundamental models are trying to produce returns years in the making – which is highly uncertain. Machines focus on the moment. A high-speed trader told me they think of stocks in real time as moving from 5s to 6s, or 6s to 5s. “Buy the 5s that are becoming 6s, and short the 6s that are becoming 5s, and never take your eye off the book – the money at risk.”

I can tell you we rarely see arbitrary scatterplots in the behavioral trends we track. There’s a mathematical symmetry everywhere that defies the idea of a random walk down Wall Street.

It would make sense if the rules required machines to set—wait a minute.  The rules DO require machines to set prices.

Every “marketable” order that wants to be the best bid to buy or offer to sell must by law be on an algorithm, a machine, so it can move fluidly around the market, much faster than humans can react.

The spreads are regulated, to the penny. The bid to buy can’t be higher than the offer to sell, nor can it be the same as the offer to sell.

That’s math. Machines can be programmed to adapt to rapidly changing prices. Simons told The New Yorker he “never overrode the model.”

And maybe that’s your answer. Nobody overrode the model.

So, then? First, investors and public companies have to stop thinking rational thought is responsible for all prices, and boards and management teams better know. Both investors and companies should become better at data analysis if they’re to compete with machines.

Second, there is irreconcilable conflict between humans focused on fundamentals and machines fixated on data patterns – and the machines have the advantage because the market has been handed to them by rules.

Third, should there be different markets for the two? We’ll have opportunity to grapple with that idea when next machines drive the market somewhere it wasn’t prepared to go.

Climbing Mountains

You’re welcome.

Had Karen and I not departed Sep 20 for Bavaria to ride bikes along the Alps, who knows what the market might have done?  There’s high statistical correlation between our debouchment abroad and a further surge for US stocks.

Stocks spent all of September above 5.5 on the 10-point ModernIR Sentiment Index. Money never paused, blowing through September expirations and defying statistics saying 80% of the time stocks decline when Sentiment peaks as derivatives lapse.

Were we committed to the interests of stock investors we’d pack our bags with laundered undergarments and return to Germany before the market stalls.

But is the market rational?

Univ. of Chicago professor Richard Thaler, who won the Nobel Prize this week for his work on behavioral economics, is as flummoxed as the rest by its disregard for risk. While Professor Thaler might skewer my certitude to knowledge quotient (you’ll have to read more about him to understand that one), I think I know why.

Machines act like people.  My Google Pixel phone constructed a very human montage of our visit to Rothenberg, a Franconian walled medieval city in the woods east of Mannheim.  I didn’t pick the photos or music. I turned on my phone the next day and it said here’s your movie.  (For awesome views of our trip click here, here, here and here.)

Google also classifies my photos by type – mountains, lakes, waterfalls, boats, cars, churches, flowers, farms, beer.

Don’t you suppose algorithms can do the same with stocks? We have long written about the capacity machines possess to make trading decisions, functionally no different than my Pixel’s facility with photographs.

For companies and investors watching headlines, it appears humans are responding.  If airline stocks are up because of good guidance from United Airlines and American, we suppose humans are doing it. But machines can use data to assemble a stock collage.

The way to sort humans from robots is by behavior. It’s subtle. If I sent around my phone’s Rothenberg Polka, where the only part I played was naming it, recipients would assume I chose photos and set them to music. Karen would look at it and say, “Get rid of that photo. I don’t like it.”

Subtleties are human. Central tendencies like flowers and waterfalls are well within machine purview. Machines don’t like or dislike things. They just mix and match.

Apply to stocks. It explains why the market is impervious to shootings, temblors, volcanic eruptions, hurricanes, geopolitical tension. Those aren’t in the algorithm.

Humans thus far uniquely grapple with fear and greed. A market that is neither greedy nor fearful is not rational. But it can climb mountains of doubt and confound game theorists. What we don’t know is how machines will treat mismatched data. We haven’t had much of it in over nine years.

The Obvious

Algorithmic trading is Wall Street’s last best hope.

So said the lead sentence in a story called Algo Wars in the May 30, 2005 edition of Investment Dealer’s Digest. That publication is gone and so is Lehman Brothers, co-leader of program-trading at the NYSE in May 2005, and computers were then rapidly displacing humans in driving it.

