How is quantitative system trading profitable?

[mazzatelli]

My post wasn't meant to be condescending.
I typed out a whole bunch but lost it in the refresh. Summary as follows:

1) Historically index related returns are leptokurtic (> mean reversion than normal) and so edge is to short vol with random entry, since this edge persists.

2)

that would be the offset or 'adjustment factor' we have discussed in previous posts? but how do we know that risk premium is 'correct', since we cant really know the probs or impact of all black swan events? since the risk premium the market puts on it is a bit of a guess , it is quite possible it is under or overstated.

Index vols are systematically rich in general - i.e. traders consistently price a premium to realized vol, but thats not to say they are rich at any given point in time. Your question is absolutely valid, most cases we don't know if the risk premium is correct until after the fact. Empirically despite the risk premium it still underprices options, hence vol skew. otm vols are a function of atm vols so when I discuss them, they are not independent of each other when modeling the surface.

3)

i am guessing that the positive expectancy would be equiv to either inflation or a risk free rate or somewhere round there?

Yeah sorry, that originally should have read positive returns. It's +ve expectancy if > inflation and risk free rate as you've stated

4)

Regardless of whether it could be traded as a system with +EV, would it not be fair to say that if options were always fairly / risk neutral priced then the net result of such a strategy should be random swings ending up at zero over a long term, not an upward sloping equity line?

As stated in point one, if this "edge" persists so one can take random entries every month, is this the case with Barclay's strategy? Or do they have a specific signal of which they trade from?
Looking at their long/short vol portfolio the triggers are based on 2 metrics a) component stocks of a sector are higher by x basis points than a sector index (e.g. ETF) and b) implied vol vs. adjusted implied vol (where they take out earnings days)
If the latter is the case, is the edge due to what you propose or their ability to discern changes in implied vol vs relative iv of sector and historical iv. I'm guessing their index signals would have some metrics linked to the VIX.

Another reason why I am interested, I see in one of their publications, they are short the variance swap and hedged using the VIX.

http://www.scribd.com/doc/35884388/Barclays-Index-Volatility-Weekly-20100809
Here they look for put spreads, based on relative value to other sectors

5)

Furthermore, it is my understanding that the whole BS model is based on the principle that the value of an option should be equivalent to the amount gained or lost by delta hedging (ignoring costs) if RV turned out to be equal to IV paid.
If that was in fact true, then the result of an unhedged ATM straddle strategy should be the same as the delta hedged strategy (long or short)?

Assuming iv=rv and optimal hedges were possible, then atm vol theoretically equates to all payouts. The probability of payout for the position whether hedged or not does not change.

The expected return could be different (as shown in your graph).

6)

quite possibly , I'm not the expert in stats

There's no need to trust me, since I know you DYOR. This is though, stats 101 regarding increasing sample sizes to reduce sampling errors. With financial data the errors are heteroskedastic, so increasing the scope (in this case years) does not improve estimates hence all the literature surrounding volatility measures that include open, close, high, low when working with daily data and on the other side high frequency data.

EDIT: lol bs this was a summary, half a page...sorry goponcho

 

[village idiot]

1) Historically index related returns are leptokurtic (> mean reversion than normal) and so edge is to short vol with random entry, since this edge persists.

Isn’t that what I have been saying and you have been disagreeing with? Or are you now referring only to returns (ie at expiry) rather than vol path on the way?

I think I maybe see the light – we have perhaps been talking at cross purposes about two slightly different things – volatility and returns. In brownian motion theory, (distribution of) returns is a direct function of volatility, so they are two sides of the same equation - if you know the vol , you can predict the return distn. But we know in the real world the leptokurtic/ mean reverting distribution of returns does not match the assumed normal distn and is not necessarily a direct function of volatility, so the two begin to diverge making two approaches possible

Where our approaches differ is that, you seem to concentrate on the volatility, I’m more focused on returns (but without ignoring the volatility path on the way).

Why is this so? Well because I am a retail punter who has to pay significant transaction costs**, which largely prevents or at least restricts me from engaging in strategies involving too much trading in and out or dynamically adjusting hedges. Thus I pretty much had to concentrate on strategies that would only require minimal transactions, so I tend to look for edges in distribution of returns v vol paid or received. Then hedge as little as possible.

Whereas your background is from a bank, with minimal transaction costs and maximum scope to data crunch and take small edges, which is more conducive to straight vol or vol arbitrage plays.

**even using Interactive Brokers, transaction costs are large enough to impact significantly on a strategy requiring frequent hedging adjustments, much as I would like to use them


Here is the article I pulled that graph from

http://www.surlytrader.com/volatility-selling-strategies/

It is short on detail and it does vaguely refer to a Barclays ‘piece’ as the source of it, which I haven’t seen but you might be able to find.

