The returns that couldn't exist.
In 1999, Northfield's returns-based analysis put the odds that Bernie Madoff's reported numbers were real at essentially zero. He was arrested nine years later.
A return stream that rose at 45 degrees and ignored the market.
In 1999, a colleague brought Northfield a track record that looked too good to be true: a smooth, upward line of monthly returns that barely wavered whether markets rose or fell. The manager was Bernie Madoff, and the numbers came straight from the materials his firm gave prospective investors.
Using the same returns-based and factor methods Northfield had published in the Financial Analysts Journal, Dan DiBartolomeo decomposed the return stream against the strategy Madoff claimed to run — a split-strike options approach that should have moved, a little better or a little worse, with the market. It didn't. It was almost indifferent to whether the market went up or down.
Within hours, the conclusion was unavoidable: no strategy Madoff described could produce those returns. The probability that they were genuine was, for practical purposes, zero.
The finding reinforced the case whistleblower Harry Markopolos was already pressing with regulators. Madoff was not arrested until December 2008 — nearly a decade later. The episode is recounted in the Wall Street Journal and in Markopolos's book No One Would Listen.
What the numbers showed
- A near-straight 45-degree ascent — returns almost without variance.
- Little to no relationship between the returns and the market they ran in.
- Performance indifferent to whether equities rose or fell.
- Volatility far below anything the stated strategy could deliver.
- No factor exposure that could reconcile the claim with the record.
The method behind it
- Returns-based and factor analysis — the analytical core Northfield commercialized.
- The manager's claimed strategy tested against what the returns actually implied.
- Grounded in methods Northfield had already published in the academic literature.
- Run in hours, from nothing more than the marketing data investors received.
- The same discipline Northfield puts into production risk work every day.
Nine years early on the largest fraud in history.
Northfield's analysis did not, on its own, end the fraud — no one acted on the warnings for years. What it demonstrates is the power of the methodology: from published data alone, the same factor and returns-based tools ALPHA runs today separated the impossible from the plausible, a decade before the market caught up. The account is part of the public record — the Wall Street Journal (December 2008) and Harry Markopolos's No One Would Listen (2010).
Exposures the record couldn't support
Decomposing the return stream into economic drivers exposed a profile no real strategy could produce.
Strategy vs. reality
The claimed split-strike approach tested against what the returns actually implied — and failed.
Not an opinion
Grounded in methods Northfield had set out in the Financial Analysts Journal, open to challenge and replication.
Hours, from public data
The whole read took hours, using only the marketing materials any prospective investor received.
Get in touch
Rigor that holds up when it counts.
The analysis was 1999. The methodology runs in production today — the same factor and risk tools, inside the institutions that can't afford to be wrong. Put it to work on your book.
Talk to us →Analysis first performed in 1999 from Bernie Madoff's own marketing materials, using returns-based methods Northfield had published in the Financial Analysts Journal. The episode is recounted in the Wall Street Journal (December 2008) and in Harry Markopolos's No One Would Listen: A True Financial Thriller (2010). Madoff was arrested in December 2008 and later pleaded guilty; figures reflect amounts reported publicly at the time.