Forums  > Trading  > How high a Sharpe is considered "good?"  
     
Page 3 of 4Goto to page: 1, 2, [3], 4 Prev Next
Display using:  

LowDD


Total Posts: 142
Joined: Feb 2009
 
Posted: 2010-01-05 17:52

@ Lion-O, mib -- Hmm... I disagree with both of you.  I find your comments trivial and cumbersome and frustrating.  MIB do you find that Millennium is 1.99 SR?  Are you running gross or net returns?

I am looking at a simple hedge fund platform investment report that I get.  I bet it has the highest fee version of both in it, which for the last 5 years is probably 5/25.  Millennium is (net of all fees) 15.18% return, 4.95% volatility since 1997.

You guys seem really insightful.  Please compute the SR on that for me.


NeroTulip


Total Posts: 997
Joined: May 2004
 
Posted: 2010-01-05 17:52

Millenium is about 15% return and 5% vol, and has a 12y track record.

Moore is 20% return and 12.5% vol, 20y track record.

I don't have the data for RTC and SAC, does anyone have it? Not sure what the risk free rates were for these periods too.

Edit: cross posts.


Inflatable trader

FDAXHunter
Founding Member

Total Posts: 8353
Joined: Mar 2004
 
Posted: 2010-01-05 17:54
LowDD:RTC, SAC, Millennium, Moore.

Moore has 1.3 since inception (19 years).




The Figs Protocol.

bluelou


Total Posts: 68
Joined: Jan 2009
 
Posted: 2010-01-05 17:59
LowDD,
Recall that Sharpe varies with the time frame used. When investors are referring to fund performance stats they're usually using monthly data series (for both net RoR and RfR). I don't have the data sets you're working with but there's a good chance that this is what explains the difference in Sharpe ratios being thrown around here.

Lou

Je suis ce que je suis, et c'est tout ce que je suis -Popeye

LowDD


Total Posts: 142
Joined: Feb 2009
 
Posted: 2010-01-05 18:07

FDAX -- Consider gross of fees?  I have seen Moore successfully market at 5/25 (and 3-yr lock, annual thereafter) the last 5 years.  I would not be surprised if they have been there for 10-yrs.  So if you gross them up one might estimate the SR becomes 1.8-2.2. 

I think it makes sense to consider gross returns.  Obviously as someone succeeds, they raise fees.  We arent trying to judge post-fee returns but the ability to produce the performance.  The fat that high SR managers extract larger and larger premiums, thus delivering somewhat lower SR to investor is just natural.  Especially given the herd mentality of institutional investors.

Another thing on Moore.  Its probably 20-vol prior to 1998 and 8-vol since.  So its a 1 SR fund that got big and evolved into an institution that has been 2.5 SR since 1999.  Just estimates, but makes sense.


mib


Total Posts: 354
Joined: Aug 2004
 
Posted: 2010-01-05 18:32
AFAIK Sharpe ratio is defined based on _excess_ returns - you can check on Wikipedia if you do not believe me! ;) Thus net return and vol are not sufficient to calculate it. Not sure why someone not aware of that would want to discuss Sharpe Ratios here.

With cash returns averaging about 5% over the period this subtracts 1 from Millenium SR. I think there are some other, less significant differences in our calculations. I can believe in a reasonable calculation putting Sharpe at a touch over 2. You can probably call this "2-4 range". But only when marketing to a very well entertained client.

Head of Mortality Management, Capital Structure Demolition LLC

LowDD


Total Posts: 142
Joined: Feb 2009
 
Posted: 2010-01-05 18:53

MIB: Again.  Really?! So whats your number?

And whats your RFR?  I show bill rates below 2% for 4 full years of the last 10 years (actually 0 for the last year).  I show them above 5% for only a brief 1-yr period in the last 10 years.  I might conservative handicap RFR at 3.5%.  Gives Millennium a SR of 2.3 (net of fees).

You said: MIB - "I have data for Millenium at hand - they are under 2 SR over 10 and 12 years".

 


hedgeQuant


Total Posts: 233
Joined: Dec 2006
 
Posted: 2010-01-05 23:52

AFAIK Sharpe ratio is defined based on _excess_ returns

Conventionally, yes. A lot depends on what you are selling and to whom. If you are selling to prop desks / managed accounts they usually want to know the SR based on  daily gross returns (without risk free - rate). Well usually you end up sending the dailies and they typically calculate the financing charges (usually 50 - 75 Basis points / year) and do the SR computation on dailies net of financing charges. (Fees: that is a "negotiating point").

If you are selling a fund I would argue that excess returns (in SR computation) make sense if the risk free return is set as a hurdle rate.

