The DyDX Odyssey: Zero to Hero, with a Side of Comedy
I discovered DyDX through a newsletter and became fascinated with automating crypto trading via its API. The appeal was irresistible — it was like setting a Roomba loose in a casino.
The Rise
The Kelly Criterion Epiphany
My first serious attempt used the Kelly Criterion formula for position sizing. Though it promised safety, the conservative approach frustrated me as profits grew slowly and trades eventually halted.
The Kelly Criterion calculates your optimal bet size based on your edge and odds. It's mathematically sound. It's also extremely boring when you think you've found a gold mine.
Trusting My Gut — The Taco Bell Syndrome
Abandoning mathematical discipline, I began trading on instinct. Initial success proved intoxicating, but the streak reversed fast. What goes up must come down — and I squandered my gains through emotional trading decisions rather than following proven risk management.
It felt exactly like a regrettable fast food run at 2am. You know it's a bad idea. You do it anyway. You regret it immediately.
The Fall and Recovery
The Fall — Murderous Monday
A catastrophic market event liquidated my $2,500 portfolio overnight.
Undeterred, and drawing parallels to bombing during stand-up comedy performances, I returned to the drawing board. I documented my analysis in a Jupyter Notebook and reintegrated the Kelly Criterion formula:
| 1 | f* = (bp - q) / b |
Where:
f*= fraction of portfolio to betb= net odds (profit per unit wagered)p= probability of winningq= probability of losing (1 - p)
Rising from Ashes
My transparent documentation gained community attention, leading to speaking opportunities where I humorously presented crypto trading parallels to dating — both involve high risk, emotional decision-making, and occasional spectacular failures.
I rebuilt my trading bot "Phoenix" with the Kelly Criterion as its foundation, eventually recovering approximately $1,000 across three months.
What's Next
What's Next? A GitHub Star or Hollywood Dreams
I published my code publicly while planning a book and exploring integration of Bayesian probability programming with Kelly sizing — creating anticipatory rather than reactive trading strategies.
The journey from zero to hero wasn't a straight line. It was more of a chaotic scribble that eventually pointed upward. And that's okay.