We are now evaluating a software product claiming to provide better pricing algorithms optimized for FX options. The product, called SuperDerivatves, is based on a "revolutionary" model which does not draw on B-S at all. Is this wise? Isn't this too risky?

this is not, since traders use BS...

As I understand it superderivatives use models for exotics that are consistent with the volatility smile. i.e. A theoretical value (TV) for barrier options is produced using a single volatility (ATM) and then this TV is adjusted to a market price by calculating the cost of a portfolio of risk reversals and butterflies such that the dvega/dvol and dvega/dspot profile of the combined portfolio (including the barrier) is flat. This isn't revolutionary and has been used in the FX markets for sometime. There is a useful paper on this in a previous thread.

Thanks.In this case, could the model used still be a differentiator in preferring one solution over the other, or should I look for SuperDerivatives' advantages elsewhere?

I've briefly looked at Superderivatives and I think it's quite impressive, however it does depend on what your looking for. You cannot trade using the system and therefore the prices are not dealable, for me that is the major drawback. The vanilla vol surface is sourced from a range of counterparties which is very transparent anyhow. This in turn generates exotic prices using their proprietary model which indicates to you what you should expect when trading in the OTC market, however you will see differences!You might want to look at the UBS system.

Thanks again. We do wish to trade using the figures, so it is indeed an issue.I tried to access UBS and could not locate the relevant information. Could you kindly point me in the right direction?Also, do you know of other solutions such as SuperDerivatives'?

UBSI dont know of any others, i'm sure some of the other IB's have web based services. Failing that you need to rely on Reuters Dealing.

UBS is actually very impressive.I was told not to trust Reuters dealing. Any idea why?

Reuters Dealing is how the market trades with one another, there isn't anything to not trust about it. It will enable you to verify and shop around for prices from a variety of counterparties. However, if your not trading frequently enough it probably wont be cost-effective for you. In which case you might want to consider the web-based system. Reuters dealing is only a secure communication tool, it wont provide any analytics or pricing models. Mostly useful for spot traders.

Thanks again.

Hi, it's quite an old thread , but anyway, does anyone know a paper which Anthony was reffering to?I'm trying to understand the way SuperDerivatives works.... any kind of info and help highly appreciated..Regards,MaTT

Hi MaTT, What Anthony referred to is a 3 vanilla portfolio replication method that many houses use to quickly compute a "reasonable" barrier price more than 10 years ago. However, the true secret lies behind the "fudge". I believe SuperDerivatives uses an advanced version of it with "fudges" tailored to each ccy pair but no one will reveal publicly their method of choice. I've personally seen more versions than I can count with my fingers. Also, don't based your test of barrier models (assuming you are also looking at other software) on EUR/USD ... any well implemented "replication" model can produce prices that are in line with the market. Check out USD/JPY and if you are happy, go for EUR/HUF. That should turn on the heat of saleguys trying to sell you a software :-)p.s. Alex Lipton briefly mentioned the approach in his "universal" model article by RISK but notice that the "adjustment" is not mentioned at all, which is expected.

I've looked at superderivatives as well a couple of months agoSometimes very impressive but sometimes kind of scary:I found a USDJPY 80 one touch that was CHEAPER than the 75 one touch... yes you read that correctly and it lasted for about 2 wks (I checked every morning!), until I told their quant guysI tried to figure out how they obtain their prices. I think they're getting their market data from some broker (5 / 10 / 25 / 50 delta at least as far as i remember), which enables them to get a very accurate smile curve, and therefore correct prices for vanillas, even for EMKFor exotic options, they must have some kind of model as described by anthony, which they fudge with real market prices.For illiquid options (such as my back end USDJPY one touch), it gets pretty bad.

You're right !Well, I have seen USD/SGD DNT getting cheaper when one of the trigger (placed far away) is moved even further. So what you observed is not surprising when dealing with ccy pairs with large skews using a fudged replication method. Problems can start to surface when spot is very close or faraway from the barrier, especially when one try to "overfudge" a model that has obvious limitations. It is easy to miss some angles, as what you have just found. Besides, as more participants start to switch to stoch vol/local vol hybrid models for G7 currencies, fudging the 3 vanilla portfolio approach (even by artificial skew dynamics parameters) will become more and more difficult for high skew ccy pairs like USD/JPY. Have you attempt to build similar models yourself ?

Thanks guys.. cpengtoh, are you aware of any other (than Lipton's) article - I'm trying to understand this "method", but definitely need more detailed explanations/examples ?Regards,MaTT