Author Topic: Beating the S&P 500 Through Risk Parity Weighting  (Read 6306 times)

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1211
  • Age: 39
  • Location: Texas
Beating the S&P 500 Through Risk Parity Weighting
« on: June 12, 2015, 08:41:50 PM »
Was reading a book tonight and found a research paper and both gave returns to the S&P by simply weighting the index by risk parity. Equal weighting the S&P has historically outperformed the cap weighted index on a total return basis but not a risk-adjusted basis. Risk parity (equal risk if you will) outperformed both on a total and risk-adjusted basis. From 1995 to 2014 performance was:

Risk Parity
12.32% return
14.55% std deviation
0.85 reward/risk ratio

S&P cap weighted
9.64% return
15.34% std deviation
0.63 reward/risk ratio

Taken from this paper: http://dailyalts.com/utilizing-risk-parity-generate-better-equity-portfolio-outcomes/

Book I read tonight tests from '99 to 2015, risk parity portfolio is rebalanced monthly:

Risk Parity
13.1% return
-48.4% drawdown

S&P cap weighted
5.2% return
-55.3% drawdown

This is not practical for portfolios of less than $1 million IMO, and you'd have to really love investing to mess with 500 individual stocks, but interesting thought experiment nonetheless.

Would love to see a backtest going back 50+ years.
« Last Edit: June 12, 2015, 08:45:24 PM by hodedofome »

surfhb

  • Guest
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #1 on: June 13, 2015, 09:30:35 AM »
It may work out for you and it may not.   

Who knows? ;)

The Beacon

  • Stubble
  • **
  • Posts: 148
    • Financial Freedom Tips
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #2 on: June 13, 2015, 10:35:37 AM »
Historical performance is not indicative of future performance.  I am not saying back testing is not useful. It is if it is as extensive or rigorous as possible.  But it still does not tell you how likely(probability wise) you will have a similar return in the future.

For me, my main portfolio should be as passive as it can be.  So I can spend my time on things I truly like such as gawking at hot babes or day trading E-mini

Indexer

  • Handlebar Stache
  • *****
  • Posts: 1410
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #3 on: June 13, 2015, 01:07:33 PM »
It's active management. 

If it is so full proof why isn't there an actively managed fund doing it and just blowing everything away?  Well because they already are... but in practice risk parity doesn't work as well as it does in a white paper.  It is easy to backtest this.  The problem is trying to measure risk in advance.  Notice in their white paper they never actually explain how they do this.  They don't use beta, but other than that the formula is unknown.  The formula that worked perfect from 1995 to 2014 might not be the same risk parity formula that works best from 1970 to 1995 or from 2015 to 2030.

forummm

  • Walrus Stache
  • *******
  • Posts: 7396
  • Senior Mustachian
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #4 on: June 13, 2015, 01:19:20 PM »
Can you provide some more info about what exactly risk parity is? I either did not find or did not understand how they specifically composed their portfolio in the article. If I had a million bucks to dedicate to this strategy, what would it look like?

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1211
  • Age: 39
  • Location: Texas
Beating the S&P 500 Through Risk Parity Weighting
« Reply #5 on: June 13, 2015, 01:40:54 PM »
I'll check the book when I get home, but at least the book describes how it was implemented. I'm pretty sure they used the past 12 months std deviation to get the volatility measure, and then weighted each stock according to the inverse of it's volatility. I'll put up an example when I get home and have some free time.

Using std deviation as the risk measure for risk parity is pretty standard, nothing special about it.


Sent from my iPhone using Tapatalk

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1211
  • Age: 39
  • Location: Texas
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #6 on: June 13, 2015, 08:59:07 PM »
Bridgewater was the one that popularized risk parity, and they use it in their hedge fund which is the largest in the world. Here's their white paper on it https://drive.google.com/open?id=0BzyyTlvGE-T2WnRndjhnUGpwLW8&authuser=0

They use it in a multi-asset portfolio, so they'll take domestic and int'l stocks, domestic and int'l bonds, real estate and commodities. Let's just call it 4 asset classes to keep it simple. Instead of weighting the portfolio by market cap or by equal weighting or by the efficient frontier in modern portfolio theory, they weight the portfolio based on how volatile each asset class is. More volatile assets get a lower weighting, and less volatile assets get a higher weighting. What they want is for each asset class to get an equal risk weight to each other.

