Jin Choi (@jinchoice) profile picture

PhD financial math + full stack software developer. I'm no longer active here. Find me on Threads @jin.choice or Mastodon @jinchoi@mstdn.ca

Kelowna, BC
Joined March 2009
Photo by Jin Choi (@jinchoice)Photo by Jin Choi (@jinchoice)Photo by Jin Choi (@jinchoice)Photo by Jin Choi (@jinchoice)Photo by Jin Choi (@jinchoice)Photo by Jin Choi (@jinchoice)Photo by Jin Choi (@jinchoice)Photo by Jin Choi (@jinchoice)Photo by Jin Choi (@jinchoice)
This will be my last post here on X/Twitter. I've decided, as many others have, to migrate to Threads. I attribute part of this decision to business logic - Twitter is dying. Why waste energy on a platform that will die? But on a more visceral level I just can't stand Musk.
1
1
351
Say what you will about other billionaires, but I don't they match his level of narcissism. So, I'm taking my business elsewhere.
1
1
157
On a separate note, I've become a registered financial advisor in Canada, and started blogging regularly on eddywealth.com/ I have very high regard for many of the people here, so I hope we can connect elsewhere.
146
Testing. Testing. (I'd changed my username a few weeks ago)
1
361
There were ~1,000 users when I joined my #MastodonSocial server (mstdn.ca) about a week and a half ago. There are now nearly 20,000 users. The poor server admin is going nuts trying to scale resources.
1
1
Unless you operated large systems, it’s hard to appreciate both how they work, and how they break. Twitter won’t go down with a big b, and it won’t go down irreparable. Things will work fine for a while - thanks to no new deploys! Then, here’s what’s likely to happen (cont’d):
Good to hear. Having no outages means there will never be outages, no new capacity issues, no new cascading failures, no bad pushes, nothing. We will freeze in time and my tweets will go into /dev/null. I wished I was smart enough to come up with this idea at work.
69
707
228
3,574
I want Twitter to go down so I don't have to suffer terrible takes like this
Twitter now has early seed stage level company chaos but with thousands of employees and billions in revenue and entirely in the public eye. Absolutely wild to watch.
"We’re accumulating technical debt much faster than before" And that's why I have zero faith that Tesla will ever achieve full self driving.
This Post is from an account that no longer exists. Learn more
1
My new blog post. I've spent way too many months writing this one, but I think it was worth it. Please give it a read!
[Paper Review] The Virtue of Complexity in Return Prediction enjine.com/blog/paper-review…
1
1
No talented engineer I know dreams of working at a big telco like Rogers. I've always wondered if the lack of tech talent would lead to issues like this one. I'm actually surprised there hasn't been similar issues at big banks.
UPDATE: Rogers still doesn't have an estimated time for when its services will return, but a new statement from the company mentions the possibility of “proactively crediting all customers” who were impacted by Friday’s outage. trib.al/YBR804O
Beggar-thy-neighbour policies are back in global food markets. And the political consequences would be pretty bad — as it was the case in 2007-08 and 2010-11. #OATT #wheat
5
46
5
173
For our most recent M6 competition submission (Apr 24), our model said $VXX had over 60% chance of placing in the top quintile of performance among all US stocks + ETFs. I thought that couldn't possibly be right, so I fudged the numbers down to 45%.🤦‍♂️
2
Surprised that huge funds would bother owning penny stocks like $GMBL *Disclosure: I own some
🤯
Not gonna lie, didn't expect "illustrators" to be one of the first human jobs that artificial intelligence completely displaces but here we are
1
Thought provoking comment from Nadal. I feel that it's applicable to startup life as well. Nadal doesn't lie to himself during tough situations, and I imagine that frees him to overcome it.
2
One surprising aspect of machine learning is the amount of customization required for a model to be useful. You look at some problems and go "hmm, a simple multilayer perceptron (or linear regression) will work". But then it doesn't.
1
You end up having to throw way more at the problem - Bayesian methods, different neural net structures, different optimizers, until you finally get a model that works well enough to be useful, all the while knowing that there's probably a better solution out there.