Forums  > Risk Management  > fraud returns  
     
Page 2 of 4Goto to page: 1, [2], 3, 4 Prev Next
Display using:  

Nonius
Founding Member
Nonius Unbound
Total Posts: 12671
Joined: Mar 2004
 
Posted: 2009-04-16 07:01

Risk Data..."quantitative beats qualitative". give me a fucking break.  let's see: unknown auditor, self-custody, self-admin, opaque, and unusually good returns.  I don't think heavy quant tools beat qualitative analysis here.  but, my fraud detector does in fact register high fraud risk for the madoff feeders.


Chiral is Tyler Durden

Tradenator


Total Posts: 1582
Joined: Sep 2006
 
Posted: 2009-04-16 07:10

I ran a CTA I know through the bias thingy and it looked like a reputable CTA.  But I know the people are shady and I am pretty sure the track record is total fiction, so the bias ratio did not help here.  Benford picked up on it, though.


Nonius
Founding Member
Nonius Unbound
Total Posts: 12671
Joined: Mar 2004
 
Posted: 2009-04-17 20:43
hmm. who were the service providers?

Chiral is Tyler Durden

Tradenator


Total Posts: 1582
Joined: Sep 2006
 
Posted: 2009-04-18 06:33
custodian = HSBC, administrator = Bank of Bermuda, legal = Walkers, auditor = EY.  no prime broker named, last audit date not given.

Nonius
Founding Member
Nonius Unbound
Total Posts: 12671
Joined: Mar 2004
 
Posted: 2009-04-18 08:30
seems like it would be difficult to cook the numbers in such a case, no? unless the EY thing is simply false (or maybe they lied about having the other service providers.)

Chiral is Tyler Durden

Tradenator


Total Posts: 1582
Joined: Sep 2006
 
Posted: 2009-04-18 08:53
Fiction, in the literal sense.  I think they just made up numbers and put them out on funds databases.  Most likely, there arent't any investors even.  They have several funds, and one is genuine but at least one other isn't.

meteor


Total Posts: 165
Joined: Feb 2007
 
Posted: 2009-04-30 05:20
Not implying anything, have you guys looked at this:
http://www.scribd.com/doc/14713748/PontaNegraFeb09
Truly impressive result: not even a down month.



malsain de corps et d'esprit

iparkins


Total Posts: 59
Joined: May 2005
 
Posted: 2009-04-30 09:44

worth reading about them at www.brontecapital.blogspot.com - looks like they belong in this thread

 


dehaan


Total Posts: 104
Joined: Oct 2008
 
Posted: 2009-05-11 09:33

>Funny, RiskData got the Financial Times to trumpet their new fraud detector (aka bias >ratio) to the skies as something which will make all your whites whiter. I'll never believe >anything I read there again. If they're just reissuing marketing crap, I may as well read >the Weekly World News.

Journalists being journalists...BR is a statistical indicator and should be understood as such. That is if it is , say, 6 on a fund, it means that there could be issues with returns adjustment. But it may be as well that the fund is clean. And you would do your qualitative analysis more closely only on those funds that have big BR, that are suspicuous, that's all. This indicator becomes handy when you screen, say, a thousand of hfs.

 


Nonius
Founding Member
Nonius Unbound
Total Posts: 12671
Joined: Mar 2004
 
Posted: 2009-07-29 23:17

we tried Benford on buttloads of in-out samples of 1000s of HF returns.  It sucks.  Aaron, if you can think of a good test for null other than Chi-Squared, I'm all ears.

my method works well, but maybe there's some weird bias.


Chiral is Tyler Durden

aaron


Total Posts: 746
Joined: Mar 2006
 
Posted: 2009-08-04 15:53
You mean it "sucks" because all hedge funds pass? That's not really surprising. Benford would only catch people making up returns in their heads, who didn't know about the concept. My guess is most frauds would report honest returns when they were acceptable, but trim losses and maybe some big gains as well. They might sometimes report a great return to stimulate more money. But since they're starting from a real number and adjusting, the digits are likely to be random.

