Post New

There's quite a bit of interest in Machine Learning over at Portfolio 123. This book sounds interesting:

Introduction to Statistical Learning
"I recently came across this book by Stanford University and it may be the best one I have seen yet on the topic."
https://community.portfolio123.com/t/introduction-...

The book is available for free at https://www.statlearning.com/
A free online companion course is available through edX.

"we're trying to launch our AI/ML factors this month in beta. It's a huge deal for us and very big, costly project."
https://community.portfolio123.com/t/can-we-have-a...


I think Machine Learning will soon be applied successfully to stock pricing. I doubt I will be the one who programs this, but am interested in the topic. What effect will Machine Learning have on Mechanical Investing in the next 5 years? It could be an opportunity as people invest in dubious ML ideas. Or the market could become more efficient.

This has been discussed a few times:

The link in this post does not work.
2020 Subject: AI investing post "I posted on the MD board some observations about AI and investing, and I used to post here and it relates to Mechanical Investing, I'll reference for friends here rather than double-post:"
http://www.datahelper.com/mi/search.phtml?nofool=y...

2017 Subject: Reinforcement Learning AI & Ln() "I'm building some reinforcement learning (RL) AI code (in Python)."
http://www.datahelper.com/mi/search.phtml?nofool=y...

2016 Subject: AI "Given the tools and principles already at hand, might it not be possible to autonomically mine available databases for ever more consistently successful strategies? "
http://www.datahelper.com/mi/search.phtml?nofool=y...

2011 Subject: OT:Could AI lead to a perfectly efficient market "Seems like science fiction and fantasy, but as the years tick by we're getting closer and closer."
http://www.datahelper.com/mi/search.phtml?nofool=y...

2002 Subject: an AI Raltive Strength Screen... "I am 100% scientifically convinced that the AI mode works better then Relative strength, and gives an added value. Secondly i am 100% scientifically convinced there is no need for monthly retraining the prove will follow."
http://www.datahelper.com/mi/search.phtml?nofool=y...

2002 Subject: An 1970-2001 AI Backtest "the matrix was fully 1970-2001 trained"
http://www.datahelper.com/mi/search.phtml?nofool=y...

2001 Subject: Jamie its time for some AI paraphrased: "Take the TOP 10 from a screen, look at the ranks (Valueline, Rank, etc.). Calculate average and standard deviation. Define TT = Average / Standard deviation. Rank the properties on TT and pick TOP TT"
http://www.datahelper.com/mi/search.phtml?nofool=y...

I think Machine Learning will soon be applied successfully to stock pricing. I doubt I will be the one who programs this, but am interested in the topic. What effect will Machine Learning have on Mechanical Investing in the next 5 years?

My own expectations:

People have been trying REALLY hard to beat the market for centuries.
Every idea that has been invented has been tried, and there is no shortage of capital to fund the attempts.
All the fancy ideas in the last few decades have been automated, many with very fancy adaptive software methods.
A new type of automation is relatively unlikely to come up with new ideas that work so well they change the landscape.

There are presumably still lots of market inefficiencies to exploit.
I suppose machine learning bots might accelerate wiping out the few remaining obvious ones that still work and are subject to being arbitraged away, if there are any left.
But other than that? Just another player in the market.

I add the phrase "subject to being arbitraged away" because some things aren't.
Momentum is the simplest example: if someone tries to front run a momentum factor, it just adds to the momentum at a very slightly shorter lookback.
The exact mix of momentum strategies that work best will continue to shift, but I can't see any way they can be eliminated by having more participants.
It's probably much the same for the end-of-month effect. When there are more participants it might start (and end) a couple of days earlier, but doesn't cease to exist because of that.

Speaking of well funded quant software projects, I guess we all wish we had invested in the Medallion Fund and been allowed to remain.
$100 invested at the start of 1988 was worth $398 million by 1998...and that's after the shocking fees of 5 and 44 for most of that time.
Allegedly: Average hold period ranging from a-day-and-a-half to a-week-and-a-half, percentage of winning trades around 50.75%.

Jim

Lizodal:
There's quite a bit of interest in Machine Learning over at Portfolio 123. This book sounds interesting:

As a P123 subscriber I've been following the flow but so far, their existing system is limited. The data vendor doesn't allow direct access to the raw data except with a very large premium. You must rely on their factor ranks for machine learning. Some of the posts that make it seem a piece of cake gloss over many of the details.

The only alternate relatively low-cost access to the required data is still Stock Investor Pro. I've slowly been trying to rebuild a historical backtesting data set for Machine Learning. Turns out it isn't as easy as it first seemed. The difficulty is compared to a backtester like GTR1 where each run you end up with a small number of ranked stocks, each of the current members have a known purchase and sales dates. Easy to accounts for slippage, splits and dividends. ML evaluation data sets assume a fixed holding period, most use either 1 month or 1 quarter, an assumed 1 month hold eliminates almost 2/3's of the dividends. I'm ending up ignoring the dividends for ML and then reevaluating the selected portfolio's performance the conventional way with friction, splits and dividends.

My favorite book for ML is 'Hands on Machine Learning with Sciket-Learn, Keras & Tensor Flow' by Aurelien Geron.

Seems like over the last 25 years for MI to remain profitable it keeps getting more and more complex. Complicated by the fact that I seem to have less drive & compulsion to squeeze a few extra percent out of the investments.

RAM

Complicated by the fact that I seem to have less drive & compulsion to squeeze a few extra percent out of the investments.

This is OT but I think that's a general tendency when getting older. Decades ago I tried with statistics and later with neural networks to make money with stocks, programming for a while day and night on my systems. Money came (by old fashioned work), age came, priorities shifted. That shift of priorities I think is the norm in the 2nd half of life. I pity people who have money and who's goal nevertheless only is to make more (other than it being a fun hobby, like it seems to be for Warren Buffet).

The Medallion fund had spectacular success, and yet is less than 1% of the total market. That makes it difficult to see the effect of Medallion on the general market. An AI that could reproduce WEB's stock picking might be possible in 20 years.

May 2023:
Berkshire Hathaway market cap $700 billion
Renaissance Technologies AUM $106 billion
total US stock market cap $44,000 billion

April 2021:
Berkshire Hathaway market cap $580 billion
Renaissance Technologies AUM $131 billion
total US stock market cap $44,000 billion

Berkshire Hathaway started earlier. WEB bought Berkshire in 1965. Medallion started in 1988.

==== links ====
"Renaissance Technologies is a hedge fund with 14 clients and discretionary assets under management (AUM) of $106,026,795,439 (Form ADV from 2023-05-01)."
https://whalewisdom.com/filer/renaissance-technolo...

The Medallion Fund Is Still Outperforming. Other Renaissance Funds Still Aren't, April 19, 2021
"At the end of the year the regulatory assets under management of Medallion, which includes leverage, amounted to $34.8 billion, according to its annual filing with the Securities and Exchange Commission. Regulatory assets for the entire firm were $130.86 billion."
https://www.institutionalinvestor.com/article/2bsw...

Top 10 largest hedge fund managers 2022
Renaissance Technologies $57 billion as of June 30, 2022
https://www.pionline.com/largest-hedge-funds/2022

"Bad ideas is good, good ideas is terrific, no ideas is terrible."

...like it seems to be for Warren Buffet

Aaahhh!
I did what I always find so ignorant when others are doing it!
Sorry, Warren Buffett!

...like it seems to be for Warren Buffet

Don't sweat it.
Buffet is easier to spell than Smörgåsbord.

Jim

Post New

US Policy FAQ

---

Contact Shrewd'm
Contact the developer of these message boards.