To be honest I am not entirely convinced by the use of risk-free rate in SR calculations. Sharpe Ratio itself makes sense if returns are normally distributed and are independent across months/days. When the risk free rate is used that begs the question that even when the returns are approximately normal whether the risk free rate follows that profile. Granted it does not change from day to day, but probably with each FOMC meeting.

More importantly is the risk free rate a good benchmark? Given the risk profile of most of the hedgefunds I would say the BBB rate is a better benchmark.

 

 


Tradenator


Total Posts: 1585
Joined: Sep 2006
 
Posted: 2010-01-06 00:20

LowDD, I think you are being way too harsh on the Nobel Prize winners.  Perhaps you really meant the prize from some bank in honor of...


pj


Total Posts: 3342
Joined: Jun 2004
 
Posted: 2010-01-06 10:05
Do you have this 2009 prize in Mathematics in mind?

вакансия "Программист Психологической службы" -але! у нас ошибко! не работает бля-бля-бля -вы хотите об этом поговорить?

FDAXHunter
Founding Member

Total Posts: 8353
Joined: Mar 2004
 
Posted: 2010-01-06 10:33
hedgeQuant: More importantly is the risk free rate a good benchmark? Given the risk profile of most of the hedgefunds I would say the BBB rate is a better benchmark.

What you're referring to is the Information Ratio.

The rate used in the Sharpe Ratio is not a benchmark. It takes away what you could have gotten with no volatility and then adjusts your excess return by the volatility of just that.

If you took some other rate, that rate itself would have volatility. So then what volatility do you use to adjust (divide)?

The Figs Protocol.

mib


Total Posts: 354
Joined: Aug 2004
 
Posted: 2010-01-06 12:18
Uups, my bad. I used the wrong risk-free rate (GBP LIBOR). I agree that Millenium has Sharpe ratio safely above 2. Sorry

Head of Mortality Management, Capital Structure Demolition LLC

hedgeQuant


Total Posts: 233
Joined: Dec 2006
 
Posted: 2010-01-06 16:17

The rate used in the Sharpe Ratio is not a benchmark. It takes away what you could have gotten with no volatility and then adjusts your excess return by the volatility of just that.

Thanks for the clarification FDAX. I was always using SR as a metric for judging potential drawdowns/ down years (ergo did not make sense to use risk free-rate). But what you said makes perfect sense.


nodoodahs


Total Posts: 227
Joined: Sep 2007
 
Posted: 2010-01-07 02:39
Actually the risk-free rate HAS volatility. From the 1960s to today, the RF has averaged 5.5% annually with about 2.6% standard deviation (calculated using 3-month T yields from stlouisfed.org). Not that that level matters much ...

Also interesting to note that, in addition to varying for other reasons, the long-term returns of U.S. stock indices and a 10-year Treasury ladder vary with the risk-free rate environment, both of them tending to earn less when the interest rate environment is high - meaning that the Sharpe ratios of the asset classes themselves are influenced by the RF which is subtracted from their returns to develop the Sharpe ratio!

RF level probably also influences returns from managed futures, but I haven't documented that yet ...

I haven’t seen a beatin’ like that since somebody stuck a banana in my pants and turned a monkey loose.

SimJimons


Total Posts: 198
Joined: Aug 2007
 
Posted: 2010-01-07 09:46

nodoodahs, what do you mean with "RF level probably also influences returns from managed futures..."? Why probably?

/Sim


Sounds great...keep me out!

nodoodahs


Total Posts: 227
Joined: Sep 2007
 
Posted: 2010-01-07 21:29

The roll yield is a product of backwardation/contango, convenience yield, interest paid, storage costs, dividends paid, and the rate at which the contract is borrowed, which is the risk-free rate. So to me, it stands to reason that the roll yield would be influenced by the RF level, and thus the excess returns of systematic trading strategies would also be influenced.

Now, I haven’t personally verified that with a strategy backtest or seen that in the literature, but to me it stands to reason.


I haven’t seen a beatin’ like that since somebody stuck a banana in my pants and turned a monkey loose.

svquant


Total Posts: 113
Joined: Apr 2007
 
Posted: 2010-01-08 18:22
The Rf does "influence the returns on managed futures" and the amount of return due to tbills in various commodity index products and CTA index has been well documented in literature. In the past the return on margin deposits has been of the same magnitude as roll-yield and trend returns looking at trend following index type products.

Now in recent times this has not been true with short term tbills paying ~0%. On the other hand the roll-yield has changed from its past historical patterns due to supply, demand, and assets under management in passive long only commodity funds.

As for Sharpe ratio it all drops out. A quick way too look would be the excess return series of some of the index products vs the tbill yield.


nodoodahs


Total Posts: 227
Joined: Sep 2007
 
Posted: 2010-01-15 03:32
It's not at all about the return due to tbills. The excess return of the GSCI itself varies with the RF rate.

