Saturday, October 30, 2010

A market for political events








Future elections, political event probabilities are tradeable now as futures at Intrade. Of course like any other betting algorithms, a positive expectancy is required to make money over time. Finally, "hobbyist political analysts" like uncle Bob has the chance to back their opinions with some dough :)


Of course on the other hand, we could reference these numbers for a more precise forecast. After all, the market does not lie.
"Intrade makes more accurate predictions than I have ever come across."
John Stossel, 20/20 ABC



Looks like that semester of Political Science at junior college was worth it after all!

Friday, October 29, 2010

The "Hot Chicks" Indicator

This is according to google trends (see above). It appears that periods where Dow Jones exceeds Hot Chicks in relative search volume by a few sigmas (standard deviations), a bullish market likely follows, and vice versa.

Easy $$$!

Thursday, October 28, 2010

Implied correlation index KCJ


KCJ is the S&P500 implied correlation index, you can see it at yahoo finance, KCJ link.

It's pretty interesting as like the VIX, it has a negative correlation to the S&P500 index. What makes it cool is that as a correlation, it's implicitly bounded. It could definitely offer an edge for directional forecasts of both underlying markets and volatilities.

So much for words, look at the chart!

Wednesday, October 27, 2010

Intro to High Frequency Finance (Book Review)


Some of the folks at university probably think I'm crazy to spend time on all this non-school related material when final exams are next week... Any way, this was a pretty good, i.e. useful book for a couple of writers with backgrounds in academia. Some of the ideas were practical in real time trading for me, especially the stuff around conditional correlations, seasonal volatility, and rolling regressions.


The good

Lots of descriptive concepts around data analysis, modeling, and trading strategy development. Some of the ideas like normalizing returns via mapping operators to take care of the skew issue with correlations could definitely help improve the trading algorithm development process. The summarized stylized facts could help any new reader become familiar with general statistics of high frequency financial time series.


The less-than-good

It felt like they might have made some things a bit more complicated than necessary.

Over all assessment

It's got some good ideas.

Sunday, October 24, 2010

College degree, not really worth it


So I've been doing some work at a prop trading firm here in Auckland, dealing with mostly Asian index futures. There's some pretty cool folks at the firm, and the IT equipment is "hardcore" enough to pull off some of my strategies where optimal return comes from low-latency execution.

But what's really bothering me is the fact that 100% of what I do to make $$ these days requires skills that I could've gotten off a few books at the local bookshop and maybe a year or two of practical experience, all for $500~ tops! Yet, I've already paid over $100k in tuition for this Bachelor of Honours degree.

Sure, some people would say that "well it's a good piece of paper, it gets you a job interview (if your personal ventures fail miserably)". Still, there're plenty of people like me sitting on the employer side who KNOW that a university degree simply means one can show up on time and maybe type up a neat little report every half year or so, without any practical experience or ability to think critically.

So yeah, is it really worth it? I'd say not.