Hype Check
12 questions
Why is machine learning on financial data especially difficult?
Little signal, a lot of noise, limited independent observations, and relationships that change over time.
Where does the professional user's edge lie?
In data, infrastructure, execution costs, and bundling many weak signals, not in the model itself.
What is non-stationarity?
Relationships change over time, partly because market participants react to their exploitation.
Where can a gain on a meme coin come from?
Exclusively from a later buyer. There's no income or value creation.
What follows from heavily concentrated holdings?
A few addresses can move the price substantially, which favors manipulation.
How should such a position be classified?
As a consumption expense with a bearable total loss, not as an investment.
Why are leaderboards skewed?
Because with many participants, some perform strikingly well by chance alone, and those who dropped out aren't visible.
What incentive does compensation by copied volume create?
More risk-taking, since the copied trader shares in the success but doesn't bear the copiers' losses.
Why do copiers' results typically lag?
Because bundled simultaneous execution shifts the price against them.
What do you technically own with an NFT?
An entry on the blockchain pointing to metadata, which in turn points to a file.
Do you acquire copyright with the purchase?
Not automatically. What matters is the specific license terms.