Does anyone know?

[King Mouse]

Does anyone know what % of winners finished within four lengths of the winner at their last start?

Also does anyone know what % of winners has their last race 2 weeks earlier?

[dinodog]

hi

according to my database and reasonably limited questioning ability, the following are my results for your query

winners within 4 lengths of their last start winner number 1965 from 2457 races, their were 18075 contenders. this sample is over sr mr and br saturday only and represents about 80%

win quinella trifecta and place all showed a lot

so far as % of winners at exactly 14 days for last start within 4 lengths we have 1927 races , 623 winners, 5165 contestants with an across the board lot and a strike rate of 32%

these two criteria obviously require further filtering

regards

dinodog

[Bhagwan]

1-7 days 31.0% win
1-14 days 67.5% win

4 lenghths last start & less that win 64%

[hermes]

But what proportion of all runners started last 7 days or 14 days? Is it a characteristic of winners?

hermes

[Bhagwan]

It`s a broad stat based on 20`000 Australian races right accross the board.

[King Mouse]

Thankyou. Could be the basis of a decent system there.

[TheDuck]

I've found that 47% of all statistics are made up on the spot.

:smile:

[hermes]

Duck,

There are many sets of stats around and they don't always agree, do they? I compile my own to compare - and to get insights into how its done and what its about.

I wonder about excessive minutae in statistics. The more I deal with stats the more I round things out to whole numbers and just look at *proportions*, not percentages. When you get a figure like 67.5% winners started last start 1-14 days, there is nothing at all significant about the exact figure. Round it off to 7/10. The exact figure will hover around there, near enough. No need to put on a white lab coat and be all scientific about it. Rough stats will do. Its the broad ratios and proportions you need to build from.

Exact figures contribute to the scientific mystique of stats when, as Homer Simpson pointed out, 52% of people know they're rubbish.

(However, he also said, "The flu on the wekend. What are the chances? A million to one!")

Hermes

[hermes]

I feel much the same about the size of samples too. Obviously in stats sample size is important, and there's the Law of Large Numbers etc, but working back through a pile of old form guides will tell you sure enough that about a third of favourites get up. Doesn't require a sample of 20,000.

This is not to devalue the stats people kindly post on this forum, just to suggest that punters compile some stats of their own, looking for rough proportions.

How many winners started 1-7 days? A stack of old form guides might tell you 35.7% and a sample of a squillion races might tell you 30.2%. Either way its about a third.

Hermes