Algorithms, computerized mathematical models for trading, are ubiquitous now not just in equities but across a spectrum of electronically traded securities ranging from currencies and options to futures and US Treasuries.

History illuminates origins. It’s the reason to be a student of it, paraphrasing the Spanish philosopher called George Santayana (his actual name is a lot longer), who made the cover of Time Magazine in 1936 and observed that those who cannot remember the past are condemned to repeat it.

Algorithms, the article says, were birthed by “market developments and regulations that made trading equities more complicated and less profitable.” It quotes Sang Lee, founder of then brand-new market consultancy Aite Group, now a thought leader on market structure, saying algorithms “emerged from this hostile institutional trading environment where it’s getting increasingly difficult to move large blocks of orders.”

That was ten years ago. I had started ModernIR a few months earlier. Josh Friedlander, author of Algo Wars, wrote near the beginning that “because the democratization of algorithmic trading has just begun, its impact on the corporate world is still uncertain,” referring to ambiguity about how algorithms would affect stocks of companies.

Friedlander also wondered if small-caps, victimized then by decimalization and a regulatory separation of research and trading, would suffer further. The JOBS Act, made law in 2012, made it easier for small firms to go public but didn’t address structural woes for small stocks.  Today analyst coverage is a Rorschach blot on the biggest 750 firms, leaving 3,000 largely in uncovered white space. And the buyside and sellside have spent billions on technologies for hiding trades in a complex market.

Exchange Traded Funds (ETFs) grew out of this milieu. Moving big orders was a problem a decade ago.  Now look at it. We have Blackrock and Vanguard with $8 trillion of assets and a stock market with $24 trillion of capitalization. ETFs are the next evolution for a market built on rules meant to fuel movement but which paradoxically paralyze it.

I looked up one of our small-cap clients with about $1 billion of market cap and compared it to one with $10 billion. The small-cap was in 58 ETFs, 15% more than the $10 billion stock, and short volumes for it are in the highest 20% marketwide. It’s not that ETFs are focused on small-caps. Our typical large-cap client with $25 billion or more of market-cap is in about 100 ETFs.  Borrowing and derivatives predominate.

What should be obvious from the IR chair upon retrospection is how little faith one can have in what’s observable on the surface of price and volume.  ETFs move positions relentlessly and without respect to news save for reactions to prices and direction where applicable. Algorithms proliferating for a decade are designed to hide intention.

If as an institutional seller you wanted to obscure your disbursements, would you employ algorithms that pressured prices?  Selling would be patently obvious and the billions spent on sleights of hand wasted.  Clients, you know we routinely observe contrarian patterns in the data – Positive sentiment signaling impending pressure, Negative sentiment a bottom and probable buying.

Let me summarize. The obvious lesson of history here is that a decade of profound stock-market transformation coupled with leviathan investment from its core participants in purpose-obfuscating trading technologies will not produce a market where you look at your price and volume and say, “I think we have a big seller.”

Every now and then that might be right. But 90% of the time what seems to be apparent is probably not what’s occurring.  Thus ModernIR thrives today and we can help anybody regardless of size or trading volume observe reality under the market’s skin.

Chasing Spoofers

Apparently the market is very unstable.

This is the message regulators are unwittingly sending with news yesterday that UK futures trader Navinder Singh Sarao working from home in West London has been arrested for precipitating an epochal US stock-market crash.

On May 6, 2010, the global economy wore a lugubrious face. The Greeks had just turned their pockets out and said, “We’re bollocks, mate.”  (Thankfully, that problem has gone away.  Oh. Wait.) The Euro was on a steep approach with the earth. Securities markets were like a kindergarten class after two hours without some electronic amusement device.

By afternoon that day, major measures were off 2% and traders were in a growing state of unease. The Wall Street Journal’s Scott Patterson writing reflectively in June 2012, interviewed Dave Cummings, founder of seminal high-frequency firm TradeBot. Heavy volume was scrambling trading systems, Patterson wrote, leading to disparities in prices quoted on various exchanges. The decline became so sharp, Cummings told Patterson, that he worried it wasn’t going to right itself. If the data was bad, TradeBot would be spreading contagion like a virus.

