A simple daily tool that may help time the current market: A Covid Market Fear Index


Like many people at the moment, I am trying to get a feel for when to get back into the market. The last real comparable situation was the 1918 Spanish Flue (H1-N1 virus) [1], but this time it seems different.

I am currently at home, on lock-down in France and recently read a Bloomberg article by Simon Flint entitled; – “Five ‘Ps’ That Will Help Determine If the Market Has Bottomed” [2]. The article made me think about constructing a simple measure for those holding cash, based on easily accessible, publicly available data, as a tool to help time re-entry into the market.

Quantitatively, there are many ways to model and measure the temperature of the market. This is however a novel shock event and therefore backtesting a model is not really an option. Instead I have used some standard risk/fear measures, but also incorporated elements specific to the Covid-19 outbreak and also tailored it to try and reflect some of the ancillary topics that the market is currently focused on, such as the oil price.

The initial result of this, is a measure I call the Covid Market Fear Index, or CMF Index. It is made up of eight equally weighted components, which are simply averaged. The higher the number, “the greater the market fear”. The current value is around 58.1 (as of the 3rd April 2020). Below is a graph showing the index and its various components over time, since the start of February 2020.

I plan to update the CMF regularly for my own proprietary use. 

If you would like to receive updates, you can contact me either via LinkedIn, email or via WhatsApp.  Below is a description of the current CMF Index methodology;

Each of the eight factor has been transformed to be approximately in the range to 0 to 100, being winsorized to 100 on the right tail . The variables were selected to reflect some of the main factors that the market appears to be focusing on at the moment, however this list is by no means exhaustive!

CMF = 50 x log(Max[Covid new deaths, 100]) – 100 + 40 x log(Covid new cases/1000) + (VIX price-17) – (100 x mean[S&P 500 & WTI drawdown]) + Min[ (Initial claims – 200,000)/10,000, 100] + (50 x (0.7 + ANFCI) + (50 – median[PMI]) +Google trend Interest

Below is an explanation of each of the factors and the transformations applied

1. & 2. Medical: New daily COVID-19 deaths & new cases

For the number of new daily reported deaths the values arebcollated in a spreadsheet [3], and transformed by taking logs and scaling. Depending on the time of day, there is generally a one day lag in the number.

source: Worldmeter
source: Worldmeter

3. Risk: The VIX

The Implied Vol of the S&P 500 is a well know fear measure [4]. The subtraction of 17 is used, as it is roughly the long term median value. Note the VIX also forms a small component of the Chicago ANFCI.

Source Bloomberg Markets

Possible improvements? It might be worth exploring changes in the Vix, other implied vol or variance measures

4. Market: S&P 500 & WTI Drawdown

I take an equal blend of the S&P 500 Index & WTI Crude future drawdowns. The drawdown for the S&P 500 is from a max of 3393.52 and for WTI the max is $53.78. Both are multiplied by x100 with the sign reversed and then averaged. Note, during pre US open, I adjust the spot price by the current one day front month S&P Future’s return, to approximate the out of opening hours value of the index.

YTD Equity drawdowns (local currency). Source Bloomberg Markets

Possible improvements? It maybe worth incorporating the tail risk of other risk assets.

5. Macro: Initial Claims (IC)

Initial Claims, shocked the market last Thursday, so It seems sensible to try and capture this effect [5]. The rather convoluted formula, is simply to rescale the IC change.

Source Bloomberg/Bureau of Labor Statistics

Min[ (Initial claims – 200,000)/10,000, 100]

Possible improvements? IC is just one of many short term macro/economic indicators

6. Economic: ANFCI

The Adjusted National Financial Conditions Index (AFNC)I is a weekly indictor published by the Federal Reserve Bank of Chicago [6]. Positive numbers tend to suggest market tightening in which case the market may underperform, while negative values imply loosening and possible inflation risk.

Source: Federal Reserve Bank of Chicago

A full list of the indictors used can be found at: https://www.chicagofed.org/~/media/publications/nfci/nfci-indicators-list-pdf.pdf.

7. Business Sentiment: PMI

I take the median of the Purchasing Managers’ Index™ (PMI) across a number of regions and countries [7].


Wherever possible I use the latest Flash Composite numbers as they are reported, however in the case of Canada and ASEAN, manufacturing numbers are used.

8 Crowd Sentiment: Google trends

There has been a lot of interest in recent years in the use of sentiment data as an alternative orthogonal data set, to more traditional price based, fundamental and macro data. A number of approaches can be used to capture sentiment, from google trend searches [8], to twitter feed data and more sophisticated approaches, such as using NLP and machine learning to parse a large corpus of text from news feeds or research reports.

