|| Data Research
I started JCP 7/2012..
Is it better to use ALL the data I've downloaded or just use full years of 2013 & 2014 ?
I am still a “rookie” with JC but my advice would be to develop your UDM’s in the 2013 database and validate them in the 2014 database. I hope that helps
|I find that that the game itself - or the game broken down into its individual aspects... public betting trends, track profile from one meet to the next, abilities of riders and trainers, tendencies passed down by specific sires and dams, etc., - is something that is constantly/slowly evolving.|
All UDMs and UPR/UserFactors have a shelf life of unknown duration beforehand.
Put another way:
Your 'work' ( UDMs, UPR, and UserFactors, etc.) are not something to be created once and used 'as is' forever.
I am convinced that as the game evolves the player whose 'work' evolves along with the game will likely outperform players whose 'work' remains static.
I find that as the game evolves older data (eventually) becomes stale.
There are always exceptions to hard and fast 'rules' but:
For the most part I have found that data older than about 18 months to be 'stale.'
I have found that using a rolling 12-14 months of data for the 'big picture' part of my modeling to be quite effective.
I have also found that merging a rolling 3-6 months of recent data into my overall process to be an effective way of staying on top of (and being in a position to exploit) new public betting trends as they emerge.
|thxs for the advice :)|