AA - Abhishek's Attic (place to store stuff to clear my head)!

Investment Learning
1

The Quant Journey

With 2 recent books on investors / traders who were predominantly quants, there is seemingly an awakening of quantitative methods amongst investors in India. The two books are The Man Who Solved The Market on Jim Simons and A Man For All Markets by Ed Thorp. Let me take some time to pen my journey down. This is a developing story and I am probably in the first chapter.

Flashback
My journey into quant systems started about two and a half years back.

image
image666Ă—720 208 KB

It showed, not sure how accurately though, the CAGR of some of the best investors (or traders) in the world. Scanning the list I gathered that I knew about nearly all of the people mentioned. What was interesting was that the names on the list came from completely different investment styles. Some were hard core bottom up stock pickers, some were macro traders, some technical traders, some commodity traders and some quants.

I have always been a curious person. And I always try to go back to first principles. So, I asked myself that if there are people who have been this amazingly successful over decades being non-fundamental, non-bottom-up investors, then am I missing something? Have I drunk too much of the Buffett kool-aid?

I decided to make the decision for myself. So, I started learning. I took a course of technical analysis. I started reading voraciously on quantitative systems, automated trading systems, mathematical and statistical indicators. The whole objective was to understand and assimilate as much as possible of these into my own accumulated knowledge of long term investing.

There were two other triggers to this. The first one was during our long walks on the beach during the VP annual meets in Goa with Hitesh bhai (Hitesh Patel) I realized that he had been very successful at blending technicals with fundamentals and was able to follow many more stocks and get very good results though with a higher amount of churn. That was worth thinking about.

The next thought that triggered was I wanted to be able to teach my son about investing and the road I took seemed very long and arduous and everyone has to start from scratch. I was looking for a structured process that could be taught and learnt relatively easily and in shorter time.

Cut to Now
The proof of the pudding is always in the eating. So, to prove and enhance my learning, I have worked on multiple algorithms and have picked two of them and have started investing for the last 3 months. They are mechanical, systematic, non-discretionary systems. I have a friend, who shall go unnamed for reasons of privacy, whom I collaborate with closely in this.

All the words are important. The system is mechanical meaning there is no human intervention during decision making. It is systematic meaning that buy-sell decisions are time based and pre-determined. Non-discretionary means once the algorithm tells me to buy something, I buy. I do not rationalize the choice.

The Results, Thus Far
September 2019: 0.81% (Complete)
October 2019: 22.06% (In Progress. Sell Date 1-Jan-20)
November 2019: 3.56% (In Progress. Sell Date 3-Feb-20)
December 2019: -1.02%(In Progress. Sell Date 2-Mar-20)
XIRR (till date): 58.8%

The results thus far have been excellent. To say the least. I have backtested this for various durations and periods. Since the availability of data is a major challenge, I have been able to backtest only for slightly more than a dozen years. What has given me confidence is that the backtest period covers the 2008 and 2018 periods.

More about the system in the next post. Also, will try to log my learnings in this thread for future reference.

DISCLOSURE: I am a SEBI registered RA and run an equity advisory service. This and all other posts are for educational purposes only. They should not be construed as investment advice.

86 Likes
3

Fascinating! While we eagerly wait for your next post, here’s a talk by PPFAS’s Rajeev Thakkar on the same subject.
PS - it’s probably best to watch Rajeev talk at 1.25x

14 Likes
4

Very important overarching principle across domains

Some such examples -

  1. Weight Training - If one studies the weight training routines of Ronnie Coleman and Dorian Yates, one can see some clear differences. The former was by far one of the strongest bodybuilders ever known who started as a powerlifter, one can see videos on youtube of him squatting 800 lbs - an insane amount of weight for a bodybuilder. Whereas Yates was a proponent of high volume training, he would smash each body part to bits with multiple high rep sets. Both were multiple Mr Olympia winners and dominated their respective eras

  2. Clearing the CAT exam - Circa 2005 I was a very active poster on the largest MBA exam prep forum in the country. I was struck by varied styles there too - some junta would smash Verbal and just about clear the cutoff in Quant while some would do exactly the opposite. What often got underweighted was pure ability while the strategy to tackle the exam was over weighted, mostly due to prep forums like TIME and Career Launcher telling junta that anyone could crack CAT. Elaborate strategies were being drawn up by aspirants without spending enough time honing basic skills! Long story short - my strategy was to go in with no particular strategy and see the exam for what it was. It meant hopping from section to section and getting a very good sense of the nature of the paper by the 1 hour mark. Of the last 1 hour I spent 40 mins on Quant which eventually made the difference in clearing the cutoff for the section.