But without that ‘piece’ I saw nothing to indicate it is for anything other than consistent periodic entries with no other filters or inputs. At least that is what I assumed, otherwise it wouldn’t be analysing the effectiveness of volatility selling strategies, it would be analysing the effectiveness of Barclay’s signals for selling vol, which as you say is a different thing. What do you see that indicates the metrics you suggest are being used, or are you referring to some other portfolio? edit'; could you post a link to the long/short portfolio you mention. thanks

Going back to what I said above about the two approaches; in the chart I posted, the unhedged ATM short straddle equity line is essentially the ‘asset returns v IV ’ approach, the delta hedged strategy equity line represents the ‘volatility arbitrage; (RV v IV)’ approach

 

[mazzatelli]

Isn’t that what I have been saying and you have been disagreeing with?

Was just summarisizng what you said.
I agree about asset returns have the characteristic of leptokurtosis. Don't agree that in itself is an "edge".

Energy, fx etc all have leptokurtosis and a risk premium priced into the ops. But sure wouldn't sell ATM vols regardless every month.

I think I maybe see the light – we have perhaps been talking at cross purposes about two slightly different things – volatility and returns. In brownian motion theory, (distribution of) returns is a direct function of volatility, so they are two sides of the same equation - if you know the vol , you can predict the return distn. But we know in the real world the leptokurtic/ mean reverting distribution of returns does not match the assumed normal distn and is not necessarily a direct function of volatility, so the two begin to diverge making two approaches possible

Leptokurtosis is still taken into account. For e.g. guy I knew on the equities desk priced ops under a smile model + Heston, but hedged under Black-Scholes.
Another example :If you look at vol modeled under GARCH - it will produce a leptokurtic distribution.

Where our approaches differ is that, you seem to concentrate on the volatility, I’m more focused on returns (but without ignoring the volatility path on the way).

It reminds me a lot of the systematic Iron Condor selling to take advantage of the fact the index "doesn't gap and tends to mean revert".

But without that ‘piece’ I saw nothing to indicate it is for anything other than consistent periodic entries with no other filters or inputs. At least that is what I assumed, otherwise it wouldn’t be analysing the effectiveness of volatility selling strategies, it would be analysing the effectiveness of Barclay’s signals for selling vol, which as you say is a different thing. What do you see that indicates the metrics you suggest are being used, or are you referring to some other portfolio?

I was referring to another portfolio, and guessing since I cannot see the background to these pictures you posted.

I was surprised, because if it was a straight vol play as you say, then the variance swap should come out on top vs. a vanilla straddle. From the article it seems they didn't price the tails 'large" enough hence the larger loss comparatively.

So are they playing vol or distribution (despite the title of the article)?

Going back to what I said above about the two approaches; in the chart I posted, the unhedged ATM short straddle equity line is essentially the ‘asset returns v IV ’ approach, the delta hedged strategy equity line represents the ‘volatility arbitrage; (RV v IV)’ approach

I thought the trading signal was the same and comparing the results using different vehicles?

 

[mazzatelli]

I was looking at returns of CTA's this morning and remembered this thread. Results were a mixed bag for those trying to capture index volatility risk premium.

LJM Partners, one of the larger ones ($300M AUM), YTD is about -4%. Ace Investments up 5%, although was -46% previous year (~both 10 year records). Cinamen Vol Arb Program opened in '09 and is ~-29% YTD. Diamond Capital on the other hand is up 2.2% YTD (6 year record) and OpHedge is up 4.66%, although they do not focus entirely on index.

ML's index which tries to capture the implied/realized spread has taken a bit of beating (though they sell the 90-day var swap instead of the straddle). Clicking on the 3 year chart looks rosy, while the 5 year, not so much
http://www.bloomberg.com/quote/MLHFEV1:IND/chart

imo It's debatable whether the risk premium adequately compensates left tail risk.

 

[village idiot]

didnt know thos existed till now so thanks for the heads up. I had a look at LJM and ML. couple of points;

1. LJM shows -4.12% for the full calendar 2011 year. i would say that is pretty acceptable for a year with an outbreak of vol like we saw in august, when any years without such an outbreak return figs like 54% 42% 48% 21% etc.

2008 to be sure was huge drawdown for any of these sorts of strategies (-49%), whether 1 huge drawdown is enough to write off a strategy that shows 17.7% annualised return over the longer term (i am using LJMs figures here,not ML) is debatable. On those figures it has a poor risk;reward but is nevertheless +EV

2. I note LJM have a 20% incentive and a 2% mgt fee, although it is not clear how often the 20% is raked. To give away 20% of your profits when you have a win, (but not get back 20% of your losses), plus 2%pa of your capital regardless, would put such a handbrake on the whole scheme that it would probably turn a strategy that was +EV into -EV. That headwind must be built in to the published performance figures, so you would be expect the performance of the actual underlying strategy, which is what we are debating here, to be considerably better than the reported fund performance.