If I take VTSMX, VGTSX, VBMFX, VGSIX and DBC as stocks, bonds, REITs and commodities; and I weight them according to their volatility (std deviation) of the past month, this is what my portfolio would look like on Monday:

VTSMX - 21.07%
VGTSX - 16.10%
VBMFX - 36.70%
VGSIX - 15.47%
DBC - 10.66%

Bonds get most of the portfolio because they are the least volatile. Commodities get the least amount of the portfolio because they've been the most volatile the past month. Volatility changes all the time so you need to rebalance this on probably a quarterly or monthly basis. I've made a quick and dirty spreadsheet so you can calculate the weights yourself. Just put in the volatility for each fund and it'll give you how much you should have in the portfolio. It's really easy, just do 1/volatility. https://drive.google.com/file/d/0BzyyTlvGE-T2VzVUcUFvcmxmYTA/view?usp=sharing

Historically at least, the portfolio has performed about as good or a little bit less than a standard 60/40 or 100% stock portfolio on a total return basis. However, the volatility has been very low. If you brought the risk parity portfolio up to the same level of volatility as a 100% stock portfolio, it handily beat the market with (theoretically) the same amount of risk. They are obviously using leverage here. The 'risk' in risk parity, at least how it is implemented in the Risk Parity funds out there, is that most of the portfolio is held in bonds. So what happens if bonds start a 30 year bear market after a 30 year bull market? What would that do to the portfolio? Tough to say, and it's one of the main reasons I haven't jumped on the bandwagon for my personal portfolio.

In any case, I haven't seen someone post a backtest to an index fund/individual stock risk parity portfolio, and I thought the results were interesting.

To do this yourself, I would think a spreadsheet/google doc that automatically updates each day would be easiest. I don't know how to do that so, for a little more manual process, go to this link http://www.barchart.com/stocks/sp500.php?view=technical
Barchart at least gives you a nice list of the S&P 500 companies, and they are even nice enough to calculate the 20 day (1 month) volatility for you. Sign up for a free account...this might be a problem for you forummm... :)  and you can dump the list to a spreadsheet. Then you need to calculate the weights for each stock based on the inverse of their volatility. It's late and I'm doing this quickly, I don't know if the math is 100% correct but I'm pretty sure this is how you'd do it here: https://drive.google.com/file/d/0BzyyTlvGE-T2SEhtRkczV3FVN1k/view?usp=sharing

So then you'd have to go out and buy each stock in the index in the 'Portfolio Weight' column. Even with a $1 million portfolio, it looks like you'd only be buying 2 shares of NFLX and 1 share of PCLN, because the price of the shares are so expensive. Then you'd have to recalculate everything on a monthly or quarterly basis and adjust the holdings. Better have a low cost broker...

Perhaps this would still work well on the DJIA. That would be only 30 stocks and be a much more manageable universe.

Edit: here's a test of a more complex form of risk parity on the ASX50 index https://drive.google.com/file/d/0BzyyTlvGE-T2LUJmU3lqYVFvUjA/view?usp=sharing
Edit: and for the Bombay stock exchange http://www.tandfonline.com/doi/pdf/10.1080/23322039.2014.890060
Edit: I can't seem to find this pdf but the summary here says they tested it on the DJIA and the returns were almost 3x as much from 2002 to 2012 https://researchbank.rmit.edu.au/view/rmit:28057
« Last Edit: June 13, 2015, 09:34:40 PM by hodedofome »

innerscorecard

  • Pencil Stache
  • ****
  • Posts: 590
    • Inner Scorecard - Where financial independence, value investing and life meet
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #7 on: June 15, 2015, 02:28:47 AM »
Are there any low-volatility-anomaly ETFs out there?

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1211
  • Age: 39
  • Location: Texas
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #8 on: June 15, 2015, 07:24:33 AM »
From what I understand, low volatility ETFs only take the stocks that have low volatility. Risk parity weighting takes all the stocks but just weights then different.


Sent from my iPhone using Tapatalk

forummm

  • Walrus Stache
  • *******
  • Posts: 7396
  • Senior Mustachian
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #9 on: June 21, 2015, 11:20:03 AM »
Thanks for all this info, hodedofome. I've slowly worked thought it. I can't pretend to understand why it seems to have worked in the limited backtests in the articles. But there were several different sets of assets where it was claimed to have worked, although during similar time periods.

It feels like the transaction costs would be significant. I don't recall seeing any of the articles mentioning that or including it in their returns calculations. You are right that it would be a mess to implement as well.

I don't fully understand what it's doing. It seems like it would tend to push you more towards stocks with lower volatility--which tend to be mature dividend paying stocks. It seems like it would also tend to push you more towards stocks that have been rapidly appreciating recently--momentum stocks.

Is there any way to construct a backtest of this ourselves? You'd need a lot of historical data.

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1211
  • Age: 39
  • Location: Texas
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #10 on: June 21, 2015, 03:10:03 PM »
Risk parity is better understood as a concept, rather than a mathematically perfect investing strategy. I think it's beneficial for most DIY investors to get the concept, as you never know when it may make sense to use it. Making everything equal risk makes intuitive sense, you just have to do enough research to know the implications of that in a portfolio.

As far as what's possible for an individual investor, I'd bet a DJIA risk parity portfolio rebalanced quarterly could be done. You'd want to pay no more than $1-2 per trade. If you had price info for all stocks in the history of the DJIA you could backtest it yourself for many decades. I don't know know where that data is for free though.