For now, I just use the average of the global daily popularity of the search of the following 5 terms [9]

recessionunemploymentGold priceCredit default swapspandemic
Interest over time for the term “unemployment” Source: Google


The material in this post is purely for educational purposes and in no way constitutes investment advice. I may hold positions in any assets mentioned in this post. Furthermore, I am not responsible for any material that is found at the end of links that may be in this text.


[1] Centre for Disease Control & Prevention, 1918 Pandemic (H1N1 virus), https://www.cdc.gov/flu/pandemic-resources/1918-pandemic-h1n1.html

[2] Flint, S, 1st April 2020, Five ‘Ps’ That Will Help Determine If the Market Has Bottomed, Bloomberg, https://www.bloomberg.com/news/articles/2020-04-01/five-ps-that-will-help-determine-if-the-market-has-bottomed

[3] Worldometer, April 2020, Cornavirus Cases: Daily Cases (worldwide), https://www.worldometers.info/coronavirus/coronavirus-cases/#daily-cases

[4] CBOE Volatility Index, Yahoo, https://finance.yahoo.com/quote/%5EVIX/

[5] FRED Economic Data, Initial Claims (ICSA), https://fred.stlouisfed.org/series/ICSA

[6] Brave, S, 2011, National Financial Conditions Index (NFCI), YouTube: Federal Reserve Bank of Chicago

[7] PMITM by IHS Markit, Accurate and timely insight into the health of the global economy. https://ihsmarkit.com/products/pmi.html

[8] Huang, Melody & Rojas, Randall & Convery, Patrick. (2019). Forecasting stock market movements using Google Trend searches. Empirical Economics. 10.1007/s00181-019-01725-1.

[9] Google Trends, https://trends.google.com/trends/explore?date=today%203-m&q=oil%20price


Is six the ‘magic’ number? AQR’s extension of the Fama-French Five Factor Model to Six.


I didn’t realise until recently that Eugene Fama & Keneith French had extended their famous three factor model to five factors. They have added RMW, the return spread of the most profitable firms minus the least profitable stocks and  CMA, the return spread of firms that invest conservatively minus aggressively to the standard size spread SMB and value spread of HML. You can find a piece discussing it in a Forbes article.

Following on from the five factor model there is an interesting piece on the AQR website at the end of last year which considers extending the five factor model to six, by adding in UMD, momentum winners and losers. It’s is well worth a read.

Adding back in noise to model private market assets

Multi Asset Investing


A lot of effort is being done on how to incorporate private markets into a public market risk framework, as more and more institutions are trying to look at multi-asset risk holistically.

I recently came across the catchily titled ‘Fisher-Geltner Unsmoothing methodology’, which I have to confess I’d never heard of before. It appears to use first order auto-correlation to recover the underlying market values from a smoothed series. Initially it was applied to commercial property, though now people are using it range of illiquid assets, except Mezzanine debt which doesn’t seem to exhibit auto-correlation in the returns.

However the downside is you still need a time series history. This implies generally a listed benchmark or fund history, which limits it use in more bespoke, innovative  and direct investments, that usually don’t have a time series history. You could use a ‘similar asset’ database to template it, but the jury is out how representative…

View original post 7 more words

GIPS compares apples with apples, but does it stifle innovation?

Multi Asset Investing

I first came across GIPS a few years ago when I was at Gulf International Bank and have recently had need to revisit it. There are clearly many benefits to being GIPS compliant as it allows a like-for- like comparison of different managers, and removes uncertainty around the calculation methodology. For a fund/strategy to be GIPS compliant, the fund requires a minimum of five years of audited performance. Furthermore some people may say that it separates the ‘wheat form the chaff’, as in a Darwinian view of the world, it means that the fund has ‘survived’ for five to ten years. Clearly several UK pension fund consultants in particular use GIPS as part of their manager selection processes.

However from another perspective one could say that in a changing world, even five years is a long time and the head wind of of a five year record, stifles innovation. A growing…

View original post 146 more words

Has passive become passé in the Multi-Asset space?

Multi Asset Investing


More and more pension funds, sovereign wealth funds and endowments are looking to move away from having passive full replication of equities in their multi-asset portfolios to a variety of proxies either to enhance performance, lower cost, increase liquidity or a combination of all three. Amongst the alternatives people are looking at are:

Blended Smart Beta

Ive written about this on the quant blog. It’s becoming ever more popular as a diversifier, risk reducer and a potential source of alpha. A lot of attention has been drawn recently to blending different alternative indexation strategies, wether load weighted (such as risk parity or fundamental weighting) or via optimisation (as in minimum variance or maximum diversification). In addition there is the potential of netting rebalance trades to reduce turnover, compared to having each strategy in a separate portfolio.

