Favourites Strike Rate per Venue
 

Ive attached a listing of the Favourites and the S/R per venue since 2002. I've only included the venues with 200 or more races

cheers

If you also add average field size for each track, you'll see a direct correlation.

So should one run a test of the strike rates of favourites per field sizes.
Then run the average field sizes per track.

No I don't think that's necessary, just average field size per track and add that to the csv output. With some exceptions, mostly there's a direct correlation.

Course details on this site.

http://www.racerate.com/Horse_Racin...es_SnapShot.htm

Vortech, do you have favourites broken down to classes eg mdns, etc.

Wet tracks eliminate a lot of favourites as does distance beyond 1400m.

Large fields also cause concern to favourites.

We need to know under what conditions favourites win

I can run a few tests tonight if you like.

Some of the data is slightly out as CP pointed out but over the 12 years you only really have a 0.03% variance I have found.

Do you want the tests done in isolation or with two filters combined?

I'd be surprised if distance over 1400m has too much effect, my guess/assumption would be that you generally find larger fields over that distance which would be the contributing factor to favourites winning less, not the distance itself.

Happy to stand corrected though.

What would be of interest is the following tests although it would be time consuming:

Favourite stats at 1400m based on size of field (i.e. 7,8,9,10,11,12,13,14 and 14+ runners)

Then do the same for 1600m and so on, I've suggested 14+ because i've found myself doing my ratings that 8-10 is the sweet spot for picking winners with ratings, 10-12 offers some value, 12-14 it gets quite difficult and 14+ poses to much potential for issues out of the jockey/trainers/horses control.

I have seen several studies which show favs have a higher hit rate on wet tracks. Probably because of smaller field sides I would guess.

SpeedyBen, Could it also possibly be that horses with proven wet track form and fitness get heavily backed into favouritsm and hence win quite a lot of races?

I guess the devil is in the detail and we'd need to see the stats. Personally I have been omitting any races classed as Heavy from my betting/laying lately as I don't believe in backing horses running on Heavy tracks in the summer season, whereas in the winter season horses are more prepped for it.

Just my personal thoughts not sure if others think the same.

Hi Eva.

I actually rang Greg Carpenter (Chief handicapper V.R.C.) on the matter of,
Of how he and his team split field size, and distances, when assessing form and weights a few months ago, he was only to happy to discuss the issues.
This is how they do it.

Field Size.

=<8=Small.
9-12= Medium
13=> Large.

Distance.

=<1300.
1301-1800
1801-2200
2201=>

I split all my ratings into these field/sizes and distances when and where applicable.

Hope this helps.

Cheers.
Garyf.

The problem with breaking the stats down too much is that your sample size becomes greatly diminished.

I agree with evajb001, Heavy tracks produce a high strike rate of favourites because proven wet performers get heavily backed, plus there are usually a wad of scratchings which increase their chances.

Average field size has a lot to do with the favourite strike rate at each track, although there are slight variances.

What is an advantage is watching for track downgradings during the meeting and dead3 or slow tracks as trainers often run the gauntlet anyway, where they would scratch if it were Heavy.

garyf,

The field size information is particularly helpful as I haven't really taken that into my account with my ratings yet. I think there's possibly some merit in adjusting staking based on field size but will have to run some tests once I have more data recorded.

In terms of the distances thats also very useful. I recently went through the process of giving my ratings an overhaul based on information i'd read on these forums. I've always put an emphasis on track conditions (Good/Dead/Slow/Heavy) but i've also combined this with distance of the race now to be more specific with my ratings. In particular I use this data for my fitness, barrier and turn variables the most. The distance splits I use in combination with track conditions are as follows:

<1101
1101-1201
1201-1401
1401-1601
1601-1801
1801-2201

For distances over 2201 i've found barrier/turn stats are negligble and apply my 1801-2201 variables for fitness.

CP,

Agree that breaking down stats too much can harm the sample size, i'd say everything still warrants investigation (if you have the time) but its important to be mindful of reading too much into small sample sizes.

In reference to track conditions changing throughout a day this is why my spreadsheet breaks my ratings down based on Good, Dead, Slow, Heavy. So that way if conditions change I can apply my ratings based on the given track conditions not just a generic rating for all conditions. Unfortunately this means I have to place my backs/lays manually before each race but I find doing it this way is more beneficial then a generic rating. Would be nice to find a bot that places bets based on track conditions.

You're on the right track Eva.


One huge thing with ratings be they free or your own is this.

Always split them into.

State.

Then whether the track in that state is.

Country.
Metropolitan.
Provincial.

This is the one biggest factor i have found in identifying,
In whether to bet or not.

With ratings it amazes me how what is good for say one state,
Is not for the other.

Today i will look at 3 venues divided like this.

Sale=(Vic Prov)
Goulburn= (Syd/Country)
R/Hampton=(Qld/Country).

This works a treat with mine, it may be different with other peoples ratings.

The order of your ratings will also come into play, with the above criteria,
As will minimum and maximum prices and the filters you posted.

Cheers.
Garyf.

Garyf, not sure if your happy to share or not but i'm interested in how you use state, metro, prov, country etc to aid in your process. Can discuss via email if you wish.

