Measuring your correlation to professional futures traders

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.
sluggo
Roundtable Knight
Roundtable Knight
Posts: 2987
Joined: Fri Jun 11, 2004 2:50 pm

Measuring your correlation to professional futures traders

Post by sluggo »

How are you different from all the other money managers out there? Why should I invest my money with YOU rather than with your competitor?

You'll probably hear these questions a lot if you solicit money from institutions, and so it would be a good idea to have an answer ready.

One answer that seems to sell pretty well is: "My performance is good, AND my correlation to other traders is very small (or better yet, negative). Therefore adding me to your stable of CTA investments will diversify your holdings and smooth your portfolio-wide equity curve."

However this requires you to
  1. calculate the correlation between (your suite of systems) and (other traders)
  2. optimize your suite for good performance and low correlation
I'll briefly touch upon #1 here, leaving #2 as an exercise for the reader. :D

You could do worse than measure the correlation between the monthly returns of your system suite, and the monthly returns of an industry benchmark index. To help you accomplish this, I have attached a modified returns series for one of the industry standard benchmark indices: the Barclay Systematic Traders Index. (google it).

It's been modified to make life a bit easier for you: Management Fees and Profit Incentive Fees have been added back. This is handy because it's easier to simulate your system without them; so if you're doing a huge optimization with thousands and thousands of tests, you skip the work of calculating fees and incentives, thousands and thousands of times.

I added back the standard 2-and-20 fee structure. 2% annual management fee, plus 20% net-new-profit incentive fee. So you can compare your simulations (with no mgmt fees and no profit incentive fees) against a no-fees version of the index. Apples to apples. It's in the file "index_before_fees.csv" below.

For those contemplating writing their own program to add back fees and incentives, I've also attached a little test case that may aid in debugging. Don't forget about the High Water Mark (the "new" part of "net-new-profit" incentive fees). You'll find it's easy to calculate after-fee returns starting from before-fee returns, and more difficult to calculate before-fee returns starting from after-fee returns. But don't despair, it can be done, in only 85 lines of code, so it can't be that complicated. Once you've got this program, you can run it and get before-fees returns from Winton Capital, or Chesapeake, or Eckhardt, or any other before-fees return series you like. Then you can easily calculate the correlation between (your trading suite) and (Winton) ; and, if you wish, you can optimize your suite to have low correlation to Winton. How cool is that?
Attachments
test_case_for_programmers.csv
Both after-fees AND before-fees returns. A test case.
(7.75 KiB) Downloaded 478 times
index_before_fees.csv
STI before fees.
(6 KiB) Downloaded 557 times
Chelonia
Roundtable Knight
Roundtable Knight
Posts: 525
Joined: Mon Apr 30, 2007 3:37 pm

Post by Chelonia »

Great idea Sluggo.

Why don´t we all put down our Correlation number between the system you (currently) use and the BSTI.

Our´s is 0.58, which is in line with most other (large) CTA´s. The mean runs around 0.60

Running this number down is possible but in most cases the correlation between the program and for example the MSCI World Index then increases which is not favorable.

Therefore you should also report the correlation between your program and the MSCI World Index.

Ours is: -0.23

So the task to which Sluggo is referring (2) is to take that (positive) correlation number (between your program and BSTI) down and leave your (negative) correlation number between your program and the MCSI World Index as is! :wink:

(if you want to use for example the S&P Index feel free to do so)


Correlation our program/BSTI = 0.58
Correlation our program/MSCI World Index = -0.23
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

On a micro Level i think what you are suggesting is okay Sluggo - although anyone who designs a trading program / or optimises it to show low correlation to other funds as opposed to designing a trend following system for robustness and profitability - first and foremost - I think is pretty crazy IMHO.
Secondly on a macro level I think all of this kind of thinkimg is way off the mark.

I have attached (I think it should be attached - if not I'll try again) the Lawrence Clarke long term trend followers index - a piece of research Ive been toying with / keeping and updating for several years
For me investors would simply be better off simply splitting their money out across all trend followers in an index of say , 20 trend followers, even if some of these supposedly have high correlations to each other.
The performance so vastly out-weighs the performance of some kind of modern day portfolio theory nonsense (stocks/ bonds / cash /alternatives)
that it looks pretty dam good to me.

