Wind Surgery

0 votes
Just wondered what effect this has on the horse and if it a long or short term i.e. should I be weighting the effects closer to the operation date and then gradually easing off? I've read that is can actually have a negative effect when horses get older so a simple linear curve wont really cut it.
asked Oct 17, 2020 by DataScientist Handicapper (820 points)

1 Answer

0 votes
The jury is still out on exactly what effect wind surgery has - some comprehensive analysis needs to be done.   I suspect age of horse when wind surgery is first performed, number of previous runs, number of interventions, racing code etc. will all play a part in this.   We've been so busy collecting the data that we haven't yet put any time into analysis(!) - we have also built some potentially useful derived variables on wind surgery into the new historic_runners_insights table, available to all subscribers here:

Note in particular prev_wind_ops, the description for which is as follows:

the daily_runners and historic_runners table now contains a field for wind_surgery_declared. This field only contains an entry for the run immediately following any sort of wind_surgery. However, on any subsequent run where the horse did not have wind surgery immediately prior to that run and post its last run, there will be no evidence that a horse has previously had wind surgery without querying back in time for the horse's history. This field provides a ready solution by showing whether or not a horse has had any wind surgery at any point in its career, and how many different surgeries have been performed. Thus it shows 0,1,2 etc for each horse

If you are inclined to analyse the effect of this would be great to publish as a blog post, let us know.
answered Oct 21, 2020 by colin Guru (14,380 points)
Thanks I will check out the insights table.
I had a quick look and tried a few different things in my model, simply flagging it as a binary variable, I tried weighting it by days since the op, both in terms of more weighting closer and more weighting further away but in all cases it added no additional predictive power.