With such a negative edge added in it is a wonder they made anything at all.

 

[mazzatelli]

1. LJM shows -4.12% for the full calendar 2011 year. i would say that is pretty acceptable for a year with an outbreak of vol like we saw in august, when any years without such an outbreak return figs like 54% 42% 48% 21% etc.

2008 to be sure was huge drawdown for any of these sorts of strategies (-49%), whether 1 huge drawdown is enough to write off a strategy that shows 17.7% annualised return over the longer term (i am using LJMs figures here,not ML) is debatable. On those figures it has a poor risk;reward but is nevertheless +EV

Isn't that the nature of short gamma strategies? Personally I'd look at standard dev of returns in addition to expected value for this sort of strategy. I'd wager the Sharpe ratios of systematic vol selling is quite low. iirc there is a paper by Waggoner showing risk-adj returns to be positive, however the skew in returns is negative.

Nick Leeson of Barings bank infamy was also 'winning' if it wasn't for the Kobe earthquake. LTCM continued to sell index vol believing it 'overpriced' (but to be fair other issues contributed to its fall).

2. I note LJM have a 20% incentive and a 2% mgt fee, although it is not clear how often the 20% is raked. To give away 20% of your profits when you have a win, (but not get back 20% of your losses), plus 2%pa of your capital regardless, would put such a handbrake on the whole scheme that it would probably turn a strategy that was +EV into -EV. That headwind must be built in to the published performance figures, so you would be expect the performance of the actual underlying strategy, which is what we are debating here, to be considerably better than the reported fund performance.

With such a negative edge added in it is a wonder they made anything at all.

20% seems to be charged quarterly.

The program for LJM I quoted is 100% discretionary signals, and as far as I can see on my side, returns for all funds are quoted pre-mgmt/incentive fees and charges for consistency. Also they are selling index strangles as far as I can tell.

Ace is 100% discretionary, Cinamen is 100% Systematic, Diamond is 80% Systematic, Ophedge is 100% discretionary.

 

[village idiot]

we might not be looking at the same thing of course - here is what I am looking at for LJM

http://www.ma-research.com/ctaprof2.php?cta_id=238

I would have thought it odd if performances was published pre fees - as post fees performance is what public actually gets. to publish ex fees might be more level in terms of strategy perfomance comparison but pretty misleading to the public. But i have been wrong before..

Now here is something that will interest you - here is another fund i clicked on almost at random - this funds says it concentrates on vol skew opportunities although it does also sell strangles . although only running 4 years it is showing 33% annualised return with no yearly drawdowns and one monthly (out of 40) drawdown of 4%. Although it did conveniently start just after the 2008 GFC

http://www.ma-research.com/ctaprof2.php?cta_id=1307

 

[village idiot]

just been going thru the other funds you mentioned and realised that the second link i put was for OptHedge which you had already pointed to. apologies

 

[mazzatelli]

I'd have to check. I'm using works portals and viewing it as a CTA/MMgr rather than an investor.

LJM, their return is +ve, stdev is much larger compared to the second one although they invest only in index vols, while the other is spread across other markets.

Anyway, looks like I've destroyed the thread for the OP.

 

[mazzatelli]

I'd have to check. I'm using works portals and viewing it as a CTA/MMgr rather than an investor.

Checked this, appears to be net of fees. Apologies folks

 

[mazzatelli]

Updated YTD, figures pre-mgmt/incentive fees:
1) Discretionary: Stock Index Only

• LJM: 31.02%
• Ace Investments: 2.52%

2) Discretionary: Diversified

• K&Q Futures: -18.55%

3) Systematic: Stock Index Only

• Diamond: 7.16%
• GrowthPoint: Iron Condor: -9.24
• Vantage: -2.39%

4) Systematic: Diversified

• Clinamen: - 3.98%, Interesting note, one of the people involved in this fund is a strategist over at Condor Options
• Clinamen VIX portfolio Hedging: -15.22%
• OptHedge: 10.84%
• White Indian Trading: Straddle Program 4.17%

Chosen LJM distribution of monthly returns since inception, because it has the largest AUM and longest track record of the above

​

 

[Wysiwyg]

One thing that I am still unclear on is where we profit in systems trading. In order for us to profit, we need to be exploiting an inefficiency in the market right? What process goes into discovering these systems which exploit these inefficiencies? Trial and error after having an idea?

This website has some direction on system trading. The trend trading model is basic with the buy on any equal or new all time high and a sell using ATR(10) but no multiple given.

Quantpedia