Pirlo_Jiang

# I use 17 parallel accounts to try to falsify my own trading strategy
> A serious experimental record about prediction markets (Polymarket): from "finding the edge" to "personally overturning my own edge"
---
## Preface
Over the past while, I’ve been doing one thing: **finding a method for stable profitability in prediction markets**.
But I don’t want to write an article that says "I found a surefire strategy"—because in this field, **the more certain you sound, the more likely you are to be wrong**.
What I want to write is something else: **a set of experiments designed to falsify itself**.
This article will cover three things:

1,278 trades I ran myself, plus 1,529 settled markets across the whole market—two sets of independent data point to the same sentence: the money doesn’t lose because of the judgment, it loses because of the cost. And 99.5 cents is the only exception.
This article has two ledgers. The first one is ugly; the second is the first thing I can confirm since I did my two-week experiment.
First trade: I ran 1,278 deduplicated trades, for a total loss of +$2,960.41.
Second trade: there’s a price level where, with our own sample plus the whole market sample, there were 218 entries in total—none of them ever failed.