U.S. District Judge Gregory H. Woods, sitting in the Southern District of New York, denied cross-motions to exclude expert testimony in the securities and commodities fraud prosecution of Leech, finding that the proposed testimony from two government experts, two defense experts, and two rebuttal witnesses meets admissibility standards in a case built almost entirely on competing quantitative analyses of more than 33,000 trades spanning a 33-month period.
The government charges that Leech ran a cherry-picking scheme from January 2021 through October 2023, waiting to see how Treasury futures and options trades performed before routing winners to his favored Macro Opportunities strategy and losers to the Core Strategies he managed for other clients, all without disclosing the practice. A November 2024 indictment charged him with investment adviser fraud, securities fraud, commodity trading advisor fraud, commodities fraud, and making false statements to the SEC.
Harvard Business School professor Lauren Cohen and NYU Courant Institute professor Petter Kolm, the government's experts, proposed to testify that trades Leech allocated to Macro Opps had an average first-day return of $243,000, while trades he sent to the Core Strategies averaged negative $309,000, a gap that their regression analyses attributed almost entirely to pre-allocation profitability rather than portfolio management needs.
The defense countered with what it called the CRA Exercise: NYU Stern professor Bruce Tuckman, a former chief economist of the Commodity Futures Trading Commission, was given dashboards of portfolio and market data for 50 randomly selected trading days and asked to allocate 935 of Leech's trades blind, without knowing first-day performance or Leech's actual decisions. Charles River Associates vice president Aaron Dolgoff, who designed the exercise, proposed to testify that Tuckman's allocations matched Leech's 56.8 percent of the time, a result he said was statistically significant, and that Tuckman's Macro Opps allocations showed a statistically significantly higher win rate than the overall trade pool.
Woods rejected the government's bid to exclude the CRA Exercise as an untested novel experiment, finding that the design, execution, and analysis reflect that both experts were employing the level of intellectual rigor that might be expected in their respective fields. He also turned aside the defense's challenge to Cohen and Kolm's use of end-of-day and recorded-allocation-time prices as proxies for when Leech actually decided to allocate trades, ruling that the assumption was not so unrealistic and contradictory as to suggest bad faith and that the defense's evidence of a lag between Leech's oral instructions and his assistants' data entry was quintessentially a factual dispute for the jury.
On the sharpest legal question, whether the government's experts crossed the line drawn by Federal Rule of Evidence 704(b) by opining on Leech's state of mind, Woods sided with the government. He found that testimony that statistical results are consistent with Mr. Leech using information about the first-day performance of his Treasury F&O trades in making allocation decisions stops short of expressly stating the inference that he acted knowingly, willingly, and with an intent to defraud. The opinion drew a direct analogy to a narcotics detective testifying that drugs found in a car were more consistent with distribution than personal use.
Woods did flag three phrasing concerns: the government experts' use of using in connection with allocation decisions, Kolm's use of influenced, and Tuckman's description of his duration and convexity targets as proxies for Leech's market views. He ordered the parties to meet and confer within two weeks on mutually agreeable language. He also invited the defense to propose a limiting instruction on evidence from the post-October 2023 period, when Leech lost discretion to allocate trades to the Core Strategies.
The trade blotters at the center of the case span 595,236 rows and 48 columns, totaling 28,571,328 cells of data, which both sides' experts have spent months mining for the statistical signature of a fraud that the government says cost Leech's Core Strategy clients hundreds of thousands of dollars per trade.