  3. Tennis - See the differing styles of RF and Nadal, they have distinct styles. Djokovic on the other hand can beat of these at their own game - his strategy changes based on the opponent

Across domains one can notice that there are multiple ways to succeed, one has to choose the optimum method based on the core skills and temperament that one has. Sometimes one has to play the field.

But then for this, one has to experiment first and fail a few times! What I find really concerning in today’s narrative dominant investing environment is the lack of appreciation of this basic fact. Anyone who comes on TV and trumpets one style of investing is basically a salesman and not an investor, he is there trying to drum up more business for his fund

A more mature approach is to do what Abhishek is doing here, this way you fail multiple times at smaller impact cost than fail big. When one tries out various approaches, the probability of seeing the holes in one’s own approach goes up significantly, more likely that the blind spots will reduce in impact if not in number.

People understand this in all other aspects of life, only when it comes to this holy cow of investing do people start debating on what is the “only” approach - value investing, growth investing or quality at any price? The ultimate objective of investing is to make money, we should not be guilty of intellectualizing and romanticizing the game too much

A word of caution - In the earlier stages as a novice investor, one is better off sticking to a tested approach and keep things simple. Once you have a good amount of experience and can actually think independently, it makes sense to experiment a bit before finally settling on what is your core approach. Do not experiment too much as a beginner, that is a sure way to silly errors and eventual frustration

42 Likes
5

While I agree with your overall thoughts, however if one looks at the broader perspective, I mean other than making your own money by whatever style, then all these growth investing, investing at quality at whatever price and even high frequency intraday trading, all of these- what they are doing is making the entire stock market just a giant ponzi scheme. The actual value of a stock today lies not in the stock or in the company and its business, but in how much it can be sold to another person. I can discuss more but that will move this thread in another direction.

Regardless, this thread is intriguing indeed and I am very interested in future posts of Abhishek related to this methodology.

6 Likes
7

You need to wait for the subsequent parts. I am not getting time to write. Hope to do it shortly. One of the reasons I started to write this thread is to dispel the common misconceptions, ignorance and lack of understanding of a different style of managing money. (The other reason of course, is to clarify my own thoughts. That’s why I write anyways).

But, to give a very short answer now, quantitative investing or trading can be used in all ways. It can be used to do value investing, deep value investing, growth investing, GARP, momentum or anything else. As long as the various aspects can be captured numerically, it can be quant-ified.

In fact, Ben Graham, in his last interview said this:

“I am no longer an advocate of elaborate techniques of security analysis in order to find superior value opportunities. This was a rewarding activity, say, 40 years ago, when our textbook “Graham and Dodd” was first published; but the situation has changed a great deal since then. In the old days any well-trained security analyst could do a good professional job of selecting undervalued issues through detailed studies; but in the light of the enormous amount of research now being carried on, I doubt whether in most cases such extensive efforts will generate sufficiently superior selections to justify their cost.”

“Essentially, a highly simplified one that applies a single criteria or perhaps two criteria to the price to assure that full value is present and that relies for its results on the performance of the portfolio as a whole–i.e., on the group results–rather than on the expectations for individual issues”

Link here - http://www.grahamanddoddsville.net/wordpress/Files/Gurus/Benjamin Graham/A Conversation with Ben Graham - Financial Analysts Journal - 1976.pdf

Anyone more interested in understanding why the godfather of value investing questioned his own techniques can read this book - Amazon.in
image

11 Likes
8

A few more examples:

EXAMPLE 1:
Walter Schloss, a legendary value investor, and one of the superinvestors of Grahamsville had these features in his investing style:

  • Cigar butt investing over several decades
  • Owned 100+ stocks
  • Not interested in underlying nature of business
  • Looking for stocks hitting new lows
  • Trading lower than book value
  • Very little debt
  • No company visits. No talking with management.

