O
Outlander
To me this is the preferred solution - use the ratings to identify possible bets, but then look in more detail at the ones that have popped up. As an example, I use my 'selections' as the basis of my efforts, because they are all runners that managed their fastest rating (for the past several months) on their latest run, and they passed additional checks - a form check that requires they have placed recently, and a betting check that checks they are in the top of the market, both of these checks are based on the simple fact that most races are won by these runners and I think you need to be a lot better at studying form to select a likely winner from the outsiders in most fields! The third check is that the stable has not got a string of zeroes from its other runners in the past fortnight, which to me is an important check to make as even the best trainers send out runners when they have a health problem with their horses - they tend to realise all their runners are underperforming rather slowly, which is no consolation to the punter who has backed them.
Some of those additional filters, by cross checking with my top rated runners, turn out to be less important depending on type of race - so I have different filters for different race types Flat/Jumps handicap/non handicap. This is, at least currently, as far as I feel I can take the process of automating runner selection.
Before deciding a selection is a good enough prospect I then have to do a more hands on look at form and other factors - has the horse run recently (today's single selection, Mosala Dundalk 5.15) was a head second a month ago), has it ever won? To me this is quite important, horses aren't machines and how they behave once they get to the front varies - being herd animals I simply put this down to a few being 'leaders', out on the plains (or whatever) these guys would be at the front of the herd, deciding which way to go, while most are happier following. I am quite sure that some horses, having been persuaded by the animated chap on their back, get to the front and think 'crap! What the hell am I doing all the way up here at the front?' ... and then go on to ease off a bit no matter how busy the jockey gets.
I look at the opposition, what I'm hoping to find is that there's nothing looking very dangerous in there, if there are a couple of runners that look to be significant opposition then I start to wind my level of confidence a bit lower. Mosala has run a couple of decent seconds, only beaten a head last time out when headed towards the end - significantly Mosala rallied and made an effort to get back in front, and is now going over an extra quarter mile. To me that's positive, while the opposition doesn't look that tough. I think the biggest danger is likely to be from one of the less fancied runners, one of those with only the odd run under their belt - there are a few in there that have run very little, and a few stable changes - to me that looks more like a problem horse, perhaps some unsoundness that keeps them off the track, but that doesn't mean they can't be got ready for a once a year big effort, especially for a new stable.
I'm not tipping Mosala here, I'm just showing the kind of thought process I put into looking at what my program chucks out - I don't look at my ratings and think they're good or bad based on the overall win rate (that sort of info I put on my thread, it's so people can judge things for themselves, I don't consider it to be a make or break metric) - to my way of thinking the ratings will produce a few winners almost every day, all I need to do is find them! Instead of looking at 400 horses a day in 40 races, I'm looking at 40 horses that my added filters reduce to 0-5, and that I then need to look over to decide if they are worth betting on or not.
I dislike the poor data from NH racing, whilst at the same time I can't help but admit that my ratings seem to do at least as well for NH as flat. I'm not at all disillusioned with the ratings, I am disillusioned perhaps with the industry and regulators who don't think it's important to ensure data is as accurate as it can be.
There are lots of other ways to look at this - one that I do not do (I spend enough time on this already), but have sometimes wondered about is simply looking for races where the top 2 or 3 are decently ahead on points, then work through that top 3 to decide which of them is the most likely to live up to their rating on the day - clearly in most races the top rated isn't winning, so is there a pointer in there somewhere that will let me pick something lower down the ratings that has a better chance than the ratings initially suggest? Maybe it's possible to calculate a probability figure for each runner to suggest how likely they are to reproduce their best figure?
Dave
