Some Questions around Momentum Investing
Since my first blog post highlighting my own move to Momentum Investing, I keep getting queries by friends and acquaintances who wish to learn more about it. The number of queries have increased in recent times though many are in telephone calls and not shareable with others. Once in a way, I do receive very thoughtful questions that I think require more than a simple answer. Yesterday, I was asked one such set of questions by Prashant.
I think these questions and my views on the same can add value to a better understanding and hence rather than it being a one to one conversation, am posting this here in the hope that it helps clear some doubts that others may have themselves.
1) First thing, would like to discuss about the conducive environment for momentum investing –
Many researchers and authors suggest keeping track of sentiment, be it breadth of the market or some kind of trend indicator (200/50 DMAs) for overall market trend. My point is one or other sector or stocks are always there which shows momentum even if broder market breadth is not good or range bound or even falling/downtrend, So is it a good idea to be invested all the time – based on backtest and purely from your experience ?
My View: In his book, Stocks on the Move, Andres Cleanow writes and I quote
I will declare the market to be bearish if the S&P 500 Index is below its 200 day moving average. That’s a very long term filter. Using such a simple approach, we immediately have a firm way to identify if the market is in a bear trend or not. Practically all equity portfolio strategies can be improved significantly by simply adding this one rule. If the index is in a bear market, just don’t buy stocks
Andres Cleanow
While theoretically this is sensible, there is an issue when you look at the data. For the Indian Markets, I shall use Nifty 500 as the Proxy. The issue is whipsaws. From 2010 to 2020, Nifty 500 has crossed below its 200 day moving average 40 times (average of 4 times a year). Using an envelope rather than the 200 average reduces the trade numbers but adds a bigger lag on either side.
Even without bothering about Timing Luck which would impact this tremendously, this constant moving to cash and getting back to full position has a cost. Personally I don’t see value in adding this filter even though this could have gotten me partially out of the market in March of this year (I would be back fully only in August and hence lost quite a lot on the way down while losing the opportunity on the way up as well).
But when market breadth as measured through say % of stocks trading above their 200 day EMA is on the lower end, the performance of Momentum like any other market strategy will be below par. Hence a better strategy will be to sit out when % of stocks trading above their 200 day EMA drops below a certain parameter and pick it up only when the broader trends are back in place.
Because it’s not as volatile as the 200 day EMA on a Index, this has a lower number of trade-offs. But trade-offs shall remain regardless for there is no Free Lunch.
Personally since I started, my portfolio has been completely invested but for one month (April 2020). I do keep working on new ideas and strategies and willing to change if evidence points to a better way to invest.
2) Second point is Exit, somewhat related to first point and of utmost importance –
Is it better to exit all positions when sentiment for the whole market turned bearish based on above trend criteria OR can follow stock specific checks be it fixed SL(5% or 10%) or Trailing SL or momentum ranking criteria (if it falls out of list lets say top 20 or 40 depending upon one’s strategy of ranking/sector allocation/volatility). What is the best practice given the kind of experience you have ?
My View: As outlined above, exits when we feel the market is bearish (based on quantitative measures) has its drawbacks. I don’t feel it compensates for what we are aiming for – lower drawdowns in any meaningful way other than once in a blue moon kind of event.
Stocks are generally volatile and stocks that are making new highs (and hence generally tend to be part of the Momentum Portfolio) are more likely to be a bit more volatile. Having a fixed percentage stop is injurious to returns. Each stock has its own volatility and trying to force every stock to align with a single stop will guarantee bad returns.
I use WorstRankHeld as the criteria to decide whether to hold the stock or exit. This is not strictly speaking required – you can just exit any stock that moves past the portfolio size and replace it with others. But the reason I use is to limit churn in the portfolio. Churn has a very high cost (both visible and invisible) and lower the churn, lower the cost we pay.