The trouble with the correlation thinking around managed futures / trend following / CTA (which ever handle you use) is that what should be more important than correlation is "are they pursuing a strategy that overt the long term makes sense?" and since so few investment strategies do - over the long term - this question is far more important than if strats are correlated - would you put a strat in your portfolio if it were negatively correlated - if it lost money ?
Secondly the problem with trend followers is surely you have kind of success skew ? If you take the only truly sensible strategy out there and correlate all the funds that do it - then you will obvioulsy get a correlation skew (positive) - beacuse they are all successful - so to some edegree thaey have to be correlated ?
Unfortunately the idea of investing in different strategies that make money at differnt times and smooth returns and display low correlation begins to smell like the pursuit of the holy grail to me ?
I honestly think investors would be better off taking say 50 Mill and splitting it out in 1mill increments amongst 50 TF's - that way you reduce all the jother risks such as credit risk / terminal risks (manager death / hurricane strike yada yada ) - but you would in my humble opinion give your self the best chance of success over the long term

One last point - its actually quite remarkable just how low the correlation is between some trend followers and investing in a basket of say 20 of them - does actually produce quite stunningly smoother returns than you may anticpate ?

C
Attachments
July Monthly.pdf
(82.17 KiB) Downloaded 613 times
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

One last point
I confront the question "why should I invest with you as opposed to other TF's" - on a regular basis
The real honest answer is that there really isnt an answer - there all good investments - the other answer is to say " invest in both" - which again genuinely is a good option.

The trouble with all of this is that there is a divergence between the truth and the objective - i.e. raising money and doing the best / right thing
AbsoluteReturn
Contributing Member
Contributing Member
Posts: 8
Joined: Sun Apr 27, 2003 10:05 am

Post by AbsoluteReturn »

@Chris67

Is the "Lawrence Clarke long term trend followers index" updated regularly anywhere on the internet? A google search didn't help to find anything about this index.
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

Its about to be
I have it going back years
Im about to post it on my website which will launch in about 1 month
I have the index broken down to other sub sectors and groups also
It took me years to put this stuff together and I think it gives a good reflection of TF performance as opposed to "CTA" or Managed Future indices - most of which are a bit too generic for my liking
I'll post a few bits and bobs here in the next few weeks
Its only really for my own purposes and may well need some comments / suggestions
Thnaks for your interest
Chelonia
Roundtable Knight
Roundtable Knight
Posts: 525
Joined: Mon Apr 30, 2007 3:37 pm

Post by Chelonia »

What do we really know about the underlying managers in any (benchmark) systematic trend following index. Did they always lived to the system, or was there a directional overlay at times. How much alpha do we need to account for if any?
The only way of finding out is to compare system output with broker statements going back over the reported time and create another measure ratio called Systematic(Beta)/Alpha. However, that would require full transparency from the CTA´s.
Moto moto
Roundtable Knight
Roundtable Knight
Posts: 427
Joined: Mon Jun 01, 2009 4:12 am
Location: once again in the UK

Post by Moto moto »

I think measuring the correlation between yourself and others is a good idea, but it does not necessarily reverberate with why some fund of funds or institution is going to invest in you. Most of that is sales.
As mentioned unless you really look behind the scenes you may not understand what causes the correlation, and then you start getting into the never ending arguments over diversification v concentration.

A few references to some of the more successful traders (as opposed to fund mangers as asset gatherers) has always been that they ignore their clients and focus on the job of making money, and that has been what has ensured their success.
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

I agree Moto Moto
The only aim is to make money
whether you are correlated to other people is of no concern !
Also dressing up about alpha / beta and all that other financial jargon and crap concelas a very important point
If in 1985 or 1990 or 1995 I had 10 Million usd to invest and I split it out into 10 Lots of 1 million and stuck it with 10 TF's of the Hawksbill / Eckhardt variety - I would today have made more money than just about any other investment on the planet. All of those with hindsight will probably tell me buying microsoft or yahoo was a better bet - and of course it was - But I am taking about as compared to putting money in mutual funds or buildinga portfolio based on modern investment theory such as stocks / bonds / cash / alternatives !
This theorty has failed so dramatically its a nonsense - the U.K Pension fund being the biggest losers of all - having lost hundread of billions over the last few decades in their portfolios of equities / property / cash and fixed income !

Incidentally the 10 Year total return on sp500 including all dividends - at the ned of August 2010 is -17% but dont worry because stocks "always" go up in the long run !!!!
AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2071
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner »

Chris67 wrote: The only aim is to make money
You are of course right. I am a supporter and practitioner of TF or, as Graham Capital rather usefully calls it, Global Systematic Macro.

