What Long Term Capital Management Can Teach us About Basis - Part 1 of 2

Michael Ippolito | July 30, 2018

Stablecoins are one of the most interesting areas of development in the blockchain ecosystem right now. Lately there's been renewed focus on price volatility due to institutional interest in the crypto, and stablecoins represent a promising solution.

Basis (formerly Basecoin) is one of the projects leading the charge. It's a pretty damn interesting project in a lot of respects. Rather than taking what might be considered the easy route and collateralizing their coin with an underlying asset, they've opted for a system based on an algorithmic 'smart bank' that controls supply and demand to maintain a peg to the US dollar.

The project is also a pioneer of the emerging multi-token economic system. The Basis system is home to three different types of tokens, which allows for greater incentive flexibility and a robust revenue model for early investors.

In addition to all of this, they've successfully raised over $133 million in funding from investors like Bain Capital Ventures, Andreessen Horowitz, Stanley Druckenmiller, Lightspeed and a slew of other big names.

Seems like everything is working for these guys right?

Not exactly. There are two MASSIVE issues with the system that I just can't wrap my head around.  And I'm not even the first one to make these observations.  In fact, several other people have already pointed them out, which kind of makes me think that this is an 'Emperor isn't Wearing any Clothes' sort of situation.

Anyway, here are the problems:

1. The 'Smart Bank' algorithm won't have a long enough history of data inputs to make good decisions

2. Absolutist economic policies like bond floors and expiration dates are poor counter-measures against a death spiral

This article will address the first of these issues concerning Basis' 'smart bank' algorithm.  A following article will be released on insignificant counter-measures to a death spiral.


Long-Term Capital Management

Some of you may remember Long-Term Capital Management (LTCM). The firm was one of the first quantitative hedge funds, which at its zenith in 1998 managed over $134 billion in assets.

The fund's trading strategy revolved around a revolutionary formula for pricing options, known as the Black-Scholes model. Myron Scholes, a professor of Economics at Stanford, together with Goldman Sachs' Fischer Black and Robert Merton posited that the price of a financial derivative depends on five variables:

1. Current market price of the stock (S)

2. The agreed future price at which the option could be exercised (X)

3. The expiration date of the option (T)

4. The risk-free rate of the economy as a whole (R)

5. The expected annual volatility of the stock (σ)

Here's the formula they came up with:

Black Scholes Formula


Confused? So was everyone else. But that was just fine as far as Scholes, Black and Merton were concerned.  In fact, the more confusing to the layman the better.

LTCM dealt primarily in fixed income arbitrage, leveraging the Black-Scholes model to exploit deviations from fair value in the relationships between liquid securities across nations and asset classes. Merton and Scholes recruited John Meriwether, a Wall Street legend who made his name running the fixed income desk at Salomon Brothers, to helm the fund and boom - they were off to the races.

The fund performed phenomenally well right out of the gate. In its first two years, LTCM posted returns of 41% and 43%, respectively. If you had invested $10 million (their minimum investment requirement) you would have wound up with $40 million just four years later.

To many, it seemed like the ushering in of a new era. Gone were the days of 'gut feeling' investments. LTCM had found a new way, a better way, of investing that relied on mathematics as oppose to irrational human decision-making.

And if you doubted the LTCM strategy? Just check the scoreboard. The proof was in the pudding, and the numbers didn't lie.


The Numbers Don't Lie?

So what happened? As it turns out, numbers don't lie, but if you don't have enough of them, sometimes they will omit certain truths.

LTCM's downfall was the result of two highly improbable financial crises: the 'Asian Contaigion' currency devaluations of 1997 and the Russian Default of 1998. To be more specific, LTCM's downfall was the result of these financial crises in conjunction with the high degree of leverage their fund was operating at.

Here's the highly abridged version of what happened: despite LTCM's extensive diversification across markets and asset classes, they were significantly exposed to one variable: volatility. In fact, at the fund's peak they had $40 million riding on each percentage point change in US equity volatility.

