CP Algorithym
 

I have been looking at yet another neural system but unfortunately kept no records because there was not much action. However, over the past week or so it has thrown up quite a few selections, and they have performed well.

For example, there were 3 selections yesterday:
Mackay R4 no.1 - second, paying $1.40
Seymour R5 no.2 - won, paying $3.60 and $1.40
Seymour R7 no.1 - won, paying $12.10 and $3.20. Amazingly, this was the pre-post fave.

The rules are:
1 ) All runners with at least three starts.
2 ) Must be the outright top selection with the neurals at default/factory settings.
3 ) Must be the outright top CP selection.
4 ) Must be the pre-post fave, second or third-fave. If under 8 runners then it must be the fave or second-fave.

I haven't checked selections that won their last two starts but I would omit them.

I've also looked at a similar method using the top selections with the JA, JT and $ algorithyms but from what I can remember the CP algorithym easily outperformed them.

As there are quite a few selections today I'll list them and see how they go...

Bendigo R2 no.3 - De Mars.
Ipswich R1 no.1 - Spectacular Pic
Ipswich R5 no.2 - Norse Code
Ipswich R6 no.1 - Simplest
Ipswich R7 no.5 - Jellicles
Ipswich R8 no.3 - Jackpot

A disappointing day yesterday. I'll give it another go today...

Rosehill R5 no.1 - Nuclear Sky
Eagle Farm R6 no.1 - Rampant Lion
Morphetville R5 no.1 - El Maze
Morphetville R7 no.6 - Magical Dane
Moonee Valley R3 no.8 - Sand Sweeper
Moonee Valley R6 no.3 - Innovation
Moonee Valley R7 no.1 - Gunfire Messiah
Benalla R1 no.8 - Moon Maiden
Gold Coast R4 no.2 - Showciti
Gold Coast R6 no.1 - Agrios
Toowoomba R1 no.1 - Golden Flute
Toowoomba R3 no.5 - Azzaland
Belmont R1 no.2 - Superscenic
Belmont R7 no.3 - Chachi.

Michael,
If you are looking for another CP plan, one may like to test this idea.

CP 3
DLR 3
PR 3

.Top selection Neural price must be $1.55.(maybe $1.60 but needs further research)
before scratchings.

.Must not have won its last 2 starts

.Selections are worked out before scratchings.

The idea here is that we are assuming a horse with rescent days racing should hold its peak current form longer than a horse which has had a longer break than average in days last start.

All selections can be worked out the night before.

Cheers.

Hi, Bhagwan.

It would seem that CF would be more appropriate than CP if we are looking at a horse with recent racing?

Michael.

Your right Michael, I made a typo.

Should read...
CF
DLR
PRC
All at 3

Cheers

I had a look at a similar method but this was before neural pricing existed on the website. The method was for horses whose last start was in the past 7 days and was also the top CF selection. Unfortunately it turned out to be very disappointing. However I might consider having a look at DLR with CP.

Bhagwan, I've just had a look at today's racing using the CP, $ and PR algorithyms, and also ignoring scratchings.

There were no selections priced at $1.55 or lower, but there were 4 odds-on selections. Three of them won for a return of $6.60 but only paid $3.80 for the Place.

There were 5 selections priced between $2.00 and $2.50 for one winner paying $2.40 but all placed for a return of $9.00.

As it looks promising I'll list the selections tomorrow limiting them to a maximum price of $2.45.

I have found that 1200m exactly worked best for runners in the last 7 days.
Overall there are not many bets once filters are placed over them.

Those figures of yours may bare some fruit.

The CF + DLR idea. had only 1 selection today Belmont 8/11 (Sat) for a $6.00/ 2.00 result NSW div but paid odds on in the other states, weird.

The CF + $ is still holding an impressive SR , it also shows how important shopping for best price is also, as shown in the above scenerio.

Cheers.

Bhagwan, I made an error in my previous posting - it is not the $ but the DLR algorithym.

Unfortunately Belmont R8 no.11 paid $1.80 and $1.20 here in NSW - it was a selection in our CF and $ system.

These are today's CP and DLR selections. I've omitted any selections that are returning from a spell.