But I do rather wonder just how accurate one can be regarding survivorship bias. Unless you are actually around during its life to collect data for a fund or trading method which eventually goes phut!, the necessary data just disappears. Take JWH's "Original" program which has now disappeared - there are so many others like it.

It is fair enough to assume that one could indeed have replicated the performance of the S&P 500 over the past 30m years. It is self adapting - as company's fade and their market cap declines they are replaced by today's stars. It is a strategy which an investor could have replicated if he so chose over the past 30 years. You have only to read the very interesting article published by ecritt to realize just how few stocks survive over the long term and how important it is to devise some method to exit such stocks before all value disappears and they crash and burn.

On the (not very probable assumption) that one could obtain data for (a) all TF funds and (b) all TF programs which crashed and burned over the past 40 years one could probably devise some method which would have got one out in many cases before the relevant program was terminated - DD exceeding certain levels perhaps. Rather in the same way that the S&P 500 is often out of a stock before it actually goes belly up. You would have to set the benchmark very high or you would find yourself out of many programs (or at least managers) which survive to this day.

Even then, the matter is not that simple. Take Tactical - as we know, in its original form the program was terminated after a 50%+ DD and was reborn as the Tactical Institutional Program which employed significantly less volatility. Would you have immediately re-invested in the new program?

I suppose it takes us to the very heart of back testing and the rather useful warning the likes of which I used to preface my recent book:


The test results contained in this book represent hypothetical performance based on the
use of computerised system logic.
Hypothetical performance results have many inherent limitations, some of which are
described below. No representation is being made that any investor will or is likely to
achieve profits or losses similar to those shown. There are frequently sharp differences
between hypothetical performance results and the actual results subsequently achieved by
any particular trading program. One of the limitations of hypothetical performance results
is that they are generally prepared with the benefit of hindsight. In addition, hypothetical
trading does not involve financial risk, and no hypothetical trading record can completely
account for the impact of financial risk in actual trading. For example, the ability to
withstand losses or to adhere to a particular trading program in spite of trading losses are
material points which can also adversely affect actual trading results. There are numerous
other factors related to the markets in general or to the implementation of any specific
trading program which cannot be fully accounted for in the preparation of hypothetical
performance results all of which can adversely affect actual trading results.
Readers are strongly advised to conduct their own rigorous testing and research before
putting any of the ideas or systems described in this book into practice (if at all) and before
taking any financial risk.


While back testing and indeed the calculation of such indices as Chris describes are indeed very useful, the realist has to place himself back at the beginning of the calculation period and say “would I really have invested in this way, would life have really have worked out like this, would I have stuck to these funds, methods, programs over 30 or 40 yearsâ€
Jez Liberty
Roundtable Knight
Roundtable Knight
Posts: 123
Joined: Tue Nov 03, 2009 8:49 am
Location: London
Contact:

Post by Jez Liberty »

Chris67 wrote:Its about to be
I have it going back years
Im about to post it on my website which will launch in about 1 month
I have the index broken down to other sub sectors and groups also
It took me years to put this stuff together and I think it gives a good reflection of TF performance as opposed to "CTA" or Managed Future indices - most of which are a bit too generic for my liking
I'll post a few bits and bobs here in the next few weeks
Its only really for my own purposes and may well need some comments / suggestions
Thnaks for your interest
Chris - looking forward to your new website and the publication of the index.

I am also following a list of Trend Following "Wizards" on a monthly basis (published herefor anybody interested). This was mostly because I could not find one at the time.
I did compile once a historical performance with an aggregated index (here) for these Wizards. The index curve was quite smooth and returned 17.21% annually over 20 years - with a MaxDD of 22.88% (on a monthly reporting basis).

There is of course, as AFJ Garner highlights, the issue of survivorship bias in such index but one thing it shows is how diversification works wonders (ie the perf of the index is better than nearly all of its individual components...).

The alternative for a TF index might be a mechanical index free from any ties to fund managers. Probably the subject for another thread but this paper is an interesting read about that concept.
Chris67 wrote: The only aim is to make money
whether you are correlated to other people is of no concern !
To go back to sluggo's initial point, I think this is not true for at least a few large Trend Followers - as exemplified by Dave Harding, clearly saying that he offers diversified source of positive returns to his clients (he emphasises on the "diversified"). Additionally he is happy to lower his returns (ie as was discussed recently in another thread) as he feels his clients will be satisifed with those levels of return (8-10%).
That man could clearly "not listen to his clients" and go for a more correlated strategy shooting for higher returns but I am not sure he would manage AUM measured in billions...
I am pretty sure most fund allocators use correlation as a main input to their model (such as the MPT) and if you want to attract them you need to take it into account.

ps: in any case, thanks sluggo for starting this thread and sharing the useful attachments!
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

Thanks for your replies
On the long term index I have I include programs that have ceased trading so as not to incurr survivorship bias
The only program to cease trading all together was JWH Global diversified - I find it amusing that this "failed" program traded 3 months shy of 20 years - i.e. 237 Months cagr = 13.80% max draw down = 45.85% and a monthly compound ror of 1.05%
So much for a failed fund !!!

