Hi All,
We’re going to be pausing our betting for a few days…
In addition to a terrible set of results in the last few days (the stop loss really helped), we have also noticed a problem in the data we produce for predicting winners.
There is a pattern emerging in the data, which we have seen for the last few years, at the beginning of the flat racing season (April). This year, we decided to start applying a more data driven approach to this, we have recently started monitoring the “quality” of the 100+ predictors we use for each runner.
We now tracking the ratio of predictor values that are either “Empty” or “Zero”. This monitoring serves 2 purposes:
a) Empty values are indicative of a potential bug,
b) A high ratio of Zero values tells us that something is changing with the current crop of horses.
We have noticed the predictor called “WinRate_Runner_2”, which measures, for each horse, the number of races it has won in the last 2 months. This value is usually 0 (zero) for about 85% of runners. As of today, 93% of these values are zero. This tells us that the current crop of horses have not run recently, we assume rested for the winter season.
In other words, until this value improves, when the current crop of runners has at least a second race, we don’t expect the results to stabilise.
Will update you again, probably in a week, maybe 2 weeks.
Cheers
Nick