Viewing profile — alpark3

You're thinking of Optiver's The Hammer, though every HFT firm has basically done or is doing something like this. [1] Jane Street's version of this was absolutely intentional. [1]…

> Since puppies turn into full grown dogs quite quickly, how often do you suggest I replace the puppy? You must complete the mission before the puppy becomes a dog. Otherwise you m…

The pattern I've noticed with a lot of open source LLMs is that they generally tend to underperform the level that their benchmarks say they should be at. I haven't tried this mode…

I agree. Good human friends should provide a mix of a positive/feedback negative loop, providing a good gradient to train one's behavior on. AI seems like just a positive feedback …

They're not that good at arb. Real arb doesn't even really exist anymore. Even when it does, it's not JS that closes it. They market make, which is different.

Because it's lying to the client?

Maybe I should clarify that I don't think it's all that matters, but that Sam might think that.

I wonder if Sam did something in the name of his own philosophy, but was financially suicide. Like vastly underestimating the costs of training/inferencing to the board, but justif…

It's true at last anecdotally. Related, there's a running joke in some subset of the industry about the types of exotic derivatives that are so complex and esoteric "only french ba…

1/100th of a pede

> He was so good at finance that at his previous job at a hedge fund, he supervised its biggest day of losses in company history. Small irrelevant detail, but nags me nonetheless. …

Honestly an idea I've had with the recent surge of interest in LLMs is some type of robotic pet. I feel like LLMs can understand human language and images(hopefully soon) to the ex…

Natenberg

Most derivatives traders I know in the industry do some version of buy-and-hold for their personal portfolios, but one of the best I know does something completely different. He st…

I think the next big innovation in LLMs (sort of like the attention mechanism) will be some method of distributing work to much smaller, specialized, and capable units, rather than…

When Meta released a highly censored LLaMa, I think it was pretty clear that over time, the market would tighten up and release progressively more powerful uncensored models until …

I think(hope?) Llama 3 will be a MoE architecture that shows >GPT-3.5 level performance. Interesting to think Meta will probably continue to spearhead the open source AI movement.

> The new voice capability is powered by a new text-to-speech model, capable of generating human-like audio from just text and a few seconds of sample speech. I'm more interested i…

Honestly, the current general state of software may be a good proxy for what AI will settle to. I can imagine major open-source models, trained and generated by nonprofit efforts i…

One of the articles that was on the front page was from koshka.love. Wonderful website brimming with personality. What are some other examples of Web 1.0 personal websites that you…

Most people confuse market making/risk holding with high frequency statistical arbitrage strategies. I'm not totally sure exactly what Knight Capital was running, but generally the…