Quant Investing

[basumallick]

@malthankar

Thanks for explaining…just thinking if we buy say 1 st April, 1 st may, 1 June…so invest money again on 1 st July… will you sell all April stock on 30th June so capital would be available or if we sell 1 st April stock in July sometime then we need additional money to invest in 1 st July …so in total I will need around 13.33 lacs as there could be overlapping of some days in July till we sell 1 st April basket…am I missing something on this?

You will sell on 1st July and buy on the same date. The constraint is that you have a fixed capital, say 10 lakhs, in this case.

[basumallick]

@Gaurav_Rungta

I was wondering since your strategy is a buy and hold for three months -done each month, do you provide higher weight age to the previous three month performance as compared to the whole year both in terms of technical and fundamental factors in the screen?

In short, yes.

@Gaurav_Rungta

Since the strategy is run every month and not just after the quarterly results, do you keep tabs on news etc related to the companies currently in the basket and may decide to alter stoploss based on some adverse events?

No. The assumption is that all news ultimately gets reflected in the price, so there is no point in trying to interpret news. And to be truthful, it is too difficult for me, at this time. I do have some thoughts on adding social media and sentiment analysis for my fundamental investing, but nothing as of now for quant.

@Gaurav_Rungta

I see that the April tranch was in the blue but you say it was stopped out and held in cash. Do you keep a trailing stop? Do you keep or plan to keep the stopped out cash in liquid fund for the remaining duration?

I have replied to the stop loss mechanism in earlier posts. It is a slightly complex one with multiple layers of stop losses. I don’t put the money in liquid funds. I keep it in cash.

[Donald]

Hi Abhishek,

Do you think you can share constituents of the Quant Portfolio performance? Whatever is possible, may be 3 months prior, as these are already executed. And not share current/previous months performance, if there are constraints.

Proof of the pudding is always in the eating. Any performance stats without relevant data points backing up, is NOT that instructive, or really value-additive for fellow-learners.

This is just my viewpoint. You can ignore it, if you think differently.
But I do hope you can accommodate that in your approach of sharing what is turning out to be a potentially promising addition to our investing strategies.

[basumallick]

Hi Donald,

I already had that in mind. Will be posting the actual quantletters once there useful life is over. Other than the list of stocks, it also has a short writeup related to quant investing, so I presume would be useful for that as well.

So, here are the Mar and April quantletters.

Disclaimer: These are for educational purposes only and not to be construed as investment advice.
Quantletter_Q30_Apr2020.pdf (301.8 KB)
Quantletter_Q30_Mar2020.pdf (222.2 KB)

[homemaker]

@Gaurav_Rungta

I have started with basic momentum and low volatility screens with a 3mo time frame for each investment set. Few screens I am using are positive to negative return days>1, low SD in daily returns over the year and past three months etc.

How to filter low volatility out of a set of say top gainers. Is there any screener, filter available or you use your software.

[Gaurav_Rungta]

I am going for minimum volatility in returns, so i calculate SD of daily returns in excel for each shortlisted stock. The toughest part is good data collection. For that I am currently using Yahoo Finance(free).
The process of volatility calculation can be easily found on google.

[hack2abi]

Hi Gaurav,

I have created this sheet where you can quickly check up on a stock’s price history.

Using Pivot Tables/Charts, Slicers & Screener to Semi-Automate Historical Peer Analysis

I am going to make this thread a repository of all the investing tool kits that I create for myself. Here is the 2nd tool kit. A google sheet that provides in-depth analysis of a stock’s price history. Functionalities as of today include: Return and CAGR since listing. All-Time high, 52-week High/Low and % difference between CMP and these milestones. Drawdown % and days. Average, Median, Standard Deviation, Maximum, Minimum, of Monthly/1/3/5/7/10 year rolling return periods. Win/Loss scenari…

[arpans]

Is there some quant strategies available which will profit as and when markets correct?

