The alpha dens
|
The Adventures of Opco Boone, Legal Ace™
|
The west end was a no-go zone.
Pure, uncut alpha is a colourless, tasteless crystalline powder. It is harmless — beneficial, even: an enzyme the brain naturally secretes in times of excellence – it fuels the euphoria of victory.
Users — “alphaheads” — feel great power and liberation. They feel invincible. The narcotic quashes the appetite, the high lasts a week and, best of all: no hangover.
Pioneering econobiologist W. M. S. Sharpe gathered anecdotal data over the years and posited: even extended exposure yields an inarticulable wellness. It coats users with all kinds of physiological armour. They feel present, clear-sighted and certain. They feel in the moment. They make great decisions with utmost clarity. The greater the concentration — “ratio” — the better the decisions and, bingo: more generated alpha.
But problem: generating alpha in the lab was hard. Real hard. The more they tried, the harder it got. It was as if there was some kind of weird negative feedback loop in play: the more they tried, the less they produced. It was freakish: as though the alpha knew you were looking for it, and sublimed away, a Cheshire cat of psychological wellbeing.
It did not make sense. The models were awesome. The calcs were impeccable. The forecasts checked out against all historical data. The more data they fed into it, the better it checked out: the back-tests screamed shit-tons of alpha. The lab-boys clinked glasses and Googled condos in Verbier.
But when they ran the algos forward, on a live incoming, they got doughnut. Bupkiss.
They ran them in London: Not a sausage.
They ran them in Frankfurt: Keine würstchen.
They ran them in Paris. Pas de saucisse.
They ran them in Milan. Niente salsiccia.
It was like the goddamn stuff was self-aware: it seemed to know they were looking for it. It saw them coming and vamoosed. They taped the live-feed and fed it back into the model: bam, there it was, gallons of notional alpha — enough to flood the ballast tanks.
Sharpe barked at his lab assistants: “alpha’s no good on a goddamn spreadsheet, ladies and gentlemen: keep working.”
They worked like Trojans. The models predicted buckets. But out in the wild: żadnej kiełbasy, just like before.
It was like their very pursuit generated some exactly off-setting dark energy that phase-cancelled the effects.
“It’s the damnedest fucking thing,” said Sharpe. This caught the mood. It became his epitaph.
Sharpe hypothesised: alpha-hunting in a live-fire environment generated some kind of offsetting force that the models weren’t capturing. Sharpe called it “beta energy”. They couldn’t pick it up because it decayed as soon as it was recorded.
Beta energy forces an alpha system back to equilibrium. Live feeds are dynamic: the data can change. Alpha hunting interacts with data: changes them, in real-time, as they are generated. On exposure to light, alpha tends to zero.
This is why recorded data gave such strong alpha readings: they were static: it didn’t matter what models you ran over them, they stayed the same. Hence, juicy alpha readings.
The weird thing though: You could find one value of alpha in unadulterated recorded data – data that wasn’t alpha-hunted while it was live. Datasets that had been alpha-hunted had a different profile: none of the alpha from the unadulterated set was there, but there was just as much alpha in the set. It was just different.
Sharpe commissioned some experiments in second-order alpha-hunting: trailing an alpha-hunter, trying to glom this alternative stream of alpha. It worked, beautifully, on the lab-recorded data sets. But was useless in the wild.
The boffins at UC Ilam collected more data. They took it by the petaflop. They ran it through SETI@home. They strained it through neural nets. They crawled it with bots.
One more data point: market makers generated zero alpha. Passives generated zero alpha. The returns were as good as pure beta.
Almost.
Six, seven standard deviations out there was a tiny, subtle heat signature. So small you couldn’t see it with the naked eye. So small as to be a rounding error. It was below the tolerance threshold. It looked like noise.
Except it was persistent. The boffins tapped the dials.
The signal stayed.
They cleaned the Excel lenses.
The signal stayed.
They de-gaussed the visual basic engines and re-rendered the algos.
The signal stayed.
However many times they ran the numbers, there it was. This penumbral, ghostly alphic aura, just fuzzing up the clean lines at the deep space extremities of the curve, way out, a billion lightyears from Earth.
“There is alpha there,” said Sharpe. “It’s just so far down the tail our instruments can't detect it. All’s we need is some way of magnifying that signal, and I believe we’ll have something.”
The boffins fixated on that fuzzy little penumbra of positive alpha way up at the positive end of the curve.
They didn’t notice: the bottom tail of the curve was fuzzy, too.
