• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering. Learn more Join!
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job. Learn more Join!
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models. Learn more Join!

Short Term Stock Price Movement Prediction Competition

Joined
7/19/07
Messages
24
Points
13
Forward the post here as some of you guys may be interested.

Traders, analysts, investors and hedge funds are always looking for techniques to better predict stock price movements. The 2010 INFORMS Data Mining Contest takes aim at this goal, requiring participants to build models that predict the movement of stock prices over the next 60 minutes.

Knowing whether a stock will increase or decrease allows traders to make better investment decisions. Moreover, good predictive models allow traders to better understand what drives stock prices, supporting better risk management. The results of this contest could have a big impact on the finance industry.

Competitors will be provided with intraday trading data showing stock price movements at five minute intervals, sectoral data, economic data, experts' predictions and indices. We have provided a training database to allow participants to build their predictive models. Participants will submit their predictions for the test database (which doesn't include the variable being predicted). The public leaderboard will be calculated based on 10 per cent of the test dataset.


The submission deadline is October 10th 2010. Final results will be announced on October 12th. The winners of this contest will be honoured at a session of the INFORMS Annual Meeting in Austin-Texas (November 7-10).

Check the detail of this competition at http://kaggle.com/informs2010
 
I always wonder how one can predict the future with the database (history)

if there's really this kind of thing, can we predict the result of next round of a tennis game ? Can we predict the result of the next round on the gambling table ? that's interesting :P
 
Lun, here is news:
With the 2010 World Cup kicking off in less than four weeks, financial analysts at JP Morgan have crunched the numbers and come up with a bold conclusion – England will win football's biggest prize for the first time in 44 years.
"Having developed a rather successful Quant Model over the years, we intend to introduce it to our readers and also use its methodology to apply it to a fruitful field for statistics: Football and the World Cup," they wrote.
http://www.guardian.co.uk/business/2010/may/18/england-win-world-cup-jp-morgan
 
I miss one word

I always wonder how one can predict the future with the database (history) ACCURATELY

it's never difficult to predict
the difficulty is accuracy

can you predict that US comes first, England comes second, in group C ? I think, none of US ibank will predict this result.

what's the point to predict if the accuracy is not reliable ?
 
it depends on how you define the accuracy, no body can predict 100% confidently, what most statistical quants do is to increase the confidence level, 60% to 65% CI is a big improvement, but still, many of their predictions are not correct.
 
Lun, here is news:
With the 2010 World Cup kicking off in less than four weeks, financial analysts at JP Morgan have crunched the numbers and come up with a bold conclusion – England will win football's biggest prize for the first time in 44 years.
"Having developed a rather successful Quant Model over the years, we intend to introduce it to our readers and also use its methodology to apply it to a fruitful field for statistics: Football and the World Cup," they wrote.
http://www.guardian.co.uk/business/2010/may/18/england-win-world-cup-jp-morgan

That model is very flawed, it falls apart in game simulation, and they don't use the right stats as inputs to their model. Discussed here. :p

---------- Post added at 12:44 PM ---------- Previous post was at 12:42 PM ----------

what's the point to predict if the accuracy is not reliable ?

Because in between predictions and actual results there is a lot of money to be made. You don't have to be 100% correct, you just have to be getting the best of it.
 
I do have a question, many pricing models are the with assumption that it's a fair game, say BS model. If you can predict (at certain level), then it's no longer a fair game, then all these pricing models will collapse or we have to rebuild these pricing models with new assumptions, right ?
 
I do have a question, many pricing models are the with assumption that it's a fair game, say BS model. If you can predict (at certain level), then it's no longer a fair game, then all these pricing models will collapse or we have to rebuild these pricing models with new assumptions, right ?

By BS model, I think you are referring to Geometric Brownian motion SDE for the underlying (e.g. stock). The pricing is built on top of notions of risk-neutrality/martingale measure. None of them require equal probability for up/down movements. In other words, there is no assumption of "fair game".
 
for martingale

E[Sj | Si] = Si where j > i

but, with prediction (successful rate <> 50%), expectation is no longer unchanged
 
Back
Top