Ah, but wait. Regulators now say mass global algorithmic pandemonium May 6, 2010 was just reaction to layered stock-futures spoofing out of Hounslow, a London borough featuring Osterly Park, Kew Bridge and a big Sikh community. If you think the Commodity Futures Trading Commission’s revelry over finding the cause of the Flash Crash just north of the Thames and west of Wimbledon stretches the bounds of credulity, you should.

Mr. Sarao is accused of plying “dynamic layering” in e-mini S&P 500 futures, a derivatives contract traded electronically representing a percentage of a standard futures contract. It’s called an ideal beginner’s derivative because it’s highly liquid, trades around the clock at the Chicago Mercantile Exchange, and offers attractive economics. (more…)

Losing Purchase

Say you were house-hunting in a hot real estate market. Would you Tweet: “I’m going to buy the house at 1342 57th Street, and I’m headed there right now.”

If you want a good deal, do you wave your hands and try to summon others to compete with you?

So why do companies announce stock repurchases?

Forget disclosure for the moment, or lowering shares outstanding to offset incentives. Sure, there’s a bit of the blue-light special. A repurchase wants attention in the sense that it signals “we think this is a great use of cash.” It blunts bad news: “Our margins were down this quarter – but guess what? The board said to buy $500 million of stock.” That’s a positive value action to forestall a negative value message.

In both cases, it’s also a marketwide Facebook post proclaiming that “we are about to spend money here.”

Let’s apply common sense now. Isn’t that antithetical to the wise deployment of shareholder resources? If you want a good deal, cut out the middlemen – don’t hail them from all over the globe to get between you and the thing you want to buy. (more…)

Reg Nemesis


Rather than Nebraska, we’re in Steamboat Springs enjoying spectacular slopes before this weekend’s big game between our Broncos and the Seahawks.

Denver quarterback Peyton Manning uses the word “Omaha” to change plays or alert his team to shifting defensive coverage. In football, both defenses and offenses try to confuse the opposition with different formations. It’s deception.

In the stock market, the principal purpose of algorithms is to deceive and algorithms execute 90% of stock orders now. The stock market isn’t supposed to contain “the opposition” but buyers and sellers who want to find each other. Why are they hiding instead?

Part of the problem is Regulation National Market System. If you didn’t see Wall Street Journal writer Jacob Bunge’s piece (if you can’t open the link, email me and I’ll send the story) on what’s popularly called “Reg NMS” yesterday, it’s required reading for every IRO, CEO and CFO at public companies.

Two vignettes from Mr. Bunge’s story: (more…)

Great Expectations

Happy New Year! Hope you spent the two-week break from these pages joyfully.

We’ve descended this week from the high Denver backbone of the continent to visit west in Santa Monica and sponsor NIRI’s Fundamentals of IR program. Following our New York trip before Christmas, we’ve marked the turn of the calendar by touching both coasts.

We’ll kick off the year with a story. I’ve just finished Charles Dickens’s Great Expectations on my Kindle. Yes, I realize it was first published in serial form in 1860 (the year the cattle ranch on which I grew up was homesteaded). I have a long reading list. It took me awhile to get around to it.

Lest I spoil excitement for the other three or four of you planning on it still, I’ll say simply that it’s a masterful narrative assemblage of plot points, the connections between which one would never fathom at the outset. Great storytelling never gets old.

The market is like that too. As you begin 2014 in the IR chair, remember that in a market dominated by algorithms – the principal purpose of which is to deceive – things are rarely as they seem.

Take trading from Dec 9-31, 2013. The US equity world it seemed was gathered in knots and pockets like people in an old west town where the gunslinger was expected anytime to ride through. Tones were hushed, gestures animated. A pregnant air of expectation hung like a storm.

Would the Fed finally taper? And if it did, what then?  (more…)

Retail Reality

Do retail investors matter?

Depends what you mean. They’re important and valuable as investors. I once headed investor relations for a company with thousands of retail holders. I was president of an IR services firm that focused on retail-targeting strategies.

