Momentum based ETF rotation - Amazing backtested results - Need Reviews

NIFTY and Gold ETFs momentum investing

Recently there has been a lot of talk about Momentum based ETF rotation strategies. I have gone through some of the strategies and was awestruck by the amount of returns these strategies claimed to be making. One such example is this article:

This book also talks about momentum investing with backtest going back upto 30+ years.

Dual-Momentum-Investing By Gary Antonacci

I decided to try out a simple momentum strategy on the Indian markets. I don’t think we have globally diversified ETFs in india as yet but we do have NIFTYBEES and GOLDBEES. These two ETFs are not positively correlated so I decided to use these for my analysis.

The Strategy:

The strategy is really simple. Here are the rules:

  1. At the start of every week, check the 30-week rate of change of NIFTYBEES and GOLDBEES.
  2. If none of the ETFs are returning more than 3% in 6 months. Do nothing.
  3. If one or both ETFs are returning more than 3%, invest your entire position in that ETF.

That’s it. I don’t think it can get any simpler than that!

The Results:

I don’t have a lot of historical data to test this with as GOLDBEES was only introduced around 2007. Keeping that in mind, the period that I tested was from 2005 – 2015

First lets have a look at the returns generated by investing only in NIFTYBEES

  1. NIFTYBEES

Period 1/1/2005 – 21/7/2015
Initial Capital 100000.00
Ending Capital 424899.97
Net Profit 324899.97
Annual Return 14.73%
Max Drawdown -57.92 %

Although the returns have been decent at 14.73, the maximum drawdown is really scary at -57.92. If we have another sustained Bear market and the returns will suffer badly.

Lets have a look at the equity curve

As you can see, Equity declined more than 50% during the GFC in 2008. It took more than 4 years to make up for the losses and go back to the equity level before the GFC.

Now lets have a look at the returns made by the momentum rotation strategy

  1. NIFTYBEES + GOLDBEES Momentum

Period 1/1/2005 – 21/7/2015
Initial Capital 100000.00
Ending Capital 1067835.31
Net Profit 967835.31
Annual Return 25.22%
Max Drawdown 21.96 %

As you can see, the returns are amazing. It’s a no contest between buying and holding NIFTYBEES and Momentum rotation. Lets have a look at the equity curve.

To me, it looks like the only reason why this strategy has performed so much better than NIFTY is because it has the ability to resist severe drawdowns. The equity curve shows that there have been no real drawdown and the ride has been really consistent.

Lets overlay both the curves on one to understand why the strategy works.

This proves my previous point. The only reason this strategy has worked so well is because it has avoided the bear markets. In fact up until 2008, both the strategies were generating the same returns.

Lets have a look at the 6 month rate of change for Gold and NIFTY

This indicator plots the NIFTY and GOLD ROC on the same chart. The green and Red histogram marks NIFTYs Rate of change and the Gold and grey line marks GOLDs rate of change. The blue line is the fixed 3% return that we can get from our banks or from LIQUIDBEES. This shows that whenever the Equity market has declined, investors have flocked towards gold and it has seen significant advances.

BTW. According to the strategy, the current outlook shows that we should sit tight and invest our money in LIQUIDBEES or Banks as both NIFTY and GOLD have been returning less than 3% on a half yearly basis.

I am sure this strategy can be taken as a Macro indicator on the general market direction. E.g. Instead of investing in NIFTYBEES, we can invest in winning stocks based on our fundamental research and recent momentum.

Please let me know if there is something I have missed in my analysis. I shared this because the results look too good to be true and I want to know the point of view of the experts.

Here is the amibroker afl code required to backtest this strategy. Backtest this on a watchlist containing only NIFTYBEES and GOLDBEES symbols on a weekly timeframe.

SetOption(“InitialEquity”,100000);

EnableRotationalTrading();
SetOption(“WorstRankHeld”,1);
SetOption(“MaxOpenPositions”,1);

returns = ROC(C,30);

PositionSize = -100; // invest 100% of equity in single security
PositionScore = returns+1000;
PositionScore = IIf(returns > 3,returns + 1000,scoreNoRotate);

Disclosure : I am very new to investing and I am not following this strategy(yet).