I don’t believe that a stock in momentum has to continuously rocket upwards all the time. Many a time stocks do move into a range before launching into another fresh rally. Constant checking and weeding out such stocks will only mean that you aren’t allowing much time for the stock to make its move.
To give a recent example – Birla Soft got into my portfolio in October 2020 (a tad too late maybe). Stock decided to drop right after my entry. But in the ranks, did not go below the WorstRank held and this ensured that the stock wasn’t dropped like a hot potato. The next month (November 2020), it moved up enough to get back to my purchase price. It was only in late December that the stock finally made its move. Without a worst rank held, this would have gone out. There is always a chance that the replacement could have done better, but could have done worse as well.
I have tested for both and find it kind of a yin and yang. Sometimes cutting off once it drops below 30 works better than Worst Rank Held while some other times, it works bad. But overall, I have seen that the difference it makes isn’t too huge and if I can get the returns without too much of a churn, I am game for that.
3) Third point is allocation –
There are quite a few theories related to this: few researchers suggest equal weight portfolio while few suggest volatility based using
20 ATR and start allocating funds from rank 1 based on volatility of stock till the fund exhausts.
Sectoral allocation – Some suggest keeping sectors in check but some suggest to invest even if all the stocks come out from only one sector.
What’s your point of view ?
My View: This is a very good question.
Let’s start with portfolio construction. Equal Weight, Weight based on Trend Strength, Inverse Volatility weights, Market Cap based weights are among the one’s generally used. I have tested for Volatility and Equal weight and found not much of a difference in returns. Bigger difference and volatility happens due to Portfolio Size. Equal weight is simple and easy to adjust when one is adding new capital and hence my preferred choice.
Sector Allocation can be controversial. In August, I made this tweet
Not much has changed since then. Pharma remains the highest weight in the current portfolio. Is this Risky – of course it’s Risky. But sectors and industries don’t generally fall off the cliff so as to speak with little time to adjust one’s position.
In early 2018, 80% of my portfolio was in Small and Mid Caps which had just then made a high that is yet to be broken. But in the course of a couple of months, the portfolio switched out to Large Cap Stocks. But this 80% exposure was what gave me 60% returns in the period between May 2017 to December 2017. While some of it was given back, a large part was retained.
Rather than an overweight in a sector, I am more concerned with being overweight in a single stock. This is because stock level risk is always much higher and hence my willingness to cut down stocks that have gone above a certain weight in the portfolio. Reversals in Individual stocks can be nasty with very little time to adjust one’s positions.
Sector concentration risk is also a reason I choose to go with 30 stocks even though it may not be the most optimal. It’s unlikely to find the top 30 stocks all belonging to the same sector. But this is entirely possible if you are buying just 10 stocks for instance.
4) Fourth point is Momentum/Ranking and about entry –
Have read many articles ranging from simple strategies like
(a) simple rate of change(max gains) to
(b) 90 days linear regression kind of complex strategies to
(c) Again rate of change over 3,6,9,12 months period.
My point of view while deciding the rank/momentum to give more weightage to short timeframe gainers – e.g.,
let’s say I want to rank on the basis of 12 months return but weighted kind of formula while ranking want to take care of all 3,6,9,12 months returns and apply some kind of function which gives more weightage to 3M returns than 6M returns than 9M returns than 12 returns and assign ranking for my universe of stocks. For one stock, I want to consider all four (3,6,9,12 M returns) timeframe in weighted terms.
Any suggestions on this ?
My View: Keep it simple.
Currently the trend is with a lookback of 6 months, previously it was with 9 months and before that it was 12 months. But the shorter the lookback, more the churn and more the volatility. Trying to optimize on everything I feel is a fool’s errand. Rather go for one that allows the maximum amount of capital to be deployed. It doesn’t matter how great your system is if you can only risk say 10% vs a lower yield system but one where you can live with a 70% exposure.
Adding too many constraints in backtests opens up risks in terms of data mining and curve fitting. There are a lot of assumptions already built into any backtest, you don’t want to load them up even further.
Recent Comments