/biz/ - Business & Finance

File: 341 KB, 1200x720, 19F95FC9-8164-43C6-936A-2553E1BF7469.jpg [View same] [iqdb] [saucenao] [google]

Let’s talk the medallion fund Anonymous Mon Sep 19 10:56:50 2022 No.51511205 [Reply] [Original]

There MUST be someone that’s willing to share their exact mathematic models and strategies. 66% of yearly performances, how the fuck we can’t track down all of their movements and make our own models based on that alone?

>>	Anonymous Mon Sep 19 11:02:32 2022 No.51511258 >>51511205 Deep web stuff, infos, anything welcome

>>	Anonymous Mon Sep 19 11:05:36 2022 No.51511292 What do their 13f filings show?

>>	Anonymous Mon Sep 19 11:08:29 2022 No.51511321 >>51511292 What?

>>	Anonymous Mon Sep 19 11:38:57 2022 No.51511593 >>51511205 Their edge comes from the nsa spying. Imagine feeding their algos with confidential data from big companies and govs around the world

Anonymous Mon Sep 19 12:52:01 2022 No.51512501
File: 505 KB, 1142x1143, 13678201557077.jpg [View same] [iqdb] [saucenao] [google]

>>51511205
I try not to get killed here:

>Non-linear dynamics and chaos
>Attractor reconstruction
>Hierarchical agent-based trading systems
>situational awareness
>human bias, trivial system bias

Climate Change is dogwhistle for Peak Oil Mon Sep 19 13:36:54 2022 No.51513157

>>51511258
>>51511205

It was never one single thing. Often they can to keep changing their techniques because the market would figure our the inefficiency they are using and then it would disappear into randomness.

Early on there was a lot of trend following.

Later it became some about using something like Hidden Markov Models called Conditional Randon Fields over the https://en.wikipedia.org/wiki/Cepstrum of market prices.

I assume even that stopped working. So I suspect they are using Deep Reinforcement Learning now.

>>	Anonymous Mon Sep 19 13:36:54 2022 No.51513156 >>51512501 where is this image from?

>>	Anonymous Mon Sep 19 13:44:37 2022 No.51513321 >>51512501 90% of this is pseudo trash. Regular math is enough, they are hardly breaking new ground over there.

>>	Anonymous Mon Sep 19 13:47:25 2022 No.51513370 >>51511205 Only way to beat the market (outside blind luck) is with inside knowledge.

>>	Anonymous Mon Sep 19 14:06:12 2022 No.51513691 >>51512501 Damn. The only thing I knew about the fund is that they use huge amount of datas but the system itself isn’t very articulated. This sounds plausible. Where is this shit from? Anything else you can share with us?

Anonymous Mon Sep 19 14:33:32 2022 No.51514085

>>51513157
I've listened to many talks by Jim.
If I remember correctly he use some form of differential geometry and non-linear PSA.
I've tried to understand these concepts but I don't think I'm smart enough (~ 127 IQ).
I can grasp parts of it but it takes a toll on my emotional state and I think my understanding is very superficial.

Anonymous Mon Sep 19 14:49:34 2022 No.51514364

Probably just a jew scammer. Of course HE is the best with the best team and the keys to the universe. The only thing guys like him do is slosh money around for mossad. Or front run retail or inside trade. Has he ever proven any of his performance in the fund that actually does massive numbers? I thought his main larger fund didn't do that well recently. Since when can mathematics predict the future

Climate Change is dogwhistle for Peak Oil Mon Sep 19 15:08:05 2022 No.51514673

>>51514085
> differential geometry and non-linear PSA

The problem with differential geometry is that is that it cannot handle randomness very well. And the markets are full of randomness.

If by non-linear PSA you mean Nonlinear principal component analysis, then yes there is a good chance Nonlinear principal component analysis is part of the analysis. But it cannot be the final step, because what it does is it chooses simpler way to explain the data. So if they are using Nonlinear principal component analysis it has to be used at the start of the pipeline where they are cleaning up the data.

In later steps some type of conditional probability estimator has to be used. The state of the art is Deep Learning. But prior to Deep Learning we had Conditional Random Fields.

I have worked with all of the above except differential geometry. I have even tried to find patterns like this but they don't last. My broker starts doing side bets with me, and the pattern disappears within a month.

Anonymous Mon Sep 19 16:22:04 2022 No.51515743
File: 53 KB, 640x796, 1662825345316322.jpg [View same] [iqdb] [saucenao] [google]

>>51514364
Probably both. They have some skill but also some shady things are going on. Like how they started out with this weird piping business is very dubious. Might as well be some form of money laundering.

Anonymous Mon Sep 19 16:47:57 2022 No.51516168

>>51511205
as general approach
>trend following
https://www.youtube.com/watch?v=EPdKS7BZ7M8 1:20
>contrarian
https://www.youtube.com/watch?v=EPdKS7BZ7M8 3:17

technically speaking
>clustering, regressions
https://www.ece.ucsb.edu/Faculty/Rabiner/ece259/Reprints/tutorial%20on%20hmm%20and%20applications.pdf

>machine learning (deep learning)
https://youtu.be/QNznD9hMEh0?t=1770

extremely good results for swinging can be achieved by simply using smoothers, trick is to find and fine tune hyper parameters to phase response of a signal (price data), something i'm currently struggling with

mind you, as much as they dealt with stocks, I believe biggest bulk of their profit came from options, which somewhere in "My Life as a Quant" by Emanuel Derman (chapter 13 I think), he describes how someone realised while stocks are stochastic processes with infinite numbers of outcomes, options future valuation is limited and more predictable, que Black–Scholes model and it's derivatives

>>51513321
also this

Advanced search
Text to find
Subject [?]Search by post subject. Leave empty for any.
Username [?]Search for user name. Leave empty for any user name.
Tripcode [?]Search for tripcode. Leave empty for any.
Email [?]Search by email. Leave empty for any.
Filename [?]Search by image filename. Leave empty for any.
From Date [?]Enter what date to start searching from. Format is YYYY-MM-DD
To Date [?]Enter what date to start searching until. Format is YYYY-MM-DD
Image hash
Search in	All Posts OPs Only
Deleted posts	Show all posts Show only deleted posts Only show non-deleted posts
Internal posts	Show all posts Show only internal posts Show only archived posts
Order	New posts first Old posts first
Capcode	All Posts Only by Users Only by Mods Only by Admins Only by Developers
Results	Posts Threads
Action	[ Simple ]