muggermuch

Thank you for this comprehensive response!

I have often found myself struggling to explain the difference between building a strategy or trading system (which reduces to a technical/intellectual challenge) and running a hedge fund (essentially running a complex information-driven business).

Your cost breakdown really puts matters into perspective.

> it bodes well for the OP that they talk about market regimes

I concur. Market regimes (modeling, detecting, reasoning about them) are too delicious of an intellectual puzzle to resist.

> Also would this approach work with crypto?

Some of it works on crypto. TBH I've stayed away from the asset class, but only because I find it difficult to build mental models and think about features (in my mind, it's a mix of commodity factors and currency factors, but I'd have to test it out).

I seem to remember coming across papers that have tested momentum factors at larger time-frames (e.g. weeklies).

> Could you recommend a base system / reference one could use to get started which explains or bakes in some of the feature / signals engineering work?

The references I put in at the end of the post will really help with this! I might actually write out a separate blog post about starting out in this space from an ML perspective. Thanks for the idea!

Thank you!

>How did you did risk management? I put in a basic position management layer (1% fixed stop). Also, the market regime module would modulate participation, i.e. in really risky environments it would dial down the number of stock picks. I can definitely do much more on this front, but I wanted to nail down the stock picking first! :)

>Have you been leveraged? No leverage.

>Have you paid for data? Yes, my monthly running costs for data are ~$1.2k.

:) hmu on LinkedIn!

>You should be running a multi-billion $ Quant fund if this is the case.

You seem to underestimate the level of effort and rigor required to achieve this level of capital allocation. In contrast, beating the market by 20% is table stakes. Folks in the industry do it all the time; the difference here simply is that I built an ML-powered engine to do it systematically.

Indeed. I haven't shut down development, just shut down the newsletter. I'm continuing to work on it.

There are commercial feeds available via Nasdaq DataLink (FKA Quandl). I also bought bulk historical data to feed through my backtester (I haven't talked about this in the post; it was getting to be a bit too long).

Thank you!

That's a good idea. I'll try it out, thanks!

It's difficult. We made a lot of pitches. Investors/allocators require a fairly long track record and are extremely reluctant to fund (what they perceive to be) black box strategies.

Great question.

If I beat the market by 20% (say SPY generated 0% for the year, very optimistic at this point), and I have allocated $100k to this, I make $20k before taxes.

That's less than minimum wage.

Meanwhile, allocators expect a track record of at least 3-5 years.

Ideally, if I have an asset, I'd like to extract as much revenue as I can.

Hope this makes sense.

Thank you for appreciating the article; I tried to disclose all that I could!

1. Yes, I did put my own money in it (low 6 figures).

2. It went as described in the article - for the capital I allocated to Didact, I beat the market (SPY) by ~20% since inception.

3. If I understand your question correctly, this would be the equivalent of the payoff on an optimal lookback option (https://en.wikipedia.org/wiki/Lookback_option). I haven't actually done that analysis, but it sounds like a nice idea.

Thanks!

Hi, fellow HN'ers! Author here, please let me know if you have any questions or thoughts!

This is excellent, thanks for sharing!

>> dated, a little wrong about some things, but not insane (like _Applied Cryptography_ is).

Why is "Applied Cryptography" insane? As an interested bystander, I'm curious to know more.

+1 absolutely - it's been one of my pet peeves. There really needs to be some thought put into "doc UX".

So people can determine for themselves whether they want to download the PDF, by reading the abstract first. This has always been a point of petty annoyance for me as well.

Your podcast is very useful - I have been a subscriber for a couple of years now!

Thank you for doing what you do.

Tensorflow supports Metal AFAIK.

I'm guessing all technological advances performed by US tech companies are post-fixed with information about starving Yemeni children whose parents were bombed out of existence. /s

Sometimes one can appreciate tech advances without apportioning blame for other societal ills at the doorsteps of folks who are focused on making lives better for all.

PS: I'm not Israeli/Palestinian. I don't have a dog in this fight, just came here to appreciate the news.

This is amazing! Thank you for the link.

This looks really cool! However, as an Evernote and Notion customer (yes, I use and pay for both), I would love to know what would make me switch tools, or add this to my subscriptions?

Thanks for the solution! It's elegant and is a great fit for my current work which involves retrofitting my codebase with type annotations.

Step 4: ???

Step 5: Profit!!

As a side note, I am amazed that SoftPanorama still exists. Nothing seems to have changed since I last saw it almost 20 years ago! This website was one of the few fast-loading ones that I had bookmarked as a high schooler trying to teach myself CS on a cheap/flaky dial-up connection back in India.

Good times.

I can attest to this in the case of the Data Science course. My wife, a non-CS PhD, was interested in changing her career track. We were referred to General Assembly (the SF location) by one of my coworkers and the curriculum looked comprehensive enough, so she signed up.

About a month into the course, to our despair, we discovered that there was almost zero conceptual training imparted (I am a data scientist by profession and can speak to this). The instructors seemed to be working there as a stopgap measure while looking for other jobs, and the students were there for standard credential-seeking. The course material meanwhile was focused on answering "How" but not "Why", which is far more important when approaching data science problems.

To fix the situation, we quickly pieced together a curriculum based on Tibshirani & Hastie's machine learning video lectures[0] and picked a few Kaggle challenges for practice. My wife went through this material over a month, and came out with enough of a mental framework to be able to tackle book-length material such as ISLR[1].

Finally, there was almost zero career support provided apart from generic advice around how networking is important and how one should try to get outside their comfort zone.

Though all's well that ends well (my wife is working as a data scientist now at a seed-stage Valley startup), in retrospect we often talk about how we shouldn't have wasted money on this course. She didn't learn anything that she couldn't have picked up on her own, not even practical skills re: data analysis with Python/R, and her student cohort was quite disinterested in making connections.

Ce la vie.

[0] http://www.dataschool.io/15-hours-of-expert-machine-learning... [1] Introduction to Statistical Learning, http://www-bcf.usc.edu/~gareth/ISL/

NewsBlur is excellent, and a great way to collate all my feeds across categories (quant finance, ML/data science, CS/tech, science, humanities). I have been a happy subscriber since 2013, and I use it on an approximately daily basis.

Re: Coinbase, I have it setup to debit my bank account every day. The transactions go through between 2.45-3 PM EST daily, with 3% fees.

However, I just tried setting up a new transaction and the fee looks to be $0.99 for a $5 investment. At ~20%, this sort of averaging would no longer be worth it. Like I said before, I guess I've been grandfathered in after the new fees schedule was imposed. Let's see how long it lasts.