Nearly, all these are traits of quant investing.

EXAMPLE 2:
Tweedy Browne Portfolios are built using the following construct, which are all quant based:

Low price in relation to asset value
Less than book value
Less than net current assets
Low price in relation to earnings
Low PE/High Earnings Yield
High Dividend yield
Low price in relation to cash flow
A significant pattern of purchases by one or more Insiders
A significant decline in a stock’s price
Small market capitalization

Details of their approach are found in https://www.tweedy.com/resources/library_docs/papers/WhatHasWorkedFundOct14Web.pdf
This is a must-read for all serious investors.

EXAMPLE 3:
Take a look at the Nifty Strategic Indices - https://www.niftyindices.com/indices/equity/strategy-indices

By definition, every index is a quantitative one (based on fixed rules and NOT on someone’s discretion). Nifty strategic indices have many strategies and provide a wonderful glimpse into the world of quants at a very very high and broad level.

image
image1252Ă—860 81.1 KB

13 Likes
9

Any investment framework needs to encompass the following:

  • Universe selection
  • Stock selection
  • Portfolio allocation
  • Selling

The focus of the majority of books on investing is only on stock selection. However, for good portfolio performance, all the 4 aspects are important. Think of times when one or more of the aspects have gone wrong and how thing unfolded. For example, allocating too little or selling too early can make a huge difference to the portfolio returns.

This is true irrespective of whether you are investing or trading. It does not depend on your time horizon or the methodology of stock selection. It is not important if you are selecting stocks based on fundamental analysis, technical analysis, macro trend analysis etc. It is also not important whether you are following the value, growth, GARP, momentum or any other style.

These 4 aspects are constant are always true. The idea of quantitative investing is to put into mathematical terms conditions for all these 4 parameters.

Let us look at Joel Greenblatt’s magic formula. For those who do not know about it, can read it here - Magic Formula Investing Definition

  • Universe Selection
    S&P 500.

  • Stock Selection
    Rank selected companies by highest earnings yields and highest return on capital. Select top 20 or 30 stocks based on ranking.

  • Portfolio Allocation
    Equal weights for each stock

  • Selling
    Sell everything after 1 year and repeat the process.

Now, this example should quell the fears of those who are ready to jump at hearing the word quant that it is hi-frequency trading or some such ignorant criticism.

For those who believe in passive investing in index ETFs like Sensex or Nifty also need to understand that, in fact, any index is also primarily a quant system.

This is the index selection criteria for Sensex. See how it is setup and has a review frequency of 3 months.

image
image1876Ă—486 87 KB

The advantage of following a quant system is:

  1. It takes away the emotional challenges and brings in a systematic approach.
  2. You can theoretically backtest the result of your system to see how it would have performed over time including times of extreme stress, volatility or exuberance.
  3. It brings a lot more discipline to the process of investing. You should objectively follow this practice of thinking through the 4 aspects highlighted here even if you are doing bottom-up fundamental investing.
21 Likes
10

Can you advise on the website or tools you are using for Backtesting. I am fed up with the lack of tools in Indian markets. Streak is useless and other websites do the basic stuff

1 Like
11

Will come to it in due course… best option is to build your own backtesting system.

1 Like
12 
For those who believe in passive investing in index ETFs like Sensex or Nifty also need to understand that, in fact, any index is also primarily a quant system.

This is a very good and easy to understand example, which has given better than average MF performance for other categories. This can be transparently verified by anyone.

image
image622Ă—684 34.4 KB

By separating emotions from investing philosophy, quant space will be an important one to track.