Casterton R2 no.6 - Portland Prophet
Casterton R3 no.1 - Pasco
Casterton R4 no.6 - Jarraconi
Cranbourne R3 no.10 - Happy Honey
Hobart R1 no.16 - aaaaa's Jubilee
Hobart R3 no.6 - Special Resolution
Kalgoorlie R1 no.3 - Voodoo Rocking
Kalgoorlie R3 no.1 - Refal Raider
Kalgoorlie R4 no.5 - Daughter's Demand
Sun Coast R9 no.1 - Mystic Vibe
Tamworth R3 no.3 - Katobar
Tamworth 8 no.1 - Sir Winalot
Taree R4 no.1 - Barons Court

Even though yesterday was a losing day of $0.80 each on both win and place betting it did not fail.

There were 13 selections for 4 winners and 8 placegetters.

There is one selection today:
Mildura R1 no.4 - Samadji. ($1.40 pre-post).

Yesterday's sole selection won paying a disappointing $1.30 and $1.04.

Today's selections:
Kembla R1 no.1 - Atlantic Air
Kembla R2 no.5 - Elastane
Kembla R4 no.1 - Validated
Kyneton R3 no.4 - Road Novel
Muswellbrook R2 no.1 - Scotch Gem
Muswellbrook R4 no.6 - Ray.

I'm changing the rules by adding the DIST algorithym. Looking at the past three days this would have reduced the selections but both Win and Place brtting would be in profit, so I'll give it a go. If it fails its then back to the drawing board...

Today's selections:
Gosford R1 no.1 - Antrim Road.
Eagle Farn R8 no.6 - Ana Me Cori
Sandown R4 no.1 - Filou Shot

From yesterday's 3 selections there was one winner of $3.90 and two placegetters paying $3.00.

Today's selections:
Ballarat R1 no.1 - Bugatti Royale
Ballarat R3 no.1 - Belong To Peri.

After a scratching there was one selection yesterday - it won paying $3.60 and $2.30.

One selection today:
Goulburn R1 no.14 - Pure Nikki


The below would have been a selection but I have omitted it because it won its past two starts:
Pt Macquarie R4 no.2 - Amboise.

michael,


Just a suggestion when you are trialling systems. You need to take into account how likely the current results are due to luck. to do this simply just keep a record of the number of expected winners. This is easy to do as it is the sum of 1/price.

For example if you had 3 horses at odds at $3.00 ,$2.00 and $2.50 you would add up 1/3.00 + 1/2.00 + 1/2.50 = 1.23

This means 1.23 winners were expected in those 3 bets. If you do this for over 50 bets it will start to be more meaningful. You can then run a chi square test to work out if it is base don luck or not.

If you not sure how to run a chi square test or can't be bothered then a general principle is that you need at least 1.5 times winners then the expected winners. For example if the expected winners are 15 after 50 bets then you are doing ok if you have hit at least 23 winners in the 50 bets. This is just a rough guide and the chi square test is a better indicator.

The good thing about this is it helps remove faulty profits which are due to luck. For example after 50 bets I might be showing a profit of 10% ($55 returned) but if the return came from 2 winners ( $25 and $30 ) and the expected winners are 10 then I know that this profit is purely based on luck.

Hope that helps you out ...

Good Luck.

Hi, Wesmip1.

I may be wrong but the Chi method explained in your posting seems to refer to "good" luck.

If this is the case, how do you then factor in to a system where only bad luck beats your horse?

Maybe it comes down to the swings and roundabouts scenario, or am I reading it wrong?

Cheers.
Michael.

michael,

Without going into an essay on it I will try and explain it.

The chi square does not take into account just the winners it takes into account the number of losers as well. What I posted was a simple quick test (not the chi square test).

An easy way to do it has been posted by AngryPixie in a previous thread. Its easy to do in excel. In cell A1 put the number of winners, in cell A2 put the number of losers (not races but races minus winners), in cell A3 put the expected number of winners (sum of 1/price) and in cell A4 put the expected number of losers (races minus expected winners).

Then in another cell (lets say B1) put in the formula

=CHITEST(A1:A2,A3:A4)

It will come out with a probability for you. Most people adjust the prices to a true 100% market but it isn't necessary. You are looking for a number less than 1% in order to know whether the system is a winner or loser. Anything higher than 10% is regarded as being heavily affected by random noise (good or bad luck).

Good Luck.

