Risk Reward Ratios

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2071
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Risk Reward Ratios

Post by AFJ Garner »

In trend following systems designed to hook on and hang on to long secular trends, I prefer not to pay too much attention to standard deviation or semi deviation of returns when it comes to deciding between taking profits or letting full position size run to the bitter end. And in my systems and my portfolios, although risk adjusted return (using these measures) is almost inevitably better with profit taking, I prefer raw CAGR.

If I recall correctly, Sluggo reported better CAGR as well as most other tests of "goodness" using profit taking on some systems/portfolios - but I have not found this to be GENERALLY the case in my own testing in longer term systems. A usual I may well be missing any number of points.

I have however noticed excellent results (using convential risk metrics) from shorter term trend following systems using low R Multiple profit taking exits on a big percentage of full position size. But I have much testing still to do in that area.

Herewith an interesting article picked up from www.hedgeworld.com which I set out in full.


"Online hedge fund databases grow as hedge funds proliferate, and investors use these databases to screen funds. But some of the data lead investors to consider funds they should reject and reject funds they should consider. Providers aren't fixing the problem and investors aren't calling them to task. It's the risk elephant in the hedge fund living room.
Let's take a case. Consider two hedge funds with the same positive but low mean return for say the past three years. Monthly returns of one are symmetrically grouped tightly around the mean; monthly returns of the other look like a fat positive head with a thin negative tail. Existing risk data (Sharpe ratio, Sortino ratio, VaR) penalize the second fund by treating positive variance from the mean as undesirable (Sharpe, VaR) or unimportant (Sortino). The result is that an investor seeking a manager who pursues opportunity while controlling risk (as opposed to passively protecting capital) will be misdirected away from the second fund and toward the first fund.

Cases like this occur because databases use statistics that are widely accepted yet deeply flawed in two elementary ways:

• Risk is misrepresented. Adjusting return with standard deviation or downside deviation is an artifice that introduces more confusion than refinement, and is unreliable due to the assumption of normality.

• Opportunity is not represented. When monthly gains are used at all, they are treated like monthly losses. The statistics that misrepresent risk also ignore the reality that opportunity and risk are opposite potential outcomes of the same strategy.

The false premise of risk without opportunity underlies current performance assessment. Opportunity and risk are distinct but inseparable in portfolio management, as in any capital investment decision, and attempts to quantify risk while leaving opportunity out of the equation don't deal with reality. Why debate the validity of a risk statistic that derives from treating a loss no differently than a gain? Why ponder an appropriate time frame and confidence level for risk while ignoring opportunity as the reason for risk? Worse yet, what is the meaning of any performance expectation based on a one-sided version of performance history? Quantifying opportunity and risk together in an intuitive way would be more informative to investors than imaginary risk statistics or artificial return adjustments. Or will investors be left to continue screening hedge funds with measures that are short on reality?

More to the point, exploiting market opportunity while avoiding market risk is the raison d'être of any genuine hedge fund, in which the portfolio manager works to achieve a steady dominance of opportunity over risk not obtainable by betting on market direction. This is a game that is fundamentally different from conventional securities investment and one that cannot be scored in any meaningful way with conventional risk measures. It's time to quantify and report the hedge fund manager's skill at playing the game by supplementing return data with a score for the opportunity/risk goal. Database providers should act responsibly to accomplish this end, and it would be easy to do so because a simple remedy is at hand.

The distribution of monthly returns offers a straightforward, meaningful number for how strongly gains dominate losses in a portfolio. The sum of positive monthly returns divided by the absolute value of the sum of negative monthly returns is an intuitive measure of the opportunity/risk balance achieved by the manager. This ratio is also the base case of the Omega Function, as described in 2002 by Keating and Shadwick. It can be termed the Quality Ratio and used as a fundamental factor in performance comparisons. It should replace flawed statistics in reporting by any hedge fund database provider interested in giving investors better numbers to screen hedge fund managers.

So why do we tolerate and even embrace existing risk statistics? Why does Columbia play Cornell? It's tradition. Existing measures have the weight of decades of use behind them. Providers are all geared to generate them and investors rely on due diligence to counteract their shortcomings. Most managers accept them as standards, are not innovative in the subject, and rely on direct marketing. Current database reporting is further rooted in common language (risk/reward, risk-adjusted return) developed by respected authors of books and articles in the mainstream of investment discourse. A key stumbling block is that the large body of literature on risk assessment and risk management largely ignores the concept of opportunity as a factor to be quantified in conjunction with risk.

The proposal here is disruptive but simple: Replace all hedge fund risk statistics and return adjustments by a measure that reflects the manager profile for pursuing opportunity while controlling risk. Such an outcome would greatly reduce confusion in performance assessment and screening of hedge funds, by discarding several misguided measures and focusing on one that is much closer to reality. That kind of constructive retooling won't be accomplished unless the industry engages in some serious dialogue.