But when people ask me if retail investors have a chance in the market, my answer is monosyllabic: No. Scottrade, the online brokerage, runs ads featuring clients claiming that “I don’t trade like everyone else. I trade like me.” That seems to suggest retail traders can craft unique schemes that stand apart. While it’s theoretically conceivable for a retail order to price the market under rules requiring “market” orders to meet at the best national offer to sell, in practice it’s remote. Retail orders are passengers on market trains.

I’ll explain. Scottrade says in its order-routing filings that 100% of its trades are non-directed – they don’t specify execution venues. The average Scottrade customer is thus statistically most likely to trade not like you or me or him or her, but like Knight Securities, since that’s where a quarter of Scottrade orders go.

Another chunk from Scottrade lands at Citadel, the high-speed hedge-fund owned platform. Citigroup gets the largest portion but it’s divided between limit orders at Lavaflow, Citi’s fast-trading facility, and market orders at Citigroup’s agency desk (suggesting Citi powers arbitrage).

About 15% of orders shoot through Direct Edge, the high-frequency-trading exchange that’s merging with BATS, another sizzling execution venue. And for options and futures and NYSE stocks, about 7% of Scottrade’s orders route to G1 Securities that until recently was owned by rival E*Trade, which has sold it Susquehanna, a quantitative trader. Yup, folks who hoped to “trade like me” were actually trading like E*Trade. (more…)

The Great Debate

The Great Debate is upon us.

No, not the presidential one tonight. The other one, about equity markets. The SEC’s technology summit yesterday aimed at finding ideas for preventing another Aug 1 Knight Capital debacle from ever happening again included mostly the folks who huddled after the 2010 Flash Crash to prevent glitches from…ever happening again.

Since the Knight glitch came after efforts to prevent glitches from, yes, ever happening again, and since glitches and one-off flash crashes are routine now, reflected in continual halts and erroneous trades (including two yesterday early, even as the SEC summit was commencing), understandably hopes for change are dim.

We’ve gotten many questions in recent days. How do we control technology? Is technology the problem? Are markets too complex?

And the simplest one: Why can’t we shut the hummer down when it goes haywire? Right? Common sense tells us it can’t be too hard if it involves electricity. Pull the plug. Yank the doohickey or whatever out of the machine.

We’ll get a “kill switch,” sure. But HFT won’t end soon because structure depends on it. The major exchanges are averaging about 4.6 billion shares of trading volume, down from over 7 billion daily in 2009, when incidentally Dodd-Frank was crafted. Add 30% more in dark pools and volume is about 6 billion shares each day now.

If we divide that into what’s Navigational – moving stuff around – and what’s Fundamental (real buyers and sellers meeting), it’s about an 85%/15% split. We’re left with 900 million shares of “real” volume, with the rest from HFT, ETF arbitrage, automated market-making and so on. This is the glitch-infested stuff.

What happens if all 85% of it disappears? Nothing, if you’re Berkshire Hathaway’s Class A shares trading 400 shares daily with no navigational volume. For exchanges selling data and services to drive profits, it’s doomsday. The largest broker-dealers would leave equity markets. So would the 50-odd large high-frequency firms. (more…)

You Never Know

An ode to erudition in professional sports, these pearls of wisdom overheard on sidelines come thanks to ESPN’s halftime report during the unfortunate demise of our Denver Broncos in Monday Night Football:

“If you hadn’ta been where you was, and did what you did, we wouldn’ta got what we got.”

“You can sum it up in one word: You never know.”

“You never know” is a good way to describe markets. And reason why market-structure analytics are essential to IR. Paul Rowady at TABB Group, the top market-structure authority today, wrote extraordinary commentary at TABB Forum yesterday saying monetary intervention by central banks poisons market data.

What’s the real price of your stock? As you ponder, Rowady says, “At any moment in time, one could argue that there simply cannot be true price discovery in any market where intervention occurs – which is most of them.”

Why? Because central banks, unlike the rest of us participants, can use unlimited money and unrestrained access to information – the Federal Reserve is not bound by “insider trading” constraints like you – to affect prices of every asset, every commodity, every currency. (more…)