alpark3

8 ms·

1.
by alpark3 1y ago
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] https://www.reuters.com/arti
2.
Interim Order in the Matter of Index Manipulation by Jane Street Group(sebi.gov.in)
7 points by alpark3 1y ago|0 comments
3.
by alpark3 1y ago
> 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 must wait 14 years until you can get another puppy.
4.
by alpark3 1y ago
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 model yet and am not in a position to for a couple day
5.
Beating the Crowd(withentropy.com)
57 points by alpark3 1y ago|11 comments
6.
by alpark3 2y ago
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 loop.
7.
by alpark3 2y ago
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.
8.
by alpark3 3y ago
Because it's lying to the client?
9.
by alpark3 3y ago
Maybe I should clarify that I don't think it's all that matters, but that Sam might think that.
10.
by alpark3 3y ago
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 justifying it to himself because it's all going towards
11.
by alpark3 3y ago
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 banks" trade them.
12.
It's Fucking Impossible to Stay Healthy(withentropy.com)
18 points by alpark3 3y ago|29 comments
13.
by alpark3 3y ago
1/100th of a pede
14.
by alpark3 3y ago
> 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. Jane Street is not a hedge fund. They don't remotely
15.
by alpark3 3y ago
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 extent that a pet should be able to. Some kind of robotic
16.
by alpark3 3y ago
Natenberg
17.
by alpark3 3y ago
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 sticks to a philosophy of scanning multiple "small"
18.
by alpark3 3y ago
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 having one giant network. We already see hints of this with
19.
by alpark3 3y ago
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 demand was met. I'm glad to see this happen in real tim
20.
by alpark3 3y ago
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.
21.
by alpark3 3y ago
> 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 in this. I wonder how it performs compared to other c
22.
by alpark3 3y ago
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 in a roughly similar fashion to Linux. Entire businesses migh
23.
Ask HN: Web 1.0 style personal websites?
1 points by alpark3 3y ago|0 comments
24.
by alpark3 3y ago
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 only "little" mistakes that would cause
25.
by alpark3 3y ago
It's interesting to note that exchanges are adding "obvious error" rules that can slash trades under certain circumstances.
26.
by alpark3 3y ago
From the title, I thought Procreate was coming out with another crappy stable diffusion-derived AI image generator. Thank god that's not the case. Very excited for this.
27.
by alpark3 3y ago
> I mean...nothing in this world is unlimited for real. Maybe don't call it unlimited storage then :D
28.
by alpark3 3y ago
_If_ 3.5 is a MoE model, doesn't that give a lot of hope to open source movements? Once a good open source MoE model comes out, maybe even some type of variation of the decoder models available(I don't know whether MoE models have
29.
by alpark3 3y ago
Some quant jobs interview both. The fermi estimation problems are much less emphasized, but they're there nonetheless.
30.
by alpark3 3y ago
>Scope: Where commitments mention particular models, they apply only to generative models that are overall more powerful than the current industry frontier (e.g. models that are overall more powerful than any currently released models, i

More ›