Brumby supplied the (potentially) useful stat: 73% of winners are within four zip points of the starred zip horse (in the Sportsman Chartform).

How do we make this figure useful? Brumby suggests perhaps we set a limit (percent) of runners that can be within 4 points of the top zip. So, for example:

Race 6 at Cheltenham last week.
Top zip (starred) = 56
Four from the star = 52
Runners within range 56-52 = 8
Number of runners = 15
i.e. 8/15 runners are in that range of those horses winning 73% of races.

In theory if there are too many such horses in a race it will be a tough one to pick and conversely if there are few of them it will be compasratively easier.

Baghwan has just pointed ouit how often favourites get beaten. Actually I think its about 70% of the time, isn't it? If we are looking for horses likely to beat the favourite look to that top group of horses within four of the top zip. (The top zip is often but not always the favourite. A good bet when it is not.)

BUT, races don't always go to plan. Because the other significant stat here is that in 27% of races winners were beyond a range of four points from top zip. These are the cases where, in review, you can't find anything in the form guide that mighgt have given you a clue that this horse would win. As Baghwan says in his posts, form will usually lead the experts to the same conclusion, with no surprises. But in fact races are full of surprises. The problem becomes:

in races with a small percentage of the field in the top four points, there is a corresponding larger percentage of the field in that group of horses causing upsets 27% of the time. The fewer the class runners the more likely a roughie will get home. Isnt that right?

So might this be a measure of what races are likely to go to roughies? If you have a field of, say, 15 but only three horses are within four points of the top zip, isn't it more likely a roughie will win, simply because there is proportionately more of them? Its an inherent problem. Races with few decent runners are easier to pick, but only up to a point, because there are also more dark horses to cause an upset. And upsets happen in what, at least 20% of all races just because the universe is made that way and no amount of form analysis can change it?

In race selection, how can we try to predict races likely to go to roughies? It doesn't take much skill to calculate the "best" horse in a race, but the "best" horse doesn't always win.

Perhaps I'm fumbling here for a system of grading races into:

*Those likely to go to the favourites
*Those likely to go to a non-favourite among the top group of horses
*Those likely to go to an outsider.

All suggestions welcome.

Brumby? How often do starred zip horses win? Better than 30%. I'm interested in any stats on the zip rating's performance. Thanks.

Cheers