Paul Wilmott introduces quantitative finance

Bits of it.
Good book, obviously advanced stuff, arb delta hedging etc.

Random behaviour of assets is a good chapter.
I'm sure most people are unaware that quant finance has randomness
as its foundation...:LOL:
 
Bits of it.
Good book, obviously advanced stuff, arb delta hedging etc.

Random behaviour of assets is a good chapter.
I'm sure most people are unaware that quant finance has randomness
as its foundation...:LOL:

Well the key idea behind black scholes and the almighty can of worms that opened was that investing in randomness can eliminate randomness (sort of). Of course, these ideas are never really that new.
 
Well the key idea behind black scholes and the almighty can of worms that opened was that investing in randomness can eliminate randomness (sort of). Of course, these ideas are never really that new.

Yeah true, it has been said before that the apparent black arts of quant finance
are not as far away as you would think from the methods you could employ using
more conventional means...
A philosophy I completely agree with :)
 
'Derivatives **** ups', has to be the best named chapter in any quant finance book I've read.

:cheesy:
Highlights the dangers of assuming things will always stay the same very well.
That danger faces all of us in one form or another.

Reliant on low costs - screwed when they increase.
Reliant on liquidity - screwed when it disappears.
Reliant on current margin rates - screwed when they increase.

Thats the problem we all face, you either try and develop a model that is robust
enough to weather the storm, or can adapt, or just kill it and implement
a different approach that suits the conditions.

Being able to change and adapt is a nice luxury.
The MGRM example in the book is a case in point.
They panicked over the increasing futures margin.
Ultimately that panic (or inability to meet the margin)
lead to the bigger losses when they were forced to close the OTC forwards.
Really they had not priced in the cost of carry.

Sounds obvious, but its all too easy too assume financing costs,
or a cost structure, you are reliant on will always be priced as is, or lower.

No revelation there though, those are the same principles that create and fuel
every bubble.
Human greed and the rush to gorge in favourable conditions.
The assumption that conditions and costs will not change,
despite the fact that history shows things always change sooner or later.
I suppose it boils down to adapt or die with the caveat that you also need to
be able to survive the apadtion process.
 
:cheesy:
...
Thats the problem we all face, you either try and develop a model that is robust enough to weather the storm, or can adapt, or just kill it and implement a different approach that suits the conditions.
...

Good post. Which would you recommend, weather, adapt or kill? I suspect your answer might be along the lines of 'it depends'. I'm usually in favour of temporary kill, because I don't like trading when I have that unsure feeling, so it's psychologically easier to kill for a day or two while I do some analysis. This is a difficult thing (for me) to decide which is best. Drawdowns happen, but they're often just a heads up something might be wrong, rather than a reason to do a great deal about it. Added to that, is that I am not automated, part discretionary, which means I also have to evaluate after every day, how well I traded and to what extent that and discretion caused the results. Not easy.

If you use a statistical property of the market, I believe you have to also have a good concept of why it is there in the first place, or else how do you determine when that property might disappear or even capitalise on it fully? It helps to have a good idea a priori of when it will likely struggle. Quantitative methods won't really reveal the reasons behind, but could set you on the right track.

So what method do you use to decide the edge has stopped working or needs to be adapted? Over what period of data? When is it just wrong, and can't be adapted? If you have a great method for this it is almost an edge in itself. BBMac wrote about knowing your edge and knowing how many losses you should expect and so on. This is a good start, but this could lead you to the wrong conclusion. I'd like a more sophisticated and accurate method
 
Last edited:
Good post. Which would you recommend, weather, adapt or kill? I suspect your answer might be along the lines of 'it depends'. I'm usually in favour of temporary kill, because I don't like trading when I have that unsure feeling. This is a difficult thing (for me) to decide which to do. Drawdowns and runs of losses happen, but they're often just a heads up something might be wrong, rather than a reason to do a great deal about it. Added to that, is that I am not automated, part discretionary, which means I also have to evaluate after every day, how well I traded and to what extent that and discretion caused the results. Not easy.

If you use a statistical property of the market, I believe you have to also have a good concept of why it is there in the first place, or else how do you determine when that property might disappear or even capitalise on it fully? It helps to have a good idea a priori of when it will likely struggle. Quantitative methods won't really reveal the reasons behind, but could set you on the right track.

So what method do you use to decide the edge has stopped working or needs to be adapted? BBMac wrote about knowing your edge and knowing how many losses you should expect and so on. This is a good start, but this could lead you to the wrong conclusion. I once had a method that was doing very well, adn I was excited about it. Came to November, then December and the results were not looking nearly so good. Looked at the stats towards the end of Dec, and decided I didn't like it so much any more. January was a great month ... for the method.

Thats the $64m question.
I suppose as you say it depends...:LOL:
Seriously though, I know exactly what you mean.
I was expecting the recent conditions to be less than favourable,
my algo has had a good run though.

I'd say a drawdown limit, monitoring range and volatility as well.
I'm aware of the conditions that may cause problems.
As I say I had expected May or June to be a potential bad month, so far it has been joint best month.

I think the primary reason is simple levels of fear, Greek election etc. was a
known upcoming event, so was priced in a more orderly fashion,
not as much panic as sep '11 (greater volatility & range).

Volatile panic stricken one way markets are my cue to sit up and take note.
Also, volatility drying up, or a continual bull or bear might knock things out.
I say that as I have not tested or run under those conditions.
So it stands to reason they could potentially be a cause of failure.
My data goes back as far as 2003.
 
Last edited:
Just realised I didn't actually answer the question :innocent:
I would aim to weather the potential storm, but keep a very close eye
on DD, range and volatility.
Or better still, pull the plug when the BBC roll out Robert Peston on the 6 'o' clock :)

------------------------------------------------------------------------------------------------------------
EDIT:
Oh and BTW, I don't recommend anything, just my view.
Everyone makes their own choices :)
 
Last edited:
Just realised I didn't actually answer the question :innocent:
I would aim to weather the potential storm, but keep a very close eye
on DD, range and volatility.
Or better still, pull the plug when the BBC roll out Robert Peston on the 6 'o' clock :)

By the time he's finished explaining anything....the next event is well under way :LOL:
 
Top