In terms of my fitness variable I haven't taken into account state etc, i've just used distance, conditions and runs this prep. However in terms of the barrier and turn variables i've used venue and distance.

For example all my barrier and turn stats for Flemington and specific to the stats recorded at flemington for distance ranges i posted above. If I don't have the stats recorded for a venue i've just been using the averages of all venues I do have recorded.

In my previous ventures with ratings I had noticed quite a difference between venues however I was recording all my bets in terms of tatts code (MR, VR, AR etc) so unfortunately for SA and WA a race at morphettville for example was recorded with the same code as a race at Port Lincoln. I'm not sure I can record it any other way with how my spreadsheet is formatted but as I said, still very interested in how you apply those details.

Here is an example of my format.

Victoria.

M/VALLEY
S/HILLSIDE
S/LAKESIDE
C/FIELD.
F/TON.

These tracks where my bets were, would be listed on my spreadsheet.
Next to those tracks would be a code.

Code= V/M meaning all these tracks were VICT-METRO.

Victoria.

BALL.
BEND.
CRAN.
GEEL+SYNTH.
WARRNAMBOOL.
MORN.
SEYM.
BENALLA.
KYNE.
MOE.
PAKEN
SALE
SW/HILL.
WERRIB.
Y/VALLEY
KILM.

These are Victorian Provincial tracks so would be coded on,
My spreadsheet as = V/P

Everything else in Victoria i class as Victoria Country = V/C.

These are your tracks like Ararat, Echuca, Wangaratta, Tatura, etc.

I do the same for SYD.ADEL,QLD. tracks.

I don't bet on Night Meetings, or P/Holiday meetings.

When i want to see how my ratings perform on,
Victoria Provincial tracks i just go to sort and filter code = V/P ,
And all those tracks appear i can then compare various filters to V/M or V/C etc.

You can put an insert column, mark it as code, and do the same thing,
Mine are these.

A/C
A/M
A/P

Q/C
Q/M
Q/P

S/C
S/M
S/P

V/C
V/M
V/P.

Hope this helps

Cheers.
Garyf.

What i would do then is something like this.


Based on my previous 2 posts.

Top rated =V/M
=<8 F/SIZE.
=<1300 DIST

Same top/rated.
9-12 F/SIZE
=<1300

Same top rated.
13=> F/SIZE
=<1300

You can then do a SYD/CNTRY using the same field/size distance etc.

Then experiment with the other distances field/sizes i mentioned earlier,
In different places .

This is only one set of filters, i use many others as well,
But i won't go to a filter until i have established a ,
STATE.
C-M-P first.

For me this makes the sorting out procedure easier.

You must have a selection(s) rating whatever it be 1st
For easier distribution.

Others may disagree, its what works for me, and has been for many years.

Cheers.
Garyf.

When testing with my previous posts you will find that,

Different sets of selections will deliver different strike/rates,
With different filters.

EG.

1-4 NEURALS.
1-4 D/SCOTT.
1-4 A.A.P PREPOST.
1-4 U/TAB RATINGS.

All these free ratings will throw up various strike/rates,
Profit/loss margins depending on the order of the selection(s)
And the filter(s) used.

Enough from me on this, someone elses turn now.

Cheers.
Garyf.

Yup that helps garyf, so over time you can see any trends and bet/not bet or stake accordingly.

I might start doing the same with regards to my ratings now, although would take some time before the data becomes meaningful. I guess it's important to record the actual conditions/variables of the race to use in filtering. Ratings take care of the horse/jockey/trainer side of things but have to be mindful of how well those ratings reply to specific venues, distances, conditions and field size.

Thanks for the reply and explanation.

EXACTLY.

Cheers.

I made a study a while ago over a period of 5 months...not much I know but was just testing out what someone wrote here that maiden favourites win quite a few of the first two races at a meeting. With a small number of filters, 1st-4th last start, within 5L of winner or won last start, started within one month of last start, age to 5yo, maidens and R65 races in New Zealand. Worked very well when using those filters and if the favourite did not qualify, the 2nd favourite was looked at for qualification. Where the dividends were too close to split (say 20c), both were backed.

Also noticed that the distance range 1950-2200m was running at 50% win strike rate.

[QUOTE=Vortech]I can run a few tests tonight if you like.

Well we triggered some good discussion mate.
Isolation would be great thanks.

Some of the data is slightly out as CP pointed out but over the 12 years you only really have a 0.03% variance I have found.

Do you want the tests done in isolation or with two filters combined?

Some more findings Kiwi

It was one line of data...

Was this for Favs over to show a profit of minus 13.39 overall.If so it is my belief if we could find the false favourites there is a profit to be made.

If you eliminated South Australia and Queensland would it make a difference.

Could you do Victoria Melbourne courses only please, I show a profit betting there, but not on favourites.

By knowing if the favourite cannot win we can make a profit opposing them which I do quite often.

I won on the first race at Flemington today purely on my perception of linebreeding even though the favourite had form at the distance.

The Excel doc has several sheets with the results per filter

Wow, found it, grateful thanks Vortech, this is hugely useful, appreciate your efforts.

Chas