23 Likes

Superb post. Have read this book, found it good. Nice to find someone actually simulating this strategy for Indian market.Kudos to you!!

Good stuff. Some immediate thoughts:

  1. Have you tried this with bear markets and bull markets exclusively as well? Also various time frames from a few months to a few years and everything in between
  2. What would the trading costs be?
  3. Instead of selling off nifbees can we buy put options? Would that give a better outcome

I have tried this with bear and bull markets exclusively. Here are the findings:

  1. Bull Markets - Strategy slightly underperforms the NIFBEES during the bull markets. The underperformance is because of whipsaws between gold and nifty during the moments of uncertainity in the markets.
  2. Bear Markets - Strategy outperforms NIFTY by a huge margin. In fact, this strategy remained profitable during the GFC and also during 2011-2012.

This is a long term strategy. Its unfair to test it out on shorter durations as the results will not be meaningful imo.

Trading costs are negligible. Backtesting shows a total of 29 trades in a 10 year duration. Also, as we are only trading ETFs, there is very little to no STT on the trades.

Taxes are real though. You will have to pay short term capital gain taxes on some of the trades. However, most of the big profits come as Long term capital gains so no taxes on those.

I dont have any knowledge of options so cant comment on the last question.

I am in the process of finalizing my long term investment plans now. Here is the strategy I will be trying to follow:

if(NIFTY ROC > 3 and NIFTY ROC > GOLD ROC)
{
Buy stocks with good fundamentals that are breaking out(ROC(20) > 30 with good volume) on the weekly charts
}

if(NIFTY ROC < 3 and GOLD ROC < 3)
{
Sell everything and move to LIQUIDBEES
}

if(GOLD ROC > 3 and GOLD ROC > NIFTY ROC)
{
Sell everything and invest in GOLD
}

I am pretty sure this strategy will outperform NIFTY during both Bull and Bear markets. It will make sure that we dont lose money during moments of extreme distress. I havent found out a way to backtest it in amibroker as it involves a bit of Rotational and a bit of normal trading.

4 Likes

Hi,

I did the back-test on excel from data obtained from Yahoo Finance for Nifty and Gold Bees from 2008. The algo applied was what you mentioned. Long on either of Gold/Nifty bees depending on which is maximum and outperforming 3% return over 210 days (30 weeks), else going long on liquid bees which will give 3%/30 return over a week. Please find attached the excel sheet.

1st sheet contains raw data. Analysis sheet contains analysis. Due to certain data not available for both indices on common dates, some weeks (36 weeks in total) have been excluded from analysis. My results for the same are 16.45% CAGR return over the period with 18.43% draw-down. Also yearly and monthly return is present on the next sheet. Please let me know in case there is anything I am missing.

Disclaimer: My 1st post here, so please excuse me in case of any mistakes.gold_nifty bees_v2.xls (503.5 KB)

1 Like

Hi,

Although I havent verified your spreadsheet. I think your analysis is correct as I am getting similar results when I run the backtest from 1/1/2008 - 22/7/2008.

Interestingly, NIFTY has returned just 4.6 during the same time period!

Nice work. You should look at Quandl if you want a richer source of data.

To put this in context, if you can find two negatively correlated quantities, then it becomes a case of asset allocation. I think it is more likely that the US Dollar and Gold are negatively correlated (when the world economy does bad, every one runs to physical assets). I think the relation between the Indian stock market and gold would be more a side-effect of what happened in US over the last decade.

While the data you ran through is a lot, it is just about one economic cycle - usually people prefer atleast 2 or 3, which usually means about 20 to 30 years of data. This is hard to get for most of us retail investors. So I would still be careful in drawing conclusions…

2 Likes

Yes My friend this shall work. in fact what needs to be done as proposed in the book is to put the overlay of trend following over and above relative strength.

I am sure you must be doing it on your individual capacity. Would love 2 collaborate on this endeavor as I have keen interest in this field.

I am interviewing Gary Antonacci for my podcast next week and will be posting it on our website www.stoicinvesting.com in the meanwhile I would request you to have a glimpse at my IIF presentation link below where I talked about Wesley gray’s 40 year back test on Alpha generated by momentum investing.
https://www.youtube.com/watch?v=ACdmkrW0Ns4

Incidentally we have also interviewed Wesley gray where he talked about this strategy in detail.

