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

Founding Member

Total Posts: 8335
Joined: Mar 2004
Posted: 2010-01-27 07:33
And, please explain how "max drawdown" isn't that hard to solve. Because, with what little experience I have that's the biggest and trickiest problem is controlling drawdown characteristics and consistency.

The Figs Protocol.


Total Posts: 228
Joined: Aug 2009
Posted: 2010-01-27 21:47
"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?"

Sorry for the delay on the answer.

It doesn't say much, only that they are the actively reporting funds and I don't have the underlying database. The report is from blackstarfunds, but they don't have it online. Cheers, Cord.

Vespertilio homo est cientificus


Total Posts: 142
Joined: Feb 2009
Posted: 2010-01-29 01:30

@ Hedgequant - Can you ask a more specific question?  Is it the "run 4 buckets... look for #5" that you dont understand (seems self-evident).  Or the latter point... "some failure condition" (first pass -10% on risk capital, second pass, losing 1/3 of our targeted ann return).

@FDAX - Solving "maxdrawdown".  Yeah, I get your point.  My point is that solving maxdrawdown is the same thing as having a good pnl stream, but most people focus on the upside and flipping things can illuminate.  This breaks out for me in a few ways:

1) If we cannot do really good returns from something with a pretty controlled low drawdown expectation... maybe we dont really know what we're doing.  I like to think about it as 3-to-1 ann return/drawdown target.

2) Then we want a basket of things going on, as I proposed in the general approach.  Each of these things should be separate and measured separately.  It should justify itself and stand alone. (PTJ - "a diverse set of highly skewed reward/risk trades" (or strategies).  It then follows if I make a mistake on 1/4 of my book and my theoretical multi-month 10% max draw gets blown through (-20% in weeks or days)... i'm only down 5% on my book.  We want to expect surprises and structure around them.

3) Then in each basket/strategy/bucket, etc... I tend to think alot about control.  I tend to favor a balanced L/S group of trades.  Of course this is wonderful in single stock equities.  But it applies just as well in generalized macro trades where it is always easy to peg a "risk long" and a "risk short".  Balancing the basket is THE best control for drawdown.  Much better than assuming liquidity (small scale, rookie mistake).  Of course, at heart this is an understanding of alpha v beta... or trying to "out-perform" risk taking.  Anyone can take risk.

Helpful?  Make sense?


liquidity peddler

Total Posts: 60
Joined: Aug 2007
Posted: 2010-01-30 02:40
For me, it is the dependency aspect between trades which makes Max Drawdown so difficult to "control". You can draw from a distribution for a parametric simulation or you can draw from a pool of data for a non-parametric simulation, but neither one tends to give you a very good idea of possible drawdowns in the future, because they ignore dependency between trades.

And the dependency tends to be non-stationary, so it doesn't model easily. What is the behaviour of trade dependency in a given strategy? When does it wax and when does it wane? It all seems to be a rather hand-wavy problem I fear.


Total Posts: 2
Joined: Jun 2017
Posted: 2017-06-29 12:20
Check this post out.
This posts explains this topic well
Previous Thread :: Next Thread 
Page 4 of 4Goto to page: 1, 2, 3, [4] Prev