I take the point one has to be careful not to pick only the best funds - that iis why i include all funds
I have a UK index also that has tracked all onshore TF investment managers since inception - this is about 13 years old and Ive been tracking it since 2000 - very impressive too and will publish soon

I think these indices are more reflective than managed futures or CTA indices which include all manner of strategies in them
jklatt
Roundtable Knight
Roundtable Knight
Posts: 102
Joined: Sat Feb 17, 2007 11:51 am

Post by jklatt »

Sluggo,

Such a great post. I've often thought how difficult it must be for a new trend following manager to raise funds. Why would an institution allocate capital to the new manager when there are quite a few "old dogs" that are far more proven and have a structural advantage? It feels like two of the most popular approachs to overcoming the hurdle are:

1) Accept smaller account sizes.
2) Sell a somewhat "curve fit" (knowingly or not) backtest with gain vs pain ratios > 2 in hopes that it continues to preform long enough to make a significant amount of coin.

It's hard to blame the mangers for doing this because they're really just giving the investors (in general) what they want.

The other option that seems like it would sell like hot cakes is what you refer to as #2... and nobody likes to talk about #2. I really don't have anything meaningful to add either, but I've often wondered where the general sweet spot is in all of this correlation mumbo jumbo?

I fired up the simulator this evening and did the following:

1) Generated 1000 different random return streams for CSI symbol CC2. Direction was picked randomly and hold times were random between 1 and 300 trading days. I also added a random wait time between 1 and 300 trading days between exit and re-entry.

2) Defined a "Pain" statistic. I chose the Ulcer Index.

3) Ran through the 1000 different equity curves from 2 to 1000 and logged the correlation between equity curves n-1 and n and the ratio of the combined pain of equity curve n-1 and n vs. the pain in equity curve n-1.

4) Calculated 4th degree (4th seemed good, maybe it's not) least squares coefficients for the data set logged in step 3. X = correlations, Y = combined pain / original pain.

5) Plotted combined pain/original pain using the 4th degree fitting.


Below is a plot of the data that was generated. For this test, it appears that pain becomes exponentially more painful the closer you get to correlation = 1. That makes sense to me. What is suprising is that as you go from 0 toward -1, pain also increases. Really? At correlation = -1 shouldn't pain be zero (a flat equity curve)? Maybe it's just a small sample size? Maybe I'm about to have an "Ah Ha!" moment?

As I said above, I really don't have anything meaningful to add just yet, but the idea is to generate a LOT of equity curves and plot the general tendancy for gain realtive to correlation and pain relative to correlation and figure out where the gain/pain sweet spot is.

EDIT - scaled image down by 50% for the benefit of those without ultra high resolution screens - Moderator.
Attachments
Shrunk to 50% of original size
Shrunk to 50% of original size
half_size.jpg (71.99 KiB) Viewed 14024 times
original_big_image.zip
The original image at full size
(284.14 KiB) Downloaded 395 times
Last edited by jklatt on Wed Sep 29, 2010 11:54 am, edited 1 time in total.
jas-105
Roundtable Knight
Roundtable Knight
Posts: 130
Joined: Sat Aug 02, 2008 2:32 am
Location: London, England.

Post by jas-105 »

Chris, if lack of correlation hasn't convinced investors to place money with you then what about reduced fees ? Or does this make a new fund appear "cheap" and/or desperate ?
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

I'm afraid that the phlosophy the fund I run has is that we dont really try and persuade anybody to do anything - I think part of the success of the some of the old school trend followers was that if i / they believe in what we do then organic growth of 10-20 Million should take it up hugely over a 5 year period. Investors either get in on our terms or they dont get in - if an investor cannot see the merits then do you really want that sort of investor on board ? They will only pull their money when you've had a 5% draw down ?