Cheers

Manish Dhawan.

I have done some work on this strategy. Let’s discuss this offline. Drop me a mail or tweet, will get in touch.

So I have been following this strategy for the last 1 year. Couldn’t have picked a worst time to start my investment journey :smile: .However, Happy to see that the the equity curve is as I was expecting for a bear market.

Interactive chart here:
https://plot.ly/~sharma.uditt/4911/portfolio-vs-nifty/

Since December 2015 I have been fully invested in liquidbees which has given me around 1.5% returns. That return is not visible on the chart so the flat line is actually a bit skewed towards the top in reality. Since March I have been 40% gold and 60% liquidbees.

3 Likes

Check my backtest.

remember this is relative strength only. Things wud improve manifolds we put a absolute overlay.

http://www.mysticwealth.in/buyhold-mf-returns/

1 Like

Hi,

Thats a nice article. However, I have not seen too much of a performance gain when I put an absolute overlay. Maybe I am doing something wrong.

The thing that has given me an immense gain(in backtests) is when I invest in stocks instead of NIFTYBEES. I am using the relative rsi as a macro indicator. When NIFTY is returning more than gold I invest in stocks satisfying the following criteria:

  1. CMP less than 400
  2. 30 week ROC more than 30
  3. CMP within 10% of its 52 week high.

I am sure adding fundamental filters to this criteria will make it even more profitable. Amibroker doesnt have good support for fundamentals so I haven’t been able to add those.

The backtests that I have done have given more than 30% CAGR during the last 16 years. I would be a very rich man if I can get half of that for such a long period.

Also, In real life, I was uncomfortable in allocating all my portfolio to just 1 asset(GOLDBEES). So when gold is beating NIFTY I invest around 40% in gold instead of the 100% as the strategy dictates.

1 Like

Hi Manish,

Nice article. But what has been the consistency ratios for the strategy. Apart from CAGR and drawdowns, what is the ratio of +ve & -ve months, quarters, years, and min and max return in a year. On seeing the graph for a 30% drawdown strategy, it still had negative years, which would be tuff to pursue when actually the drawdown hits the portfolio.

Ankush

I have been averaging 30+ in real from last 5 years. Negative is of course that I have missed the entire recovery rally recently but I m not complaining since I hv more than trippled my corpus. ;)

3 Likes

Hey guys,
nice thread, very interesting. Quick question, I was trying to backtest the strategy using COMEX gold prices (in $) vs Nifty bees and that does not seem to generate the same robust returns as using GoldBees. This made me wonder whether a large part of the gains of the gold part of the strategy are infact attributable to INR depreciating against the USD (and hence gold prices rising in INR terms). Supporting this is the time period of the strategy from 2007 to 2015 (as Gold bees was introduced in 2007). This was coincidentally the time when Rupee started depreciating against the USD. Over this 8 year period it fell from 44 levels to 66 levels ie. about 50% - creating an extra annualised alpha of about 6% whenever gold was held in the portfolio. It further makes me wonder whether the alpha of the strategy would contract if INR depreciation against the dollar were to stabilise of reduce

Just a thought! Would appreciate feedback

Cheers

Sharmaudi, Spot on My friend. U r onto something. This is word by word what Minervi prescribes in his amazing book. Please read it if u havent already.

With Strict stop loss and a bull run. this strategy can fetch serious money.

check out this guy https://weekendinvesting.wordpress.com
he is running a MI50 based on momentum strategy. I am doing the same on my portfolio but not as diversified as him and therefore big gyrations on both side.

Adding fundamentals help in reducing the need to over diversify.

give me a call sometime.

1 Like

Hi Manish can u share the MI50 momentum based strategy if possible

I don’t have it sir, a guy named Alok runs it. twitter handle @rohiniglobal

nice book by wesley gray, applying momentum on individual stocks and very nice review written by gary anntonacci of dual momentum on that. Please check it out. https://goo.gl/ipUKi6

He emphasizes that momentum works better on Index and asset classes rather than individual stocks.
And that is the reason I feel u need a thadka of fundamentals to generate serious alpha like minervi and william o neil did.

I am doing just that on my PF