Schroeder: Putting theory to the test

Shamil Chandaria puts his money where his mouth is. The former global head of structured finance at Deutsche Bank has been telling audiences across Europe to adjust their portfolios for reduced volatility over the coming two years, and by signing on as chairman of London-managed (and Cayman-based)

The fund is managed by former McKinsey & Company strategy consultant Karsten Schroeder and employs a team of six researchers, mainly recruited from Germany's quantitative world. The strategy isn't designed to capture large moves, but rather to skim off the overage after a market overshoots and then reverts.

"My reasoning is that if you only have fundamental players who understand the situation in the market, and some news comes out that should move the price from, say, 100 to 110, then that is what will happen," he says. "But in reality, technical traders will observe the beginning of that move and jump on the trend bringing the price to 120 or 130, at which point the fundamental traders will say the market is overpriced. So they start bringing it back into line, but the trend guys see we are in a downward trend and they bring the price below 110."

Although that strategy is behind several off-the-shelf indicators, Schroeder says Amplitude's approach is 100% proprietary. "We developed all of our indicators in-house and you will find none of them in a text book," he says. "We are in 16 different instruments, basically one commodity, crude oil, and 15 financials. We apply our strategies differently for each market so the same inventory is not applied across the board."

He says the heavy volume of trading minimizes the danger of overfitting, which is what happens when a statistical model has too many parameters. "We hold trades for an average of 1.5 days but slightly longer for winners and shorter for losers," he explains. "That's a lot of trading. If you run on a medium' to long-term basis and only generate a few trades per year, then it's dangerous to optimize for each market because the risk of overfitting is high."

After five years of preparation, the fund started trading one year ago with $4 million they put up themselves, then opened to customer money in June. They say they can handle $750 million and already have nearly $50 million under management, which is symptomatic of the new breed of fund run by men with the kinds of pedigrees trusted by institutions.

But Schroeder's pedigree goes beyond the McKinsey aspect. He's traded futures for his own account since the late 1990s, and unlike the stereotypical consultant has always been leery of untested theory. "One thing I became convinced of, even in school, is that people often try to explain in a scientific way things that don't connect with reality," he says. "As a trader, you have to be pragmatic about theory and observe what works. Then you apply the working part and disregard the non-working part."

Like most traders, his early strategy was haphazard and discretionary. He moved through a series of subjective technical indicators such as candlesticks and Fibonacci retracements and finally settled on the combination of trend momentum and mean reversion that is the basis of Amplitude's approach today.

He's also fiddled with the various indicators used to evaluate traders and finds fault with using the Sharpe Ratio to measure a trader's risk level. The widely-used tool assumes that big wins can only come at the cost of great risk, essentially equating volatility, up or down, with risk, and then offers a ratio of that "risk" to performance.

"Whatever you invest in, you don't care about upside volatility," he says. "You should only look at downside volatility." So he advocates the Sortino Ratio, which only measures downside volatility against returns. "Even that's not perfect," he says. "It still assumes independent and identically distributed returns," another of the standard measures employed by probability theorists that Schroeder finds fault with.

For Schroeder's money, the simple measure of biggest and average drawdowns are the best measures of risk and in Amplitude's case, the biggest drawdown in five years of backtested data, allowing for slippage on $750 million under management is less than 5% stretched out throughout six months, while volatility is constant at 6.4%. Since going live in November 2003, the fund is up roughly 8% with 0.98% in September and 2.36% in October.

That's just more than half their targeted annual average, so true believers can hope to cash in on a rebound next year, although sticklers like Schroeder might call such a projection less than scientific.