Testing a Neural System

[michaelg]

Since Monday I've been looking at a neural system using the default settings.

The selection is the horse with the highest points at each venue. There have been no filters.

Today's selections are:

Wellington 3/7
Toowoomba 1/7
Rosehill 2/1
Caulfield 5/7
Dombeen 1/3
Morphetville 5/1
Colac 2/1
Gold Coast 1/3
Goulburn 6/3
Newcastle 7/14
Ascot 4/1

Yesterday was the only losing day for Win betting (lost $2.30), but it is still providing an overall profit. The highest divvy was $12.00.

It has also shown a profit with Place betting. It has won all five days. The results are based on NSW TAB divvies.

We'll see how it goes today.

[michaelg]

Not a losing day yesterday, but the profit was minimal - $0.80 for win and $0.30 for place betting.

Today's selections:

Cranbourne 1/2
Gawler 2/6
Geraldton 5/2
Hobart 5/2
Sun Coast 8/9
Wangaratta 3/4
Warwick Farm 6/7
Wellington 8/8

[The Ocho]

Interesting michaelg. I suppose Saturdays may be the hardest for the Neurals as there are a better class of horse competing against each other. Have any been really short priced apart from Black Caviar or are they pretty reasonable prices?

[michaelg]

I wasn't too unhappy about yesterday's results because I believe Saturday is probably the most inconsistent day for the neurals.

It would be expected that the selections would generally be the shortest fave for each meeting. In the testing period there were 31 selections for 17 winners. 7 of these have been odds-on. The highest winner has been $12.00, and the highest placegetter has been $5.40. So the prices of the selections haven't been too bad.

Today at Warwick Farm it would be expected that Hay List which is $1.60 in the Tele would be the selection, but its R6 no.7 at $4.40...???

[jose]

Michaelg, I have

WFarm R5-1 @ 233.3 points
WFarm R6-7 @ 163.3 points

???????????????

[michaelg]

Jose, its not uncommon for the neurals to change after the first race. I think this has happened because race 5/1 now has 273, while 6/7 has 270 points.

I double checked it this morning because I was surprised Hay List was not the selection.

[jose]

Mine still has my original figures from the previous post?
Beats me.

The reason I noticed is that I have been looking at something very similar.
The selections are not the same as yours but are very close.
From yesterday I had 8 bets for 3 winners and 6 placegetters.

8/3 & 6. $10:30 & $9:10.

Bit of a worry when the #'s change tho!

[michaelg]

Jose, the points quoted in my last post are also unchanged.

I have all the neural categories on a setting of 3 except WT which is on zero.

[jose]

That's the difference!

Today I have:

Hobart 5-2
WF 5-1
Wang 7-3
Gaw 8-6
Cran 9-3
Well 8-8
Sun Cst 8-9

So far so good, although no fancy prices amongst them.

[jose]

Even sort of day for me today.
No real damage tho as I tend to bet 1 x 4. Helps to overcome the shortish prices.
Great day for your stuff tho.

[michaelg]

Yes, but not with Place betting. There were 6 placegetters from the 8 selections for a loss of $0.20. It's the first day it has lost with Place betting.

But a very good day with Win betting - there were 6 winners for a profit of $8.50.

[jose]

On Unitab you would have made a $0:30 profit just betting for the place. What TAB are you using?
At 1 x 4 you would have made $9:70 profit on Unitab.
These are for a $1 stake.
I know the staking argument has been done to death, but I still maintain that staking 1 x 4 is the way to go if your selections are on the short side.
Try looking back thru your results and see if it helps, I suspect it will.

[michaelg]

I bet with NSW TAB. I think the difference between the TABs isn't too great, except for the NSW TAB with very short Place divvies because of their method of calculating the dividends. The NSW TAB take-out is marginally higher than Unitab which can adversely affect the very short divvies.

I'm not a real fan of your staking plan. Yes, it will have a higher Place profit than 1 unit, in fact fourfold, but only if the method is showing a profit at level stakes. However, if the method is showing a loss at level stakes, then the loss will be fourfold.

However, if Place betting is providing a profit I would prefer your staking plan because if the Win component shows a loss for a particular day, then the Place bet could give an overall profit for that day.

[michaelg]

Today's selections are:

Taree 4/5
Ararat 3/5

[michaelg]

Yesterday's two selections placed for a loss of $2.00 on Win but a profit of $1.00 on Place betting.

Today's selections are:

Sale 7/3
Scone 2/2
Kalgoorlie 5/2