Forums  > General  > How hedge funds measure the capacity of a strategy?  
     
Page 1 of 1
Display using:  

colabold


Total Posts: 2
Joined: Mar 2018
 
Posted: 2018-07-25 22:59
I have understood from other NP's threads that having a strategy that works well on new data is not enough for successful trading because strategies "do not scale". If so how can a potential hedge fund founder convince an investor that one has a working strategy?

TSWP


Total Posts: 401
Joined: May 2012
 
Posted: 2018-07-27 14:09
Maybe I can be helpful here...

Strategies do not scale: well, that depends on, there are some objective criteria to establish if a strategy can scale or not, for example, how much AUM it can handle?

"Potential Hedge Fund founder" must convince the investor with any means in his possession, words, data, papers, track records, etc., but it depends on who is the investor, how sophisticated and smart and experienced it is, is is an individual or an institution, etc.

Institutions have rules (and quants) to assess where to invest, individuals may be all sort of individuals (young people with hot money, old people with just money, etc.) so you are more on personal relationship turf with them, which is why there are so many ripoffs and Ponzi schemes, because individuals are usually not very capable to evaluate strategies.

Try to meet your investors and try to understand what would convince them to invest in your fund, nothing better than hearing directly from them what they want.


MCMC


Total Posts: 1
Joined: Nov 2016
 
Posted: 2018-07-30 15:29
What if instead of convincing external agents, you try to assess the scaling ability of your strategy for yourself (to know if you can allocate more funds to this particular strategy for example) ? What are the statistics / variables that you need to take into considerations ? I guess that market impact is one of the elements ?

urnash


Total Posts: 564
Joined: Sep 2006
 
Posted: 2018-08-07 22:17
About 10 years ago I was at a derivatives conference in Paris where Aaron Brown gave a very nice talk on the 'size of things'. He refered to an article about the size of animals and why a mouse can't be 10x its size (ratio between weight of bones and total weight, or something like that). And then he used that as an example why a fund and strategy has to have a certain size (not too big, or too small). It was very interesting and I can't remember much of it :-(

I just tried googling for this article, but couldn't find it. Not even in http://www.eraider.com/articles.php. Hope this nevertheless helps.

Jim Simmons is finalizing a c. 1000 page tome on Renaissance trading methods and is looking for a publisher. He was turned down by Wiley. They said he's taxing the attention span of their readers -- LongTheta

Scotty


Total Posts: 725
Joined: Jun 2004
 
Posted: 2018-08-08 02:24
I guess this suggests that domain knowledge - understanding the market you are trading - would be helpful?

This idea of necessary conditions being domain knowledge + data + model

On a related topic, does anyone have a view on the ability of institutional investors (family office up to pension funds) to select good managers? My priors are that there is a lack of true rigour along with the consequent impact of biases.

“Whatever you do, or dream you can, begin it. Boldness has genius and power and magic in it.”

TSWP


Total Posts: 401
Joined: May 2012
 
Posted: 2018-08-08 09:42
> does anyone have a view on the ability of institutional investors (family office up to pension funds) to select good managers?

If you exclude hedge funds, I think many of the other allocators have limited ability to understand what to look for in a manager, unless they do hire people (quants) to do that job for them, but then these "selectors" are not the owners so they are just doing a job and most likely worried with continuation of their career which in turn pushes them to try to limit damage, i.e. do not take any risks and do not hire any manager that does not tick all the boxes, it becomes just a hiring compliance exercise.

ronin


Total Posts: 318
Joined: May 2006
 
Posted: 2018-08-08 23:17
> Aaron Brown gave a very nice talk on the 'size of things'

I have never heard Aaron Brown's talk, but here is the rough outline of what may have been in it.

You calculate how much you make per day/month/year on average. You calculate the number of shares you trade in order to get to that. You divide one by the other. This ratio is called return per share, and it is a basic unit of profitability. Say the return per trade of your strategy is 5 bp.

Then you look at something called market impact. There is a million and sixteen models of market impact. Probably more by the time I finished typing this. They all give you some cost of trading as a function of your size or participation. So you calculate how much you need in AUM to afford the finer things in life (that seventh Italian villa, for example), you work out how much you would trade if you had that AUM, you run that through your favourite market impact model, and it gives you a number. Say your impact would be 2 bp.

So suddenly your theoretical return of 5 bp has to pay 2 bp of market impact. At that size, you have lost 40% of your return and 40% of your Sharpe.

So then, like @tswp says, you discuss that with your investors. Are they still interested if you are only showing 60% of your theoretical Sharpe and theoretical returns? If they are, you go ahead and ramp up. If they are not, you don't. Maybe you go looking for other investors who like that proposal. Maybe there is some medium ground. Maybe the whole thing just isn't interesting.


> does anyone have a view on the ability of institutional investors (family office up to pension funds) to select good managers?

Family offices and pension funds are basically a bunch of monkeys with a dartboard. They have no ability to pick good managers consistently. If they are lucky, they get some diversification without losing too much, and everybody is happy.

To be fair to family offices and pension funds, these days venerable big quant hedge funds outright lose to any monkey with a dartboard. The list of venerable big quant hedge funds that are still any good is shorter every day.


"There is a SIX am?" -- Arthur

TSWP


Total Posts: 401
Joined: May 2012
 
Posted: 2018-08-09 09:39
Maybe hedge fund managers could try to put themselves together and define criteria that they think are correct to evaluate a manager.

I am sure someone will disagree, but if done honestly it's a worthwhile exercise to create a common framework that could then be used by investors.

I am thinking about a sort of shared knowledge of what work and what doesn't work, but coming from practical day by day experience of fund managers, not the details of course, but the broad view.

Let's face it: only (some) managers have an idea of what can work and what cannot work.

Investors are only capable of evaluating strategies using metrics and other parameters but it is a wrong approach, maybe it should be managers to tell them what to check for in a manager.

gmetric_Flow


Total Posts: 3
Joined: Oct 2016
 
Posted: 2018-08-09 18:11
A shared knowledge network is certainly an interesting suggestion, but I’m not sure how easy something like this would be to adapt (what incentive do HFs have to do something like this). Excuse my naiveté, but why would an investor be interested in a manager’s style? As long as the fund’s results are in-line or surpass expectations (by whatever quantitative metrics the investor deems paramount), worrying about these subjective analytics seems silly.

deeds


Total Posts: 395
Joined: Dec 2008
 
Posted: 2018-08-09 19:57

@gmetric Flow - Most investors have money invested with multiple managers. Need to know the styles to risk manage across portfolios.

gmetric_Flow


Total Posts: 3
Joined: Oct 2016
 
Posted: 2018-08-09 20:54
@deeds - Of course, yet referring back to the idea that investors do evaluate strategies using "metrics and other parameters", what additional value would information from managers about other managers add?

If anything, it should be easier to evaluate risk across multiple managers from purely quantitative components.

Surely risk should be estimated using a model that is disjoint from "the pricing model", but there are other ways you could go about it without the aforementioned information. If anything, it seems that aggregating and leveraging this type of data would be cumbersome, especially since people are predisposed to bias- something which can create more noise than value.

I am so skeptical since I have thought about these sorts of things and I have always concluded that it would be useless, but it seems like others (much wiser than I) do not think so!

goldorak


Total Posts: 1042
Joined: Nov 2004
 
Posted: 2018-08-13 22:40
@gmetric_Flow: some people still need to justify their position and their salary Evil Grin

If you are not living on the edge you are taking up too much space.
Previous Thread :: Next Thread 
Page 1 of 1