14

Microsoft Excel is best for doing backtesting

15

https://www.quant-investing.com/strategies
This might be useful for someone trying quantitative investing.
I personally prefer momentum, low volatility and quality over value and small cap. In value investing one will have to use other factors to prevent value traps.
Quality portfolio will not have finance companies as roce and debt to equity parameters will be used. Other than that it performs well over long term.
Nifty alpha low volatility is by far the best performing index with respect to risk and return over 5yr periods.
Passive investors can use quant ETF launched recently.

3 Likes
16

The Current System

I have been working on multiple strategies across disciplines for the last 2 years based on value, growth and momentum. The objective is to put a part of the portfolio into 1-3 such strategies to generate hopefully better than market returns and then focus and concentrate on the discretionary part of the portfolio for generating alpha.

The current strategy I am testing out for the last 4 months now and which I have backtested for the last 12 years of data, is as follows:

Universe Selection - Nifty 500

Stock Selection - Pick the top 10 stocks from the filtered list thrown by the system

Portfolio Allocation - Equi-weighted, ie. 10% each

Selling - Rebalance after 3 months

To reduce risk and increase diversification, more because I am new to this, I have decided to run this strategy every month with 1/3rd my capital allocated for this strategy.

Illustration:
Assume, I want to allocate Rs 100,000 to this strategy. I will then buy 10 stocks given by the system on Jan 1. I will allocate Rs 33,333 on this block. Sell them on Apr 1.

On Feb 1, I will buy the second tranche of Rs 33,333 of 10 stocks again. Sell them on May 1.

I will buy the last and final tranche of 10 stocks on Mar 1, with the last Rs 33,333, which I will sell on Jun 1.

In effect, every block of 10 stocks remains in the portfolio for 3 months. From the third month, the portfolio will have a maximum of 30 stocks. The maximum initial allocation for a stock can be 10% (if it comes every month for 3 months in the list of top 10 stocks).

As of now, I do not have a stop loss system in place. But I am working on how to enhance this by taking into account overall market movement and individual stock movements.

The backtested returns have been slightly better than 45%+ cagr over the backtest period with a maximum continuous drawdown of around 43% between Dec 2007 to Mar 2009. Maximum drawdown for a quarter has been 33%.

The actual results to date have been better than expected. I will post the month-end results after the 31st.

12 Likes
19

Seems interesting. I HV two Q’s to ask.how would u select 10 stocks out of 500 every time.some of them, you may like to continue with? Considering that you are making profits,all these transactions would invite short term gains tax.

20

The selection of the top 10 is part of the algo. And yes, it would mean the entire set of gains will be short term capital gains in nature. Given that today there is not much of difference between short and long term capital gains tax rates, I am not considering it much. If the situation changes I will have to adjust accordingly.

I am also working on another longer term strategy which will have a reset period of 1 year. If it works out then a combination of such strategies could work better.

21

Great sir.keep sharing these wonderful ideas and thanks for yr quick detailed reply

1 Like
22

Hi @basumallick
in the strategy backtest, how does one account for the changes being made in the Nifty 500 components throughout these 12 years of data that you have back tested?

Does the algorithm time the entry and exit of all these components accordingly? For instance if a stock has performed well and is added to the Nifty 500 index on January 1, does the backtest record gains from January 1 or later (Assuming its returns are part of top 10%)?

1 Like
23

@basumallick dada , which are your favorite non investment related list of books for multidisciplinary learning or otherwise even just for entertainment? Many thanks

24

Is this Momentum Investing?

25

dada that is really good this will eliminate any emotional bias , Could you please share what filters are you are using to converge in to final 10 picks .Or the heck list you are going to utilise to arrive at final 10 value theme in terms of valuation or growth in terms of sector or stock in domain. It is very difficult and if I am not wrong it is almost next to impossible to find the stock with 100% accuracy that will rise 5% or di 5% . I like your strategy to allocate the capital in steps in phased manner. But suppose a stock like if chosen like ITC or HUL or Nestle which is not in your first 10 picks falls hypothetically 12 to 15% should one replace the bottom performer with the steady compounder ? or stick with the top 10 .
Regards