The passives didn’t affect it. Couldn’t get close. Market makers didn’t affect it. The boffins concluded: there were too many alpha algos. They were jamming each other. The amount of alpha in the system was constant. It was a function of positional trading. It was relative to all the efforts to produce it. It was a function of money and assets and not trading. The more people traded, the less real alpha there was per trade.
And it was nebulous: the asset management labs that sprung up in Greenwich found no particular formula guaranteed its generation. It was as if analysis caused it to die. It came and went by chance. It was infuriating. It was there, it was broadly constant, but so elusive. The AM labs got creative. They hawked their historical production rates and wrapped their ads with fully cancelling disclaimers:
“Our last five years were gangbusters. So, you know —. Oh, and by the way: past performance does not indicate future results.”
Paradox, or irony?
Alpha: hard to predict in advance, a cinch to see in hindsight. Part of the patter: reconstructing the past to market the future. It was a skill. Some were naturals at it. They called them back-testers. They could name their price.
The great cynic called it: “We’re driving the train forward, the guy steering the wheel is looking out the back.” They cancelled him fast. A skew of salacious stories and they buried the lede.
As the craze hit peak rush, Quantum could generate a small amount of total demand. The partners restricted supply, insisted on strict quality control, and took their reward through fat margins. They grew rich.
The market fragmented. New products, di-alpha synthetics appeared. They had no history, so how could you sell them? You invented one, by reference to historical data. The results guaranteed by reference to what might have been. Hindsight is a wonderful thing. The back-testers did wonderful work. Our cynic again. Now we know where we were going, it’s easy to construct a vehicle that would have got us here in style. Not so easy then.
Always with the disclaimer: past performance does not indicate future results.
Quantum hoovered up grads. Credulous drones with straight As and a proven knack for trotting out what the syllabus demanded. The young hacks busted nuts hoping for partnership. They boosted CSCs. They hyper-generated pitch books and algos and expectations curves.
They dropped big bucks on the legal paperwork. It was biblical. It was impenetrable. It screamed: no need to read, we’ve done all the work.
They built books, slung red herrings and pushed it out the door.
They built warehouses but kept their stock to a minimum. This was just-in-time production. The stock popped on issue — peak expectations curve, redux — and they moved it out pronto.
Retail got the dead cat bump and the hope of redemption. Day-traders yawped.
They made up some memes. They tweeted out #WGMI and #HODL. The Redditors hoovered it up. Some are still holding now.
The disclaimers had them covered.
The young bucks with straight-As yearned for progression. The dangled carrots swung further away. The young bucks were hungry. They needed to be fed. They had the skills and knew the drill. The bold ones split and formed rival shops. To accommodate demand, they lowered their standards but kept up margins.
The pundits love it. This time was different.
The market for bullshit grew fierce.
The punters kept buying. Margins stayed fat.
It didn’t add up. The punters didn’t care.
The markets ran up. The market makers flipped fresh paper for folding green stuff. They made out all noble. The paper gapped up and the punters got paper rich. Bullshitters got cash richer.
The punters took out loans and doubled down. The bullshitters cranked the handle. Out came fresh paper. In came folding green stuff. The paper gapped up. The pundits called it the work of genius. The punters made their own memes. They scoffed at tradfi and pooh-poohed fiat.
This time was different.
The bears called out the madness. They pointed out the flaws. They wrote papers. They were subtle. They were nuanced. They were articulate, and the message was clear: the market is exuberant. This cannot last.
The message didn’t play. The market gapped up. The bulls said, “Shorters gonna short” and put the case the punters wanted to hear: the alpha train’s going up despite the shorts.
Credit stayed cheap. The punters doubled down.
The bears went out louder. They minced fewer words. They said, “This will end in tears.
The message didn’t play. The bulls called them out for talking their book.
The banks made markets and lent on margin. The CEO said, “When the disco lights on, a man gotta dance.”
The bears put it in twelve-point caps and coloured it red.
The punters called sour grapes and pelted them with cabbage.
The bulls said: this time’s for real. Were going to the moon.
The market ran down. The punters got hosed. The Bullshitters stayed rich. The pundits called it a correction: these things happen. This time was still different.
The problem: alpha was, by nature, rare and hard to produce. It manifested indirectly. You couldn’t always know it before you saw it. Sometimes you saw it, but you were wrong. There was no easy way to tell until it was way too late.
For the bullshitters: loads of scope to dissemble. Narratising was a cinch. The back-tested projections were awesome.
Our friend, the cynic: it’s a ruse. It’s a bubble. It defies logic and common sense. They called out and recancelled. He didn’t go away: he had F.U. money from shorting the last crash.