What Ive found over the years and I think at some point on this forum I offered to run a seminar on the topic (although I was shot down as it was constrewed I was running a seminar on TB which I wouldnt do as I do not perceive myself to be an expert on TB - Just enough to make money) - was to do a seminar on raising money for those who wanted to pursue TF professionally and to try to pass over some of the pit falls to people
These pitfalls are typically that if you think you are going to raise money from institutional investors then you would be better off trying to crack the meaning of life or split an atom - its virtually impossible in the TF world from an institutional perspective.
A great example is mulvaney Capital in London - an awsome TF program - 10 year track record of about 20% CAGR - but stuck at 100 Mill give or take for many years - I m not sure by choice - but investors wont take the draw dwons
Institutional investors are light years away from understanding how to get a long term return on their money - they still look for the holy grail / no volatility / no draw down / un correlated to everyone else approach - I think this is failry nonsensical
I believe th ebest bet for raising mmoney are contacts and individuals who make the decisions on behalf of their own money on the spot.
The trouble with institutions is that it "aint" their money - they have other concerns which dont necessarily revolve around making a return (the old cant go wrong buying IBM theory)
Its a bit like why the financial mess developed in 2008 - all the people in th einstitutions dont really givbe a shit about 3-10 years - they care about their bonuses in 6 months time and not foing something their mate down the road didnt do !!
If I'm 100% correlated to Milburn Ridgefield then you know what ... bring it on !!! that shoul mean im making 20% per year for the next 35 years !!! If an investor doesnt like this and wants to go invest with Milburn ridgefiled then bloody good luck to him - I'm sure he'll make a fortune
Thats the sad truth and the almost impossibility in this business
TK
Roundtable Knight
Roundtable Knight
Posts: 167
Joined: Tue Apr 15, 2003 5:45 pm

Post by TK »

Chris67 wrote:A great example is mulvaney Capital in London - an awsome TF program - 10 year track record of about 20% CAGR - but stuck at 100 Mill give or take for many years - I m not sure by choice - but investors wont take the draw dwons
As of now, their CAGR is closer to 13%.
Attachments
mulvaney.png
mulvaney.png (15.29 KiB) Viewed 14058 times
Chelonia
Roundtable Knight
Roundtable Knight
Posts: 525
Joined: Mon Apr 30, 2007 3:37 pm

Post by Chelonia »

No wonder they can´t attract any institutional funds with a -41% drawdown. If they would take 0.25% of their current leverage instead they might have a chance. Deleveraging could also open up less liquid markets for more diversification.
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

Chelonia - once again - all you focus on is draw down ? You WILL NOT make money without accepting draw downs ? Mulvaney could de-lvereage and then what ???? make 5 % per year ?? who would put their mony in that when you can get that in fixed rate savings ??
Also why is 90 % of teh Worlds institutional money happy to track the SP500 which has had a 51% draw down and a 6 year D/D max length

Have a look at som eother funds out there ??? How about Mark Mobious ? templeton - doesnt he have a cagr of about 1.5 % / max d/d of about 50-60% and he's the big guru on CNBC ???? You should get hold of the HSBC private banking review of all hedge funds - covers about 1000 funds across all sectors and about 800 Bill in assets
There is currently about 200 Bill in hedge funds with a negative 10 year CAGR with d/d's of about 50-90%
Accepting a draw dwon is fine if you are actually making money ?

If an investor wont accept a 41% draw down to make 13% per year but they will accept a 51% draw down to make -1% per year (most equities over the last 10 years) then what does that say about the intelligence / ambitions of the average investor ?
Please show me an investment that has no draw down and no volatility and has made an average cagr > 15 % / 20 % over the last 20 years ??? there aren't any ?
Chelonia
Roundtable Knight
Roundtable Knight
Posts: 525
Joined: Mon Apr 30, 2007 3:37 pm

Post by Chelonia »

Your right Chris, all i focus on is drawdown and that will never change. I´m not saying i don´t accept them.

Can you tell me where i can get a 5% fixed rate saving :lol:
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

Chelonia
A lot of offers in the UK at present for fixed rate savings bonds - about 3 year lock in gets you about 4.8 % - thats a good deal admitedly - rates are definately creeping up
I did a 3 year lock in recently for about 4.25%
From what I can see fixed rate UK savings are anywhere beyween 2.25% and 4.8% depending on how long you lock in for

Be great if rates went up to 10% again in the U.K as some are predicting over the next few years - how will that leave many short term hedge funds who only aim to make 7-10 % ?? If I could get a risk free 10% I know where my money would go ( risk free in the context of a crappy U.K bank that is)
Post Reply