The reason this was allowed is because, according to the quants, the probability of an event occurring that would lead to LTCM losing more than $45 million in a day was almost inconceivable. Therefore, what might seem like a risky position was deemed by the quant's 'Value at Risk' (VaR) models to be perfectly reasonable.

What they weren't counting on was the Russian default. It was a black swan event that caused a 'flight to quality' across multiple asset classes. LTCM's strategy of exploiting deviations in fair value meant that they held almost exclusively 'non-quality' assets that, within the span of about four months, lost significant value and liquidity.

The result was devastating. On Friday, August 21, 1998, LTCM lost over $550 million, an outcome that was virtually impossible according to their Long-Term Risk Models. The fund was bailed out in an arrangement brokered by the Fed in October of 1998, wiping the multi-billion dollar gains of the founders out in less than one quarter.


Short-Term Capital Management

LTCM is a cautionary tale that, despite the amount of attention its collapse received, many in the world of finance (and technology) seem all to eager to forget.

There were a number of factors that led to the downfall of the once invincible quant-fund, but one stood out to historians as particularly significant.

LTCM's VaR models were only drawing on five years of historical data in their calculations. An analysis of the available five years of financial data predicted that the loss LTCM sustained on that Friday in August was so unlikely it ought not to have happened in the history of the universe.

If their data had extended back to the stock market crash of 1987, or better yet to the first Russian default of 1917, their models might have calculated a different risk exposure. John Meriwether himself admitted that, 'If I had lived through the Depression, I would have been put in a better position to understand events.'

LTCM's fatal flaw also may have been the implicit assumption that human beings function as rational actors in financial markets. John Maynard Keynes has famously said that, "Markets can remain irrational longer than you can remain solvent."

LTCM's algorithms were predicated on the rational notion of a return to fair value. Ultimately, they were correct, most assets that they held would eventually converge with fair market value. Unfortunately, the volatility that occurred in the meantime depleted their stores of capital before that could happen.

As Niall Ferguson observed in his seminal novel "The Ascent of Money," LTCM's reliance on a short history of data and their mischaracterization of financial markets implied a different name for their firm: Short-Term Capital Management.


Back to Basis: What Does LTCM Have to do with Stablecoins?

Basis, like LTCM's fund, relies on a formula. Albeit, Basis has produced an algorithm far more complex than even the Black-Scholes model of pricing options, but the fatal flaw might be the same.

The hamartia lurking at the heart of Basis may be its reliance on an algorithm for which there is insufficient historical data to accurately predict future behavior.

The 'smart bank' algorithm is, by Basis' own description, an algorithmic central bank. This 'smart bank' has one mandate - maintain the peg to the USD. It accomplishes this by controlling supply and demand (i.e issuing new coins to increase supply, and issuing bonds to contract supply).

The 'smart bank's mandate is similar to that of the US Federal Reserve, albeit with a more narrow scope. The difference, though, is that the Federal Reserve has over a hundred years of data to draw upon. Data from 'black swan' events such as the Great Depression of the 2008 Recession are utilized to develop financial policy.

Basis, on the other hand, doesn't have that advantage. They are building an entirely new ecosystem, in which participants will conceivably exhibit entirely different behavior patterns.

Can Basis' 'smart bank' analyze real world market and economic data? Sure. Does that mean it'll be accurate? Absolutely not.

And yes, I understand that Basis is a trailblazer. The type of data they would theoretically need to power their financial ecosystem doesn't exist, so you might argue that it's not fair of me to criticize them for not having it.

But this isn't really about being fair. It's about determining whether or not Basis is a good long term investment. Long-Term Capital Management is just one example of an organization built around an ostensibly 'indestructible' algorithm or idea that turned out to be facile.

As Cathy O'Neil, author of Weapons of Math Destruction, presciently states: "Algorithms replace human processes, but they aren't held to the same standards. We trust them too much."

I've already said this but Basis might be doomed to fail because the task that they are doing might necessitate that the first movers fail so that the second wave can get it right.

Something to consider if you're thinking about investing.


*Note: Niall Ferguson's "Ascent of Money" was used as source material for this article

View All News Posts

Michael Ippolito

Co-Founder of BlockWorks Group