Michael,
Here is an example to help you out that is familiar to you:

These are your selections up until you changed the selection criteria in this thread along with the prices from unitab.

Kembla R1 no.1 - Atlantic Air = $3.40
Kembla R2 no.5 - Elastane = $9.10
Kyneton R3 no.4 - Road Novel = $3.20
Muswellbrook R2 no.1 - Scotch Gem = $2.70
Muswellbrook R4 no.6 - Ray. = $4.20
Mildura R1 no.4 - Samadji. ($1.40 pre-post). = $1.30
Casterton R2 no.6 - Portland Prophet = $2.30
Casterton R3 no.1 - Pasco = $2.10
Casterton R4 no.6 - Jarraconi = $3.20
Cranbourne R3 no.10 - Happy Honey = $8.60
Hobart R1 no.16 - aaaaa's Jubilee = $6.80
Hobart R3 no.6 - Special Resolution = $3.10
Kalgoorlie R1 no.3 - Voodoo Rocking = $1.80
Kalgoorlie R4 no.5 - Daughter's Demand = $6.80
Sun Coast R9 no.1 - Mystic Vibe = $2.90
Tamworth R3 no.3 - Katobar = $7.00
Tamworth 8 no.1 - Sir Winalot = $3.60
Taree R4 no.1 - Barons Court = $1.70
Rosehill R5 no.1 - Nuclear Sky = $3.90
Eagle Farm R6 no.1 - Rampant Lion = $4.20
Morphetville R5 no.1 - El Maze = $4.50
Morphetville R7 no.6 - Magical Dane = $4.30
Moonee Valley R3 no.8 - Sand Sweeper = $2.40
Moonee Valley R6 no.3 - Innovation = $8.30
Moonee Valley R7 no.1 - Gunfire Messiah = $3.50
Benalla R1 no.8 - Moon Maiden = $3.80
Gold Coast R4 no.2 - Showciti = $2.30
Gold Coast R6 no.1 - Agrios = $6.20
Toowoomba R1 no.1 - Golden Flute = $2.00
Toowoomba R3 no.5 - Azzaland = $1.50
Belmont R1 no.2 - Superscenic = $1.60
Belmont R7 no.3 - Chachi. = $2.20
Bendigo R2 no.3 - De Mars. = $3.30
Ipswich R1 no.1 - Spectacular Pic = $7.60
Ipswich R5 no.2 - Norse Code = $3.30
Ipswich R6 no.1 - Simplest = $2.30
Ipswich R7 no.5 - Jellicles = $1.40
Ipswich R8 no.3 - Jackpot = $5.10
Mackay R4 no.1 - = $3.00
Seymour R5 no.2 - = $4.20
Seymour R7 no.1 - = $13.00

Total number of winners = 14
Expected winners (sum of 1/price) = 13.57

CHI SQUARE RESULT = 88% chance that the results were due to luck.

Based on that there is no reason to either keep or throw away the system Its still an unknown.

Hope that helps you out ....

Good Luck

I understand it now. It's quite intriguing.

Unfortunately I know little about computers and am unfamiliar with Excel even though I hear/read of it quite often.

If the system is still in profit after about 50 selections I will calculate it manually as shown in your previous post. I suppose it can also be applied to Place betting?

There are some websites out there that will calulate it for you:

http://www.graphpad.com/quickcalcs/chisquared1.cfm

Yes you can use it for the place as well.

Good Luck.

Thanks, Wesmio1 - I'll check it out.

Yesterday's sole selection placed paying $1.50.

I'm having problems accessing the neural website. However, I was able to look at a few meetings, and what I've got so far is:

Randwick R1 no.1 - Star Of Crown
Cheltenahm R1 no.1 - Lucky Fox
Hawkesbury R3 no.9 - Mon Choix.

One more selection:

Belmont R5 no.1 - Centrum.

A very bad day yesterday - from the 4 selections thre was one winner of $2.60 and one placegetter of $1.30 (2 selections ran fourth).

Today's selections:
Hamilton R2 no.3 - Judgement Justified
Hamilton R7 no.1 - Isenikenoff
Mornington R3 no.3 - Belong To Peri
Sun Coast R1 no.1 - Kirra Lyn
Devenport 3 no.1 - Concurrent
Kalgoorlie R6 no.7 - Daughter's Demand