Analysis and Development

Prediction models: up to 40% accuracy for stocks

The below is a simple model of a prediction dashboard. This skeleton and model could be used for a number of exercises such as stocks (as below), CPC, CPM, energy prices. This is a minor skeleton that should be built out with an exceeding number of parameters depending on the subject.

The source code can be found here.

Prediction model: Applying to Sports

Building on the above model, I have worked to create one whereby different parameters of race horses are using to give prediction of racing positions. This came out at 25% accuracy. I will continue my work on this via github and my personal blog

The first race I used to test this model was 6th Jan, Cork 13:57 .

Abbeydale: The average normalized position is 0.271429, suggesting that, on average, Abbeydale tends to finish in a position better than approximately 27% of the total number of horses in the race.

Cato's Revenge: The average normalized position is 0.076923, indicating that, on average, Cato's Revenge tends to finish in a position better than approximately 7.7% of the total number of horses in the race.

Missiee: The average normalized position is 0.266667, suggesting that, on average, Missiee tends to finish in a position better than approximately 26.7% of the total number of horses in the race.

Rebel Early: The average normalized position is 1.0, meaning that, on average, Rebel Early tends to finish in the exact middle position relative to the total number of horses in the race.

Query V race results:

2 out of my top 4 came into the top 3. The winner on the day did not feature in my results.

Additional testing (26th Jan)


15:00 Cheltenham,
Saturday 27 January 2024

Here looking at a more specific distance and courses type containing ‘soft’ ground not just ‘soft’ ground.


Document Templates.org

A space to host free templates for everyday use. I find there’s a lack of simple document templates for folks to follow.
As well as templates, I have created a prioritisation tool, whereby anyone can pop in a list of their tasks and the tool will give them the priority they need to follow

Year
2022 -2023