The foregoing is only about measuring portfolio performance, not about managing it, but there should not be a disconnect among objectives, measures and management. Most hedge funds state dual objectives to preserve capital and achieve excess returns, and all portfolio managers must balance pursuit of opportunity against control of risk. A gain/loss ratio of monthly returns is a measure that reflects how well opportunity/risk prospects have been managed to satisfy the capital preservation objective. In screening hedge funds, such a measure should take its place alongside return. It's the mouse that spooks the elephant.

Chuck Hage is compliance officer for Mohican Financial Management LLC, Cooperstown, N.Y., a convertible arbitrage hedge fund established in 2003. "
sluggo
Roundtable Knight
Roundtable Knight
Posts: 2987
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo »

In the computer industry there's a saying: The nice thing about standards is, there's so many of them!

The reason why there are so many ways to measure money manager performance, is because everybody likes their own way of measurement and nobody likes anybody else's way.

I recommend a three step process:
  1. Figure out what is truly important ("mission critical") to your managed money clients
  2. Figure out a performance measurement that emphasizes what is truly important
  3. Arrange your trading to optimize that performance measurement
Many people think "maximum drawdown" is what is truly important ... because that's what causes clients to withdraw their funds. So they use performance measurements that emphasize maximum drawdown.

Others think "lack of correlation with benchmark B" is what is truly important. Notice that benchmark B might be "the S&P 500" or "the Mount Lucas Trendfollowers Index" or anything else your heart may desire.

Others think "absence of pain" is what is truly important. So they use performance measurements such as "% of days not in a drawdown" (or "average length of drawdown" or "length of longest drawdown") as their measurement.

Just because some pompous pontificator wrote an article telling you what to do, doesn't mean you have to pay attention.
Old European
Roundtable Knight
Roundtable Knight
Posts: 144
Joined: Mon Apr 19, 2004 3:04 pm
Location: Old Europe

Post by Old European »

Indeed
kianti
Roundtable Knight
Roundtable Knight
Posts: 335
Joined: Fri May 02, 2003 6:10 am
Location: Florence - Italy

Post by kianti »

In the late 1920s Warner Heisenberg stunned the scientific world with his uncertainty principle of quantum mechanics. He showed that you can look closer and see less.

Adding a little bit of Behavioral Finance you could notice how utility function or initial state of wealth modifies judgement of numbers and perception of risk.

A fund manager could invent a new ratio to show how good he was in the past, a finance researcher could come up with a new ratio to show how clever he is and get nice a job from a bank, a portfolio manager could invest in the best ratios and in case things go wrong blame some unexpected event..... I could start the month with a maximum drawdown target of 1% and the very next day think that I should risk at least 10%.



------------------------------------------------------------------
"So far as the law of mathematics refers to reality, they are not certain. And so far as they are certain, they do not refer to reality".
ALBERT EINSTEIN
GEOMETRY AND EXPERIENCE
AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2071
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner »

I suspect that one does have to listen to the pontificators or at least to the general drift that the pontificating is taking.

Fashions take hold in the fund management industry as much as in any other. And if you are in the thankless business of getting blood out of a stone by raising institutional interest in trend following, then buzzwords and the latest ratios might be just the ticket.
Jason
Roundtable Fellow
Roundtable Fellow
Posts: 57
Joined: Sat Apr 26, 2003 10:31 pm
Location: Oakville, Canada
Contact:

Post by Jason »

I believe that the optimum trading model and the optimal business model are two different things.

In my opinion, those focused on the optimum trading model are going to be focused on maxing CAGR while (possibly) minimizing the risk of elimination. As I am buliding a business, I aim to strike a balance between running a fund in a way that keeps me engaged and listening to what clients want. Specifically, in this space of mechanical trading, it seems that clients are looking for positive months and what losing months you have should be in the single digits. Naturally, they also look at all the other standard risk adjusted measures but I am surprised at how closely each individual month is scrutinzed.
Christian Smart
Roundtable Fellow
Roundtable Fellow
Posts: 50
Joined: Fri Apr 18, 2003 8:53 pm
Location: Huntsville, AL
Contact:

Post by Christian Smart »

Hedge fund managers have to please their clients, the majority of whom seem to prefer a steady stream of returns rather than absolute returns. Look at Toby Crabel's performance - great in risk-adjusted metrics, not so great in absolute returns. The net result is that his is one of the bigger hedge funds around. In the book HedgeHogging, Barton Biggs discusses a client presentation where someone points out that most hedge fund investors would pall at a 10% down month. This seems different to the interests of most individual speculators, many of whom are aiming for absolute returns rather than great risk-adjusted returns.
Post Reply