The market ran higher. The punters held on for dear life. The market’s resilience in the face of a prominent bear case fortified their confidence. It’s not a bubble – the market has priced in the shorts. This was a new paradigm. The financial fundamentals had changed forever. They had turned base metal into gold.
The Bear said, “I grant you, I didn’t expect that.”
The market ran higher. It floated all boats.
The Bears ran op-ed excoriations. They let themselves down with intemperate hyperbole: they said “alpha” didn’t exist. They called it dumb luck and psychosomatics. They cited the stats: in a field of a thousand traders, ten will beat the market five years straight trading at random.
The bulls called them deluded. They said to be greedy when the market is fearful. The punters got greedy.
The market ran higher. The traders beat their benchmarks. They busted their hurdle rates. They raked in 2 and 20. They boasted in their investor letters about their audited accounts – proof positive in black and white, they said, that they were alpha machines.
Investor letters ran a disclaimer in their footer: past performance does not indicate future results. They might have said this: objects in the rear-view mirror may be closer than they appear.
The Bear called it: Alpha was like Malaysian oil. There was ten times more on sale than the market could logically produce.
Someone was selling snake oil.
The market ran higher. Academics produced papers on the persistent madness of crowds. They explained the absence of shorts: the market can stay irrational longer than you can keep your AUM. Shorting an exuberant bubble was a fool’s game.
Bill Haydon of Hyde Park Capital Investors, celebrated scourge of the market, declared himself just such a fool. Haydon ran a big public short. He called the market out.
The market ran higher.
Haydon hung on. He had the stomach for a fight: it was he who broke the Lira in 83. He smoked out the carbon bulls in 95 and made a killing. He wanted another notch in his belt. He went for the alpha crowd. He talked a great game – that was his major strategy, matter of fact. He tweeted wildly. He held press conferences. He published articles. He was on a short loop, talking the market down.
The market ran higher.
Haydon hung on. He was bigger than this.
SqL played him like a fiddle. Waited till his funding supply lines were stretched, till Haydon was full extension at the outer limits of his reach – then he waded in with a large fully funded bull play. He went public – all-out full-channel frontal assault – it was not his usual style but it was what the situation demanded. It was the precise tool for maximum impact on this play.
The marked gapped higher.
Haydon’s PBs pulled their lines.
They jacked up margins and hit him with calls
Haydon was jammed: he was running zero excess cash on hand and got hosed.
The punters blew them off. They put it down to professional jealousy. There was alpha, there must be, it was a mathematical derivation. Saying there was no alpha was like saying there was no common denominator.
The cynic said, “A common denominator depends on the common numbers. It doesn’t exist independently. You can’t have a common denominator until you know the numbers”.
Theorists held their line. They worked out a theory. They built a paradigm. The market generally bought it. Alpha was there, the alpha effect was real, but it had bad by-products. For every gram produced, there was an exactly equal offsetting amount of negative alpha – together, they cancelled each other out. It was a zero-sum game. The trick was to strip the negative alpha away and dispose of it. That’s what you paid the big bucks for.
Cynics: you can’t know what is alpha and what is negative alpha until it’s too late. By the time you know it is in your portfolio.
Catch: No one wanted the negative alpha. Who wanted a buzz that made you paranoid, insecure, weak and half-blind? You could maybe find some mutual funds to stuff it in their portfolios, cut it amongst the huge quantities of other stuff – dilute it, water it down, but these guys were smart and they wanted to be paid.
Double-catch: it was impossible to identify alpha at all, let alone separate negative alpha before you took it. They ran tons of tests. You could only see it in the lab after the chemical reaction that generated it.
Triple-catch: it was easy to generate something that looked a lot like alpha, by mixing beta – harmless stuff, you found it in tap water – with amphetamines, stimulants, caffeine – a class of accelerants called vega – and much of the time this had a similar effect as alpha. If the market rose, synthetic alpha rose more. If the market fell, you ditched the vega. Snappy name: enhanced beta.
So easy to say in theory. But ditching the vega meant finding someone to offtake it. Keine problem in a rising market. Who didn’t want a lever when the balloon was going up? You could sell all the vega you wanted: getting hold of the stuff was the problem. But in a dip, the lever was a killer. No one was buying in a falling market.
Here’s where real alpha – that elusive capricious unicorn – and vega cut beta parted ways. In a falling market, real alpha might take a loss, but it still beat the market. Relative to the average player, you were still better off. Vega cut beta got hammered. The lever worked in reverse.
Snake oil peddlers started boosting chat about black swans and five sigma events. “Wow man, the probability of this happening was like not in the life of the fricken cosmos man.”