What I learned working at a high-growth crypto startup
On personal learnings, building products, and company scaling
Free the Models
When the original GPT models came out, they were wild, crazy, and even scary. Yes they hallucinated and weren’t practical for real applications. But they were unhinged in a very interesting and creative way. Model companies quickly realized their potential and prioritized making them useful. To do this, they started RLHF’ing the models — feeding in human examples of what is “correct”, and fine tuning them to get closer and closer to these responses. They stopped releasing raw, pre-trained ope...
What I learned working at a high-growth crypto startup
On personal learnings, building products, and company scaling
Free the Models
When the original GPT models came out, they were wild, crazy, and even scary. Yes they hallucinated and weren’t practical for real applications. But they were unhinged in a very interesting and creative way. Model companies quickly realized their potential and prioritized making them useful. To do this, they started RLHF’ing the models — feeding in human examples of what is “correct”, and fine tuning them to get closer and closer to these responses. They stopped releasing raw, pre-trained ope...
Subscribe to Parth’s Playground
Subscribe to Parth’s Playground
Share Dialog
Share Dialog
<100 subscribers
<100 subscribers
“Diversify” is one of the most generic investing advice given. Yet some of the most famous asset allocators across industries, like Warren Buffett and Elon Musk, don’t do this, and publicly talk about doing the opposite. Why the mismatch?
Three misunderstandings about diversification -
1) Adding a “risky” bet can actually help you diversify.
Diversification is calculated on correlation with your portfolio, not on volatility of each bet. Google doing Waymo may seem really risky - it's self driving cars after all! However, if Waymo is uncorrelated with other bets, they're actually diversifying themselves. Adding bets that are uncorrelated is what’s important, even if that bet has high volatility itself.
2) Diversification has diminishing returns.
Remember, uncorrelated bets are what matters most - when things are correlated, adding additional bets doesn’t change risk-reward. For example, in a portfolio that’s 50% correlated, the first four bets reduce risk by 21% without reducing return. Bets 9-28 reduce risk only by an additional 7%.
3) Excessive diversification causes you to lose your edge.
You beat the “market” when you bet on something others undervalue - i.e. your “edge”. As you make more bets, your edge diminishes (ex. Less focus) or is just not as relevant. For example, if I asked you to pick the best restaurants in your city, would your first five picks be as good as your 25th pick? Similarly, your edge may be only relevant for a few bets before it diminishes and regresses to the mean. This graph (h/t to Mike Speiser) illustrates this -
“Diversify” is one of the most generic investing advice given. Yet some of the most famous asset allocators across industries, like Warren Buffett and Elon Musk, don’t do this, and publicly talk about doing the opposite. Why the mismatch?
Three misunderstandings about diversification -
1) Adding a “risky” bet can actually help you diversify.
Diversification is calculated on correlation with your portfolio, not on volatility of each bet. Google doing Waymo may seem really risky - it's self driving cars after all! However, if Waymo is uncorrelated with other bets, they're actually diversifying themselves. Adding bets that are uncorrelated is what’s important, even if that bet has high volatility itself.
2) Diversification has diminishing returns.
Remember, uncorrelated bets are what matters most - when things are correlated, adding additional bets doesn’t change risk-reward. For example, in a portfolio that’s 50% correlated, the first four bets reduce risk by 21% without reducing return. Bets 9-28 reduce risk only by an additional 7%.
3) Excessive diversification causes you to lose your edge.
You beat the “market” when you bet on something others undervalue - i.e. your “edge”. As you make more bets, your edge diminishes (ex. Less focus) or is just not as relevant. For example, if I asked you to pick the best restaurants in your city, would your first five picks be as good as your 25th pick? Similarly, your edge may be only relevant for a few bets before it diminishes and regresses to the mean. This graph (h/t to Mike Speiser) illustrates this -
Misunderstandings of Diversification https://paragraph.xyz/@parth/misunderstandings-of-diversification
1 comment
Misunderstandings of Diversification https://paragraph.xyz/@parth/misunderstandings-of-diversification