Sent from my iPhone using Tapatalk

forummm

  • Walrus Stache
  • *******
  • Posts: 7396
  • Senior Mustachian
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #11 on: June 21, 2015, 04:05:16 PM »
Risk parity is better understood as a concept, rather than a mathematically perfect investing strategy. I think it's beneficial for most DIY investors to get the concept, as you never know when it may make sense to use it. Making everything equal risk makes intuitive sense, you just have to do enough research to know the implications of that in a portfolio.

As far as what's possible for an individual investor, I'd bet a DJIA risk parity portfolio rebalanced quarterly could be done. You'd want to pay no more than $1-2 per trade. If you had price info for all stocks in the history of the DJIA you could backtest it yourself for many decades. I don't know know where that data is for free though.

You'd want daily price history, right? That's why I was thinking it would be quite a lot of data to find somewhere.

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1211
  • Age: 39
  • Location: Texas
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #12 on: June 22, 2015, 05:21:07 AM »
Yes, you want daily prices in order to calculate the standard deviation.


Sent from my iPhone using Tapatalk

TomTX

  • Magnum Stache
  • ******
  • Posts: 3609
  • Location: Texas
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #13 on: June 22, 2015, 05:28:44 AM »
Risk parity is better understood as a concept, rather than a mathematically perfect investing strategy. I think it's beneficial for most DIY investors to get the concept, as you never know when it may make sense to use it. Making everything equal risk makes intuitive sense, you just have to do enough research to know the implications of that in a portfolio.

As far as what's possible for an individual investor, I'd bet a DJIA risk parity portfolio rebalanced quarterly could be done. You'd want to pay no more than $1-2 per trade. If you had price info for all stocks in the history of the DJIA you could backtest it yourself for many decades. I don't know know where that data is for free though.

You'd want daily price history, right? That's why I was thinking it would be quite a lot of data to find somewhere.

Presumably for each day you need both closing price and high/low to measure both daily volatility and longer term volatility.

I wonder if you need some kind of curve-matching if a stock consistently is going up....

Either way, I'm not gonna bother.

forummm

  • Walrus Stache
  • *******
  • Posts: 7396
  • Senior Mustachian
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #14 on: June 22, 2015, 08:32:25 AM »
This feels like one of those things that "worked" in certain places at certain times, but is super hard to replicate or test out-of-sample. And who knows if it will work going forward. From reading the articles, I'm not even sure I could replicate their findings on the first try even if they gave me the data.

dungoofed

  • Pencil Stache
  • ****
  • Posts: 661
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #15 on: June 23, 2015, 07:45:30 AM »
FWIW risk parity has been discussed at length among the Permanent Portfolio crowd. Indeed I think I read somewhere that Bridgewater's All Weather portfolios were based on the PP.

There are models for PP which attempt to reverse engineer All Weather portfolios or simply use risk parity in order to determine asset allocations (essentially decrease gold and stocks exposure, increase bond and cash exposure).

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1211
  • Age: 39
  • Location: Texas
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #16 on: June 23, 2015, 09:35:46 AM »
Agree that Dalio probably took inspiration for his asset allocation from the PP.

Risk Parity isn't intentionally trying to overweight bonds, it's just trying to make the volatility contribution equal for all the asset classes in the portfolio. If you only had long-duration bonds in your portfolio, the weights would be much more equal, as those are about as volatile as stocks.

NorCal

  • Pencil Stache
  • ****
  • Posts: 745
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #17 on: June 23, 2015, 09:55:33 AM »
It's active management. 

If it is so full proof why isn't there an actively managed fund doing it and just blowing everything away?  Well because they already are... but in practice risk parity doesn't work as well as it does in a white paper.  It is easy to backtest this.  The problem is trying to measure risk in advance.  Notice in their white paper they never actually explain how they do this.  They don't use beta, but other than that the formula is unknown.  The formula that worked perfect from 1995 to 2014 might not be the same risk parity formula that works best from 1970 to 1995 or from 2015 to 2030.

No, this is NOT active management.  I'm not sure where this idea came from that anything other than a cap-weighted fund is active management.  This is a blatantly false statement.

Passive management is the idea of not picking winners and losers.  That is all.  No more, no less.

How you determine which stocks go in a passive portfolio, and how you weight those stocks are the most important decisions a passive manager can make.

Modern Portfolio Theory and most related finance literature actually agrees that cap-weighting is sub-optimal for both overall return purposes and Sharpe-ratio measurements.  However, cap-weighting is the most tax efficient and easiest to manage, which brings down fees.


hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1211
  • Age: 39
  • Location: Texas
Re: Beating the S&P 500 Through Risk Parity Weighting
« Reply #18 on: June 24, 2015, 08:14:47 AM »
Buying 50 random S&P 500 stocks each month and weighting them by risk parity http://www.followingthetrend.com/2015/06/a-random-ass-kicking-of-wall-street/