Her marker hovered over the third line. She tilted her head, read it again. Then she circled a small cluster of numbers near the bottom of Harrington's original equation. A decay constant buried inside a summation. "This is wrong," she said quietly.
Harrington uncrossed his arms. "Excuse me?"
"The decay constant," Whitney said, tapping the board. "It is set at 0.07, but the underlying asset class you're modeling, high-yield corporate bonds, has a historical default decay closer to 0.043. You've inflated the decay rate, which compresses the tail distribution and makes your model appear more stable than it actually is." Nobody in the room breathed. "In plain terms," Whitney continued, still facing the board, "your model tells investors the risk is low, but it's not. It's hidden. And when the market corrects, and it will, the losses won't be 15%, they'll be closer to 38." She wrote the corrected number on the board, drew a line through Harrington's, then she kept solving. Harrington pushed off the table. "That number was verified by my entire research division. 12 PhDs. You're telling me you know better than 12 PhDs?"
Whitney didn't turn around. I'm telling you the number is wrong, sir. Based on what? Your waitressing degree? He looked at his guests for support.
A few smiled, but fewer than before. The room was shifting, and Harrington could feel it. Like sand moving under his feet. The silver-haired investor, the one who had checked his laptop earlier, stood up. His name was Richard Callaway.
He managed 11 billion dollars in pension funds. He walked to the whiteboard, adjusted his glasses, and followed Whitney's math line by line. 30 seconds passed. A minute. Callaway traced her work with his finger, pausing at each correction. Then he turned to Harrington.
She's right. Two words. They hit the room like a brick through glass. Murmurs broke out. A woman in pearls set down her champagne and picked up her phone.
Two investors at the back table exchanged a glance. The kind that says, "We need to talk about our money." Harrington's jaw tightened. The smirk was gone. "That's one error," he said. His voice was clipped, strained.
"One. Anyone can find one mistake if they stare at a board long enough."
"It's not one error, sir." Whitney was still writing. She hadn't turned around. "The inflated decay constant feeds into your correlation matrix on the second panel. That skews the copula function, which means your joint default probabilities are off by a factor of 2.3. Every downstream projection in your model inherits that bias." She drew three arrows connecting the cascade from the decay constant to the correlation matrix to the final output.