Conversation

And here comes Sam. He's willing to dispute a time-tested and well respected probability theory formula in order to make a point that what? Uniswap sucks and Serum is our savior? Puhleeze Sam.

Quote Tweet

SBF

@SBF_FTX

Dec 2, 2020

Replying to @_charlienoyes and @danrobinson

It's trying to maximize EV[log(wealth)] instead of EV[wealth]. This is probably wrong in and of itself as an assumption to make. But if you *do* want to make that assumption, then e.g. you have to consider all of your assets that don't have anything to do with the pool.

Replying to

I'm so confused! what "probability theory" am I disputing? The part of Kelly I'm disputing is not the math part, it's the question of what utility function you decide to maximize in the first place.

Replying to

and

Kelly tells you which utility function should be used and also tells you that log(wealth) dominates any other strategy bar none.

Replying to

and

It even comes with a mathematical proof that in the long run it beats any conceivable strategy.

SBF

@SBF_FTX

Replying to

@Bcdrdi

it comes with a proof that, given a very specific set of assumptions (that you must always use the same strategy and not adapt, that the process is repeated *literally infinitely*, etc.), as time goes on, the odds kelly wins go up and the amount it is losing by in EV go up too.

1:31 AM · Dec 3, 2020Twitter Web App

Replying to

and

Using the same strategy and not adapting is not a problem whatsoever if that strategy gives you the most money in the long run. The infinite repetition is not a requirement, it's just that the longer you keep betting the less the influence of randomness on your wealth.

Replying to

yes, and the more and more true it is that: a) kelly is likely to win b) if kelly loses it loses by infinite amounts c) kelly falls behind in EV d) kelly gets ahead in median