The dangers of inefficient programming

[Justin7]

When I write modeling programs quickly, I don't worry about speed. Most of today's computers can churn through any data, no matter how poorly the code is written.

I'm testing that premise. I now have an "X^3" algorithm, and my game sample space is about 10k games. I wish I had paid a bit more attention to programming at the start, instead of having to redo it midway through.

[acw]

KISS (Keep It Super Simple)

Any interesting factor that came out of the X^3 ?

http://forum.sbrforum.com/players-talk/87462-me-usa-mls-fields-looks-lot-smaller-than-european-fields.html

This is the last time I gave a new factor a go just to discover that it was again useless.

And trust me, no matter how good your model is, you will always make more money if you can find guys that will bet into you at 9/10 on events that are widely available at Even Money than you will ever make betting yourself.

Anyhow Good Luck!

[Red_Sux]

econometrics is mostly stats and regression. at one point i was thinking about using arima model to predict a slump in mlb. never got anywhere because it all boils down to curve fitting.

and this got me thinking. what ever model you come up with...either based on theoretical or observational assumptions are good but how can you be confident on predicting results. backtesting might give you really good results, but it might just be an illusion when you optimizing the model that you're actually data mining. for example, you come up with a model that only tend to work with certain series of historical data. live testing takes time, unless you're one hardcore pro who got a pipeline of models to test, no one has the patience to do this. at the end you come up with a model that does great in backtesting, but how confident are you putting couple of cents on its calculations? things can change in a hurry...your model become unprofitable.

it takes a special personality to become a winning gambler. people who come up with models and good with numbers doesn't always end up as successful gamblers because they might be too risk adverse. on the other end you have patty who is as risk loving as you can get. to be a winning gambler you gotta have some good intuition about all the factors but not too boggled down with numbers. -6.7 pts vs -8.5...****, players are human, not computers. getting too mathy is not always a good trait. you need the balls to pull the trigger at the right moment. models are nice, but your final analysis comes from your head, not a computer print out. you decide based on experience and intuition plus guidance with model outputs. every gambler is different, so every successful gambler will have their own unique heuristic process of determing betting options. this is okay because there is more than one way to win money, as long as it works, stick to it.

[acw]

every gambler is different, so every successful gambler will have their own unique heuristic process of determing betting options. this is okay because there is more than one way to win money, as long as it works, stick to it.

Imagine the massive advantage that you have when combining those!