Remember, unless you're Madoff, you have to show people numbers that add up, and some of which match publicly available ones. So you cheat by having some "ABC Trust" or exotic derivative that you assign a high value to, then add it up with all the rest.

I think Chi-square is the good test for the null hypothesis, but you have to be sure to adjust the Benford prediction for the range of returns.

doctorwes


Total Posts: 576
Joined: May 2005
 
Posted: 2009-08-04 19:32
Suppose somebody took a genuine return series (e.g. for the S&P) and added a small constant x% to every value. What kind of test would pick that up, assuming you did not have access to the original series?



aaron


Total Posts: 746
Joined: Mar 2006
 
Posted: 2009-08-05 01:06
None. Even with the original series, there are strategies expected to have a similar return pattern (if the number were exactly the same every day, of course, it would be suspicious).

However, when people cheat, they're more likely to reduce the variance than increase the mean; or they do both. Reducing maximum drawdown and volatility can make the fund seem much more attractive, and means you don't need a lot of fictional cash for a long period of time.

adas


Total Posts: 185
Joined: Sep 2004
 
Posted: 2009-08-05 02:29
Nonius - where are you sampling your fund returns from: HFR, TASS, CISDM or elsewhere?

Have you considered smoothing your returns to remove serial correlation? The Benford test fails if returns are clustered. Suppose a fund reports the following returns: 8, 8, 9, 13, 9, 9, 12 - clearly Benford's test is not suitable here.

I don't think, therefore I ham.

Nonius
Founding Member
Nonius Unbound
Total Posts: 12671
Joined: Mar 2004
 
Posted: 2009-08-05 21:06

You mean it "sucks" because all hedge funds pass?

no meaning sucks as in for the sample of funds that we know are fraudulent and not fraudulent, there seems to be no correlation between pass or fail and fraud and not fraud.  anyway, I'd prefer a model that assigns a probability of fraud, not some metric that you'd have to trust with a digital answer assuming the Chi Squared rejection works.

adas, i'm using an in-house DB, which is actually better than those commercially available DBs.  I didn't think about removing serial correlation although I did notice that with my fraud detector, it is probably giving false signals on returns with high autocorrelation.  (actually the two theories behind autocorr are a) PL smoothing, which may be fraudulent and b) illiquidity of assets.)


Chiral is Tyler Durden

adas


Total Posts: 185
Joined: Sep 2004
 
Posted: 2009-09-01 03:58
Forthcoming JFQA paper on a potential method to screen for hedge fund fraud:

Bollen and Pool

Applies the conditional returns smoothing technique that gutenberg and others suggested earlier on in the thread.

I don't think, therefore I ham.

purbani


Total Posts: 89
Joined: Apr 2005
 
Posted: 2009-09-02 05:59

The problem with this approach is that there are several possible reasons why HF return series may be autocorrelated other than fraud. In fact autocorrelation or 'stale pricing' may well be a significant source of alpha in its own right relating to informational inefficiencies in the market place that hedge funds are able to exploit. You are thus left with the same problem as with the Bias ratio where the mere presence of suspiciously / relatively high positive skewness could be evidence of either skill or fraud or pro-forma in sample backtests (fraud lite ?) published as real returns. All worth flagging to be sure but not conclusive.

Our research indicates that about 1/3 of all HF suffer some some form of autocorrelation and that in general this tends to understate volatility by about 1/3 for those funds. Can be corrected using any number of means incl Blundell (note not Blume as shown) Wald - Kalman filter as in the linked presentation.

http://www.infiniti-analytics.com/kb/kb/article/quantitative_methods_in_hfof_construction

IMO - the assumption of i.i.d a far more serious transgression that flies in the face of all common sense, behavioural finance herding, momentum and empirical evidence. It just aint so not even for equities.


chrigl


Total Posts: 5
Joined: Nov 2009
 
Posted: 2009-12-02 14:33

I would like to play around with return time-series of fraudulent and non-fraudulent funds, using pattern recognition algorithms (e.g. bagging algos) to see, if it is possible to get any discriminatory power. Is there anyone willing to share such historical data? Or is there any public database where one could get it (e.g. at academic institutions)?


jslade


Total Posts: 1089
Joined: Feb 2007
 
Posted: 2009-12-03 09:08
You can find a large collection of monthlies at
http://www.iasg.com (registration is free). You have to screen scrape though. They don't tell you which ones are frauds either.