Explicit risk premia/factor overlays

Smart Beta largely works through the manifestation of indirect…

View original post 203 more words

Smart Beta benchmarks compared to their real world implementation

Everybody seems to be looking at smart beta these days, particularly within the equity space as an alternative to pure passive. I’m not a great fan of the term smart beta, as if your definition is ‘a non free float cap weighted index’ then price weighted indices such as the Dow 30 and Nikkei 225 are smart beta. Alternative indexation is a better term, but a bit of a mouthful.

Given cap weighted indices by their very nature have a fairly low turnover, a big issue with smart beta is the difference between a smart beta benchmark as calculated by an index provider that everyone looks at and a real world portfolio implementation of that benchmark, which may have a very significant two way turnover when rebalanced. Take for example a minimum variance portfolio, which can have a very significant turnover. The problem becomes progressively worse as the investable universe becomes less liquid, as transaction costs increase and as more non major fx’s are involved. An example of this is the potentially significant difference between benchmark performance and the implementation drag of say an S&P 500 minimum variance portfolio and a minimum variance broad emerging markets portfolio.

Relative GTAA from a UK investors perspective

One of the many areas that fascinate me is relative strength multi-assets dynamic overlays. As a quant focused investor, my interest in GTAA was sparked by Mebane Faber ‘seminal’ SSRN paper . Following on from this, the guys at GestaltU, CSS Analytics and Michael Kapler at systematic investor have also added a lot, particularly in relation to portfolio construction within GTAA.

I have been interested for a while in the implementation of global tactical asset allocation (GTAA) from the perspective of a UK ETF investor. A lot of interesting work has been done in this space on long only global tactical asset allocation, but most of it has been focused on US based ETF’s.

When implementing such models as a UK based sterling investor, the liquid sterling denominated ETF’s available to a UK investor are different from the US. This becomes particularly pertinent when it comes to using FX unhedged low vol assets, such as bonds.

There are a few articles focusing on UK buy and hold strategies such as the ‘Lazy portfolios‘, outlined on the monevator website, but these are strategic, rather than tactical.

Below are the investment rules I use to construct the the UK GTAA portfolio. They are by no means ‘optimal’ either in terms of the asset mix or portfolio construction methodology.

The investment rules

1. Consider eight ‘groups’ of assets:

US equities
European equities
Asian+EM equities
Global sector equities
Bonds+inflation linked assets
Other illiquid assets

2. Each group contains various GBp denominated ETF’s. Within each group pick the asset with the highest sum of the six, three and one month return

3. If the highest sum for a group is negative or the vol of the top asset is very high (subjective, but say > 20), set the group to cash

4. Rank the highest picked assets from the groups, and go long the top four

5. Simple risk parity weight the portfolio (based on the inverse 3 month daily vol) and rebalance monthly

Portfolio Composition

Below is the portfolio composition on the 5th June struck as the end of April for a one million pound portfolio. I’ve added 1 share of Faber’s GTAA ETF (priced in sterling) and a Barclays balanced ETF purely for performance comparsion.



To monitor the strategy performance, I have used the Morningstar UK portfolio manager. Clearly it is over a month since the last rebalance. I will most probably rebalance it in the next few days.

The benchmark Index assumes 40% BarCap Sterling Agg & 60% MSCI Wld Free.



Disclosure: The views expressed in this blog are purely my personal opinion and do not in any way constitute investment advice. Furthermore I may hold positions in my personal portfolio in any of the ETF’s or stocks mentioned in this article.

Click here to sign up for email updates to my blog

Will the real Diversified Growth Fund please stand up

I was chatting with someone about diversified growth funds (DGF’s), which are very popular at the moment in the UK. They have been around for over a decade, with the most well known one probably being Standard Life’s GARS fund. This got me thinking about what actually defines a DGF.

Passive long only portfolios are generally defined as having a tracking error less than 25bps, enhanced indices are a bit more vague (say between 25bps and 50-100 bps), above that is active and hedge funds are generally trying to achieve an absolute return.

So far closest I have come too is :
‘A total return multi-asset product which is trying to achieve equity type returns, but at lower vol (generally targeting two third of equity vol or less).’

If anyone has a more accurate definition, let me know. For those of you who are interested in learning more about DFG’s, an useful overview is:

Aberdeen’s Independent guide on DGF’s for investors