Tag: quant

Does Momentum Trend or Mean-Revert?

Outline

We consider long-only momentum returns to be composed of market returns plus excess returns.

If excess returns over a specified period, n-days, (say, 5- or 10-days) either trends or mean-reverts, then that can be used to trade the momentum portfolio.

One way to check if a time-series is trending is to calculate the Hurst exponent (H) over a rolling window (say, 5-years) of n-day excess returns . If H < 0.5, then the time-series is mean-reverting; if H > 0.5, then it is trending, else it is random.

A simple strategy would be, for H < 0.5 (mean-reverting), if excess returns is greater than its median, then exit or if excess returns is less than its median, then enter.

The problem boils down to specifying the excess-return calculation periods (n-days) and the Hurst exponent rolling windows so that it makes sense (avoid data-mining.)

Setup

We use the Barclays Euro-zone, UK, Japan and US momentum index data-sets to run our experiment. Since they provide both an excess-return index and a total-return index, we can use the former to time entries and exits in the latter.

We ran for two n-day configurations: 5 and 10. We set the Hurst (and median) rolling window at 5-years.

We expected to find H to be either consistently above or below 0.5. My personal expectation was that H would be above 0.5 (trend.)

Results

Using the Hurst exponent did not improve momentum returns. In both 5-day and 10-day configurations, a strategy that went long if n-day returns were less than their median out-performed those that incorporated H.

The back-test using 5-day returns mostly worked on Euro-zone and US momentum indices. So we are skeptical that this approach can be generalized and it will likely fall prey to data-mining.

The back-test using 10-day returns saw buy-and-hold emerge a consistent winner.

The 5-day H toggled between trending and mean-reverting but spent most of its time trending. The 10-day H was consistently trending. For the specifications that we tested, momentum excess returns trends.

Code and Charts

The R code is here: 5-day and 10-day config. You can login to pluto and play around with the lookback and statWindow variables to see how H, median and back-test results change.

Questions? Ask them over at our slack workspace.

Categories Tags ,

pluto: Your Research Velocity

pluto is our compute cloud for exploratory financial data analysis (intro). We built it, in part, to scratch our own itch and to offer an intuitive platform for financial market research that abstracts away most of the drudge work involved in data acquisition, storage and maintenance. The end goal is the increase the speed at which reproducible and shareable research occurs. Here’s a recent example.

VIX-Adjusted Momentum

On 10:13 AM · Jul 11, 2019, Darren (@ReformedTrader) tweeted out a link to CSSA that discussed a momentum strategy on the S&P 500 index. It divided the daily returns of the index by the day’s VIX – a poor man’s volatility adjustment, if you will. The back-test result was interesting and we wanted to reproduce it.

We started work on it at 2 PM. Using pluto’s Indices data-set, we could quickly setup the code and reproduce the results within the hour. See the github history of vix-adjusted-momentum-US.R notebook if you don’t believe it.

Next: if it worked for S&P 500, could it work for NIFTY 50? We fired up pluto again at 5:15 PM and quickly ran the strategy for different look-back periods (vix-adjusted-momentum-INDIA.R.md) before concluding that it doesn’t. Time taken: 15 minutes.

A quick glance at the annual return chart of the S&P 500 back-test showed that the out-performance occurs in periods of persistent high-volatility, like in 2008. But regimes change and signals fade. If you removed 2008 from the back-test, the strategy’s overall out-performance degrades considerably.

Next: does it beat a simple SMA system? vix-adjusted-momentum-and-SMA-INDIA.R answers that question in 20 minutes.

So basically, within an hour, anybody who had a passing interest in systems trading/investing could reproduce a strategy, check for applicability and extend it.

We will continue to add more data-sets to pluto and make it easier to use so that you can increase your research velocity.

Categories , Tags

Quant Model in Mutual Fund Wrapper

Most quant/smart-beta model based portfolios in India are built on direct-equity platforms – PMS, RIA, Themes and DIY. Their first major drawback is the 15% capital gains tax that needs to be paid the piper every year. The second one is the ability to track the “all-in” cost of maintaining the portfolio. This is where mutual funds have an advantage. Their pass-through status means that they don’t have to pay capital gains tax on portfolio sales and the end-of-day NAV gives investors the fully baked-in value of their portfolio. That said, mutual funds that wrap quantitative models have been few and far between. A new one has entered the fray: the DSP Quant Fund.

They were gracious enough to share their backtest. What follows is a 30,000 foot analysis.

Cumulative performance looks vs. a broad-market cap index looks good

cumulative performance vs. NIFTY 100 TR

However, excess returns seem to be tapering off…

excess returns over NIFTY 100 TR

Value factor seems to be a drag

If you regress the Quant Fund against the market-cap index and NIFTY strategy indices representing quality and value, you can see that returns have been primarily driven by the market (beta) and quality. Value seems to contribute negatively to overall returns. Part of the diminishing excess returns could be explained by the increasing influence of market beta to the fund’s returns.

drivers of returns

Why not just buy the NIFTY 200 Quality 30 Index Fund/ETF?

cumulative performance vs. NIFTY 200 Quality 30 TR

The SBI Quality ETF that tracks the NIFTY 200 Quality 30 Index has an expense ratio of 50bps. So while comparing the index against the Quant Fund, we need to haircut the index performance by that amount. Also, the Quant Fund comes out at 40bps for direct investors. The former is an ETF with minimal liquidity whereas the latter is an open-ended fund that can be redeemed at NAV – matters when you want to exit.

Qualitatively speaking…

DSP’s Quant Fund is a low-cost alternative to investors who want something more than market beta but not a full-fledged actively managed fund. It is tax efficient compared to other direct-equity platform solutions that over-weight the quality factor. And it is of comparable cost to most other quant/smart-beta funds/etfs for direct investors. Passive investors should definitely give it a strong look.

Code and charts are on github.

Categories Tags , ,

No Silver Bullets

Most of the time, beta swamps alpha. Take the case of the Roubini Country Insights model, for example. It claims to “rank countries based on an analysis of over 2500 data points from institutions such as the Bank of International Settlements and the World Bank.” Also, “these data points cover each country’s demographics, quality of education, healthcare and ability to innovate, and will look at the country’s growth potential in political and social spheres, as well as its top-down macro-economic situation.” It sounds like it does everything that a smart investment manager with a long-only global equities mandate should be doing. And you would expect such a smart model to add significant alpha.

Thanks to Barclays, a bunch of equity indices based on this model have been available for a while now. We were curious as to how these performed vs. their corresponding plain-vanilla market-cap weighted cousins.

market-cap vs. maro-quant

Developed markets: MSCI World (black) vs. Insights (green)
Emerging markets: MSCI EM (red) vs. Insights (blue)

The value add from the smart-beta quantitative “Insights” model, roughly about 1% a year, seems skinny compared to all the work that must have gone into it. 2500 data points is a big dataset but it looks like most of them have no effect on equity returns. This also ties into the curse of dimensionality when dealing with complex adaptive systems – more data typically subtracts from the model.

As an investor, it probably would have been easier to stay invested in one of the cap-weighted indices, just accepting the beta, rather than reach for that 1% extra with fancy sounding strategies.

Categories Tags