Bagging ... can you use that on different kinds of learners? Seems like it would be iffy on monthly time series data in general, compared to something like Benford or one of the Kalman thingees in the literature.

"Learning, n. The kind of ignorance distinguishing the studious."

chrigl


Total Posts: 5
Joined: Nov 2009
 
Posted: 2009-12-03 10:41

Bagging could be used for different classification algorithms, yes. However, the idea was to start with a collection of potential statistics & indicators (you named some) as input variables, not plain return series.

PS the great book by Hastie, Tibshirani and Friedman is now available as PDF: http://www-stat-class.stanford.edu/~tibs/ElemStatLearn/).

 

 


Nonius
Founding Member
Nonius Unbound
Total Posts: 12671
Joined: Mar 2004
 
Posted: 2009-12-22 07:43
08
You can find a large collection of monthlies at
http://www.iasg.com (registration is free). You have to screen scrape though. They don't tell you which ones are frauds either.

Bagging ... can you use that on different kinds of learners? Seems like it would be iffy on monthly time series data in general, compared to something like Benford or one of the Kalman thingees in the literature.

on the front page, they list Valhalla....does this mean they mix fraud and non-fraud returns in the DB? 


Chiral is Tyler Durden

Tradenator


Total Posts: 1582
Joined: Sep 2006
 
Posted: 2010-01-07 03:41

Morningstar has stepped in to the fray with a scheme that interprets serial correlation to flag returns with high risk of manipulation.  I only know this because mine are flagged as guilty of large serial correlation.  Under this logic, how would cash returns be rated?

 Attached File: Morningstar Hedge Fund Operatonal Risk Flag Methodology.pdf

Operational Risk Flag

As a compliment to Morningstar's performance-based Morningstar Rating, we assign an Operational Risk to all hedge funds. To read our methodology, please click here.

 Auditor risk warning      Auditor risk warning

 Administrator risk warning     Administrator risk warning

 No registration information     No registration information

 Unusual serial correlation patterns     Unusual serial correlation patterns


Todd


Total Posts: 1
Joined: Mar 2010
 
Posted: 2010-03-18 17:44

Thing about all CTA reporting sites- they simply regurgitate the data provided by the CTAs.

That's not to say that database operators can't offer insight, but pursuit of that information quickly wanders into the qualitative rather than quantitative realm.

A program hits my screen as the result of returns, but I only make an allocation after significant discussion...

Probably oversimple for this group, but here are some points I try to cover: http://mkmanagedfutures.com/selecting.html

What it comes down to (for me, anyway) is to gain an understanding of the source of returns and a grasp what exposure a client will experience. Without these two, therest is largely useless to me.

Just my two cents worth...


urnash


Total Posts: 560
Joined: Sep 2006
 
Posted: 2010-04-12 09:48
adas alread mentioned a paper by Bollen & Pool, but according tto http://allaboutalpha.com/blog/2010/04/11/academics-dabble-in-vexillology-to-better-predict-hedge-fund-fraud/ they've written a new paper. Haven't read it; I'm only posting it for interested NP-ers (at their own risk...)

understanding shit will always be worth something -- filthy

amateur


Total Posts: 147
Joined: Mar 2010
 
Posted: 2010-05-11 06:19
Has someone ever read the below paper?Any comments?

Estimating Operational Risk for Hedge Funds: The ω-Score

hxxp://papers.ssrn.com/sol3/papers.cfm?abstract_id=1335449


“unnecessary complex models should not be preferred to simpler ones. However . . . more complex models always fit the data better”
Previous Thread :: Next Thread 
Page 2 of 4Goto to page: 1, [2], 3, 4 Prev Next