4 races run at wyong & sandown, 2 at Bris......I make it 9 favs from 10.
Get stuck in !!!!!

And yes I would believe.

Wyong R5....a bolter......who would have thought.
Love it when a plan comes together.

To circumvent any misunderstanding, I won't be on Bris until R5.

Mumble mumble....something about horses and water..........mumble mumble.

I must admit, I never go on what happened in previous races, only previous days, and then only if it was a massive deviation from the average expectation.
However, the trouble comes when it goes on for days at a time, not so common here, but quite frequent in the UK.

Another note or debate on the side of this topic is in relation to confidence levels.

If you have a system with 50 selections and you manage to get 30 winners (60% SR) with an average return of $3.00 you have a profit of $40.00

Now using any interval calculator http://www.surveysystem.com/sscalc.htm you have an interval of 17.87. Meaning the Strike rate can range from 42.13% to 77.87% with a 99% confidence.

Based on this, at worst you have 42.13% strike rate @ $3.00 average winner is still a 26.39% profit.

Is this logical in racing terms?

I should have added the USA!!!
Been making very good money on both UK and AUS the last week, but USA has hit hard :(

Agreed, the US was very hard going this morning.
I worked 10 races for 5 favs, 3 2nd favs, a 3rd fav and 1 result.

Reckon you need a much bigger sample Vortech, to go forward with any confidence.

Today was not extraordinary for the USA, this has been going on steady for the last week. I wonder if it's the average small field size.
At this stage I'm praying for rain, as that's when we get the nice results in the mud there ;)

To give you some yardstick:
AUS profit 25.32 units after commission
USA loss of 14 units after commission

Head above water, but USA is definitely on a sustained run of short winners.

Chrome

What's the US liquidity like these days? Last time I looked it was pretty poor - similar to the NZ markets.

Can you make a decent buck there?

Isn't this why we use a confidence intervals.
The larger the sample the less the interval (+/-)

Liquidity is fine except for the outsiders, or those outside the top few prices.

Not too sure about any of that Vortech, never really looked at it - personally I don't proceed with any method/system until I have a substantial result set that passes a chi-test & plots a positive bell curve.

Do you have an example of how the chi-test works?

Here ...

http://www.propun.com.au/racing_for...74&postcount=30

Using my test sample

You have 30 winners from 50 selections

30 winners, 20 losers. Avg Dividend $4.00
Expected number of winners is 1/4 = 0.25
Normalise the market to from 115% is 0.25/1.15 = 0.21739

50 Selections so multiply by 0.21739 to get the expected winners

Gives 10.87 winners expected

Expected losers is 50 - 10.87 = 39.13

Observed winners = 30 A1
Observed losers = 20 B1
Expected winners = 10.87 A2
Expected losers = 39.13 B2

Put this into excel as CHITEST =CHITEST(A1:B1,A2:B2) on get something like 0.0000005%

Are these the chances of the system continuing? :) I'm very interested in hearing others thoughts

Even making the Avg Dividend like $3.00 more realistic is still giving a very very low %

No it's the chance that the observed results where not due to luck. Frankly though you don't need a Chi-Test to tell you that 60% winners at ave div of $4.00 isn't due to luck ;)

So what is a punter trying to achieve from these results. More confidence in a system long term with a higher % from the Chi-Test?

More confidence with a low %. The lower the % the less chance the results were due to luck.

All looks good Vortech...but I still think your sample size is way too small as imho you need at least 500 events to get any sense out of a chi-test.

Vortech, the Chi-Test is just a tool and shows you "what was". There's no promise of "what will be".

Norisk, you should read some Nassim Nicholas Taleb on the subject of normal distribution. He's reputed to have made a fortune from the GFC by pretty much throwing the Bell Curve in the bin.

cheers AngryPixie, will give it a look

Would add that I view all these tests etc just as a guide to get me going in the right direction, they certainly do not offer any sort of guarantee that one has nailed the holy grail or the like, as isnt the golden rule 'rubbish in, rubbish out'.

Use this link here, much quicker and easier.

Binomial Confidence Interval
http://statpages.org/confint.html

Will tell you your worst and best case expected SR from your current sample size.

Thanks Thorns.
Basically set to 99.9% confidence.
Take the min Strike rate and multiple by avg. dividend.