[basumallick]

You can device quant strategies for anything that can be expressed in numbers. However, it is difficult in India as short selling is not allowed. Only shorting is possible using futures and options.

There are lot of breakdown strategies possible even using index futures. Simple Donchian channel or Bollinger Band breakdown strategies, when applied consistently along with a regime filter will give decent results.

[eyesice]

I have a basic query regarding selection of stocks for backtesting. Obviously one cannot choose the current index constituents because backtesting them would mean our strategy is likely to perform well. They are there in the Index for a reason!

So how do I get a historical list of NIFTY50/SENSEX 30 constituents so that I can use those stocks at various points in time while running the backtest?

[basumallick]

These would be available from the different data providers or also in the bse archive data.

[vithalpai]

@basumallick Great Thread. Thanks for posting this.

How this strategy will perform in sideways market. More than 50 % of time market is moving sideways.

[basumallick]

Your observation that 50% of the time market is moving sideways is based on recent observations of the last 2-3 years. Historically, that has not been the case. We see very strong bull and bear markets with some intervening time in a trading range (sideways market).

Any growth or momentum-based strategy will perform very well in a trending market but no so well in a sideways market. But in general, is should do better than normal index returns during most times.

[basumallick]

Using a Regime Filter

A regime filter or a market regime filter is a tool to help us conceptually understand the kind of market we are in. As a systematic investor, we can increase our odds of success by adding a regime filter to our arsenal. It tells us, based on how we have defined it if we are in a bull market or a bear market. We would think differently about market risk in different market scenarios.

A simple example of a regime filter is using the 200 day moving average. If the index of your choice is above the 200 day moving average, then you define it as a bull market and below it as a bear market. You can design your portfolio strategy to hold full allocations in stocks if you are in a bull market and 50% allocated in a bear market.
So, with that basic logic you can start constructing a slightly more realistic and slightly more nuanced regime filter.

First, define the market conditions you want to address – superbull, bull, bear, superbear. The reason for doing that is you want to be cautious in the market extremes of superbear and superbull conditions and aggressive in the bear and bull conditions (for long-short strategies). Then use a combination of indicators like RSI and 50 & 200 day moving average to define the selected conditions. For example, above 200 dma and 70 RSI you define as superbull and above 200 dma and above 50 RSI as bull phase.

Another trick that can be used is to use multiple indices. For example, you can use the average of Nifty, Nifty Next 50 and Nifty 500 in equal proportions to define your market. For a long only investor, it may increase the odds of success to be buyer only when the regime filter is indicating a bull market.

[basumallick]

Adding the May 2020 Quantletter. This basket is now closed.

The May basket was up +24.53% including brokerage (0.15% both ways … we use a discount broker).

Note: these are real results and NOT backtest results.

Quantletter_Q30_May2020.pdf (267.8 KB)

[Gaurav_Rungta]

@basumallick @vivek_mashrani
What are your views on risk parity and portfolio % allocation strategies.
I think beginners are better off with an equal allocation model (Keeping it simple works better most of the times ). But do you guys allocate based on say: retuns volatility , ATR etc.?

[basumallick]

Risk parity is more a portfolio allocation strategy and usually not used in single asset class portfolios. There is no study internationally which proves conclusively the high volatility strategies are higher risk and consequently lower return.

[bhavik.1991]

Very interesting thread sir, thank you for sharing.
Can you please share quant letter June month as Reset period is already passed?

[basumallick]

Sorry… I have been really busy with multiple things and have not been able to keep up with the thread here. Will continue to post on quant thoughts in the upcoming months.

Meanwhile, here is the June Quantletter.

Quantletter_Q30_June2020.pdf (712.7 KB)

[agarwalk2]

Loved this initiative Dada! I have a simple question: My observation is that this system captures much of the upside while the market goes up. In your opinion, what pretends the severe drawdown during market downturns(as observed during backtest)?