Take the 40-year monthly history of the GSCI and split the months into halves, based on RF rate. The months in the higher half of RF rate will have (1) higher average EXCESS return, (2) lower standard deviation of EXCESS return, (3) positive skewness OF EXCESS RETURN instead of negative skewness, and (4) higher excess kurtosis OF EXCESS RETURN.

Edited to add the emphasis on excess return, didn't want the issue of higher collateral yield to be confused. Yes, they ALSO have higher collateral yield, but the excess is influenced by the RF rate (which is in reality a factor of government response to current economic conditions including multi-year lookbacks on CPI changes).

I haven’t seen a beatin’ like that since somebody stuck a banana in my pants and turned a monkey loose.

SimJimons


Total Posts: 198
Joined: Aug 2007
 
Posted: 2010-01-15 11:31

svquant, I must be missing your point here! To me it sounds like you are saying that the RF-rate has an impact on CTA absolute returns?! But isn't that a given? I mean 90%+ of the portfolio is invested in bills! What am I missing?

nodoodahs, interesting numbers. Would you say that it's only the level that is of interest, or is the slope, or the slope given a level, also of interest? Can this effect primarily be explained by inflation hedging, and thus a massive flow into commodities while inflation/rates are rising/high, or do you see some other explanatory variables? Off topic, sorry!

/Sim


Sounds great...keep me out!

nodoodahs


Total Posts: 227
Joined: Sep 2007
 
Posted: 2010-01-15 13:18
I haven't looked at term structure of interest rates or rate of change for the RF yet, and may not, only so far looked at level of RF rates. Also haven't looked yet to see if it's an effect on spot trend or roll yield. I find it most interesting that the skewness is so different. To me, the larger question is, does this type of impact provide sufficient additional pricing information to incorporate it into a system, when one is already using other signals to determine position(s) in commodities?

Bringing it back to topic somewhat, the RF doesn't just "drop out" of return calcs. The RF itself has variance, and the level of RF (as well as its TREND) influences the returns of different asset classes. This should be self-evident when dealing with FI assets and currencies, should be quick to notice in equities, and definitely occurs in the GSCI.

So "how good a Sharpe?" may depend on the rate regime in which it was earned, in addition to the previously noted variance in return of different strategies (which probably isn't all related to RF rate regimes).

I haven’t seen a beatin’ like that since somebody stuck a banana in my pants and turned a monkey loose.

svquant


Total Posts: 113
Joined: Apr 2007
 
Posted: 2010-01-19 08:05
Sim - my point was to be aware of the contribution of tbill return on that 90%+ of the portfolio to any CTAs quoted rate of returns.

Nodoodahs - we were talking a different language since the original post did not specify excess returns and the RF does drop out in the SR calc based on total returns. Now your observation that excess returns are related to RF rate regimes - perhaps so or perhaps the "RF rate regime" is a proxy for (or components of) a risk aversion measure and one is just seeing how commodities perform under risk seeking and risk adverse regimes. Or if we are talking about GSCI really how energy performs under these regimes...

Dynamic Turtle


Total Posts: 165
Joined: Sep 2006
 
Posted: 2010-01-19 12:58

Cordura, the source data for that chart - does it say much about the constituents and their strategy? Is it a mix of discretionary CTAs and/or systematic trend followers?


hedgeQuant


Total Posts: 233
Joined: Dec 2006
 
Posted: 2010-01-26 02:28

Slight thread jack: I was computing my strategy's  SR based on monthly net returns and I am getting 1.8 with annualized returns of 9%. This is based on a 5 - year track record. For the moment I have not used a risk free rate.

This is a market neutral equity strategy focusing entirely on liquid US equities. Worst case drawdown was 6% with no down years. Any idea what the appetite would be for such a strategy in the current environment?


LowDD


Total Posts: 142
Joined: Feb 2009
 
Posted: 2010-01-27 04:11

Depends... on your background, capacity, structure, investment drivers.

There are lots of silly people who pick-up this kinda thing.  I would say... develop a bucket of 4 underlying strategies, each of which should target 2-to-1 return/max drawdown.  Vol and drawdown are not that hard of problems to solve...

I encourage analysts like you to consider their job... "developing new strategies".  So we watch the 4 on-going, and we look for #5.  Of course we have some criteria that allows the existing bucket to "fail"... which it always will.  So then we add new stuff.


hedgeQuant


Total Posts: 233
Joined: Dec 2006
 
Posted: 2010-01-27 04:43

 So we watch the 4 on-going, and we look for #5.  Of course we have some criteria that allows the existing bucket to "fail"... which it always will.  So then we add new stuff.

Could you please explain this a little more?

Previous Thread :: Next Thread 
Page 3 of 4Goto to page: 1, 2, [3], 4 Prev Next