Friday, May 22, 2009

Why my betting appliction didn't work out well

For the last two years I was working on online betting predictions. First I started with simple laying on horses with relatively high odd, bellow 1.6. The strategy behind was simple and can be found on various websites which are revealing so called "holly grail". The horses which are potential winners are often not winning, so laying on them brings nice profit. But it is not that easy and I was trying to use my statistical and other knowledge to build up an application, which would combine three major parameters: current odd (later on when Betfair added near price prediction, I used that odd), intelligence information (website which was predicting winners) and history results. First notion in my blog about this activity was in post Free Forex and betting signal over Twitter.

System was great when it got enough historical cases, i.e. around 100 events. But there was a little problem... Based on assumption, that I lay on odds 1.6 and lower, I ended laying on horses in interval of 2.5. - 3.1 (which was flexible of course, since my application was learning from new data). This interval is no longer high profit and my predictions seems to be very accurate. But if I missed it, there was major loss. This was no go, and after several months of posting my prediction on Twitter, I closed this story.


And went to next challenge, which was Premiership soccer games. Since I started it some months ago, it still didn't got enough data, but I got some very good results out of it. My strategy is betting on matches to have Over 2.5 goals. A very popular option of betting, since it doesn't matter who wins and scores, but only to have 3 goals on the single match. And surprisingly it happens many times, and with a little help of my application, results were even more accurate. So for next season I plan to rebuild my prediction application and get really profit of it. I am optimistic, as results shows high correlations with my signal, also posted on my Twitter page.

Lesson learned: methods and ways to make money on Internet are numerous. But everything must be tested first. And maybe most important thing: set your goals, and don't change them in the middle of a project. If you get bet results, try something else, from scratch.