Tag: quant

Equity Returns at the Turn of the Month

The Turn of the Month Effect

A recent paper in the Financial Analysts Journal looks at the Turn of the Month effect on equities. Equity Returns at the Turn of the Month, John J. McConnell and Wei Xu:

The turn-of-the-month effect in U.S. equities is found to be so powerful in the 1926–2005 period that, on average, investors received no reward for bearing market risk except at turns of the month. The effect is not confined to small-capitalization or low-price stocks, to calendar year-ends or quarter ends, or to the United States: This study finds that it occurs in 31 of the 35 countries examined. Furthermore, it is not caused by month-end buying pressure as measured by trading volume or net flows to equity funds. This persistent peculiarity in returns remains a puzzle in search of an answer.

Does it apply to Indian markets?

The study skips over the Indian markets. So we did a quick test on the CNX 100 index to check if the effect holds. Here’s the cumulative return chart between a Buy-and-Hold CNX 100 strategy (B&H, black) and a Turn-of-the-Month CNX 100 strategy (TOM, red):

CNX100.TOM

Although the TOM strategy has lower-drawdowns, the B&H wins – both in terms of tax advantage and trading costs. The Turn-of-the-Month effect doesn’t seem to apply to Indian equities.

Broadcast Dates on NSE

This is more of a “note to self” post. I was trying to do some earnings event analysis and got tripped by this. If you look at NSE’s quarterly earnings broadcast, the ‘broadcast date/time’ they provide lags the time at which earnings are actually announced.

If you take a look at DCBBANK, for example, this is what NSE says:

dcb broadcast

However, results were announced the previous day, during market hours:

dcb announce

So if you are looking to match up market action with announcements, then NSE’s broadcast time-stamp is no good.

Mutual Fund Alpha Charts

Charting Alpha

Most mutual fund investors chase recent performance. However, experience shows that alpha, or out-performance, is rarely sustained in the fund universe. To visualize how alpha fluctuates across different time-periods, we extended the relative strength spread notion of stocks to mutual funds. By normalizing performance across multiple funds vs. a single benchmark, the CNX 500 index, we can get a sense for how stable the alpha is.

Exhibits

Have a look at the RS-Spread chart of the HDFC Growth Fund:
hdfc.growth.mf.relstrength

Notice how 1-year alpha was below zero between Oct’2012 and May’2014 and is now back below zero again. This should indicate that whatever strategy the fund is employing is not that great in generating sustainable alpha. Now compare that to the Birla Sun Life Frontline Equity Fund:
birla.frontline.mf.relstrength

Notice how the fund has managed to outperform over the last 5-years. Here’s the MNC fund’s RS-Spread chart:

Birla.MNC.mf.relstrength

The FundCompare Tool

We update these charts daily for more than 100 funds. You can access them through our FundCompare tool. If you have any questions, give us a call or Whatsapp us!

Relative Strength Spread Charts

What is Relative Strength Spread?

Relative Strength Spread of a stock is the relative cumulative performance of the stock vs. the CNX 500 index over the same period.

We had discussed how the Relative Strength Spread Index is a coincident indicator of momentum earlier. With a new update, we are publishing charts of relative strengths of individual stocks over multiple lookback periods. This can be found under the ‘Quant’ tab of the equity pages.

How to interpret Relative Strength Spread?

Relative Strength Spread is a measure of historical out-performance. By observing the RS-Spread over different lookback periods, investors can get a sense of future direction of momentum. For example, if you look at the RS-Spread of TATAMOTORS, you can observe how the stock lagged the broad market for the better part of the year before suddenly turning around in the last few days.

TATAMOTORS.relstrength

Now contrast that to BHARATFORG:

BHARATFORG.relstrength

Unlike a price or return chart that looks at an individual stock, the RS-Spread chart incorporates information from a broad set of stocks and compares it to the market average. It has more information content than the ‘Relative Momentum’ chart as well.

ARMA + GARCH to Predict VIX

GARCH(1,1)

GARCH(1,1) is a common approach for modeling volatility. They were developed by Robert Engle to deal with the problem of auto-correlated residuals (which occurs when you have volatility clustering, for example) in time-series regression.

What we did:

  1. Picked the best fit ARIMA(p,d,q) model for historical VIX over different look back periods
  2. Created a GARCH(1,1) model based on ARMA(p,q)
  3. Predicted t+1 VIX

500-day lookback

We found that modeling based on the previous 500-day VIX closing levels gave us the least prediction errors. The appendix has the charts for other lookback periods.

Prediction vs. Actual

VIX.prediction.500

Note how in some periods, the predicted value (red) is just the previous value.

Prediction error

VIX.prediction.pctError.500

Values less than zero implies that the model prediction overshoots the actual VIX level the next day.

Prediction vs. Actual Density Plot

VIX.prediction.density.500

The model bias towards higher estimation of VIX is made explicit here.

Next steps

We will integrate this model to our morning ‘Options Daily’ posts so that we get an idea of both the current state of VIX and the expected modeled behavior.

Caveats:

  1. The 500-day lookback is purely empirical. Maybe some other look-back period that we have not tested would have been ideal to model. We will never know.
  2. Only the known history can be modeled. The outputs should be used along with market determined proxies of expected volatility.
  3. There is always a probability distribution around a predicted value. We will publish this in our daily posts.

Appendix

VIX.pacf

VIX.acf

VIX Model vs. Actual across various lookback periods. (pdf)

quant.stackexchange

Volatility Forecasting I: GARCH Models, Rob Reider (pdf)