Punting 1.01

[curious]

6 minutes ago, mardigras said:

In NZ/OZ, I tend to look at the horses history to identify how they go about getting the level of fitness to be where they want it. It's not ideal.

It's not just fitness either and I've done it myself here. But where you have a niggly difficult to ascertain soundness problem it's tempting to use races as a test of the current hypothesis and solution.

[mardigras]

Just now, curious said:

It's not just fitness either and I've done it myself here. But where you have a niggly difficult to ascertain soundness problem it's tempting to use races as a test of the current hypothesis and solution.

Brings in the old 'gear change' scenario - to test something. But without knowing what the result will actually be. I've read some punters believe so strongly in some of these gear changes, like blinkers on, they add 20% regardless of whether the individual horse might be better suited to them or not.

[hesi]

I read some data somewhere, that said the optimum field size for highest turnover per runner , was 14.

I think it was in an NZTR Annual report a while back

Might have been a NZRB one

Edited December 26, 2018 by hesi

[curious]

9 minutes ago, hesi said:

I read some data somewhere, that said the optimum field size for highest turnover per runner , was 14.

I think it was in an NZTR Annual report a while back

Might have been a NZRB one

Not so last I looked. Think 7 was the highest per runner. Not saying that is cost effective but 2 x 7 horse fields produce considerably higher turnover than a 14 horse field.

[curious]

1 hour ago, mardigras said:

The only part I've considered not using a normal distribution is when I model the times. Rather than have a mean where 50% of the performances are faster than the mean and 50% are slower then the mean (excluding those that equal the mean), when the confidence is lower, I've been thinking I'd prefer the high and low times to be the same (i.e. +/- 3 sd of the normal mean), but where the resulting mean time of the distribution is higher than the normally distributed mean. A slightly weighted distribution based on my lower confidence under the race conditions. 

Another factor I use for confidence is on race starts. Fewer starts reduces confidence, as there is less ability to cater for outlier type performance data.

OK. I had to read this post a few times to grasp it but I think I'm getting the issue now. What you are thinking you might want to use is not a normal distribution centred on the mean but a distribution centred on the midpoint of the range (excluding outliers)?

[mardigras]

16 minutes ago, hesi said:

I read some data somewhere, that said the optimum field size for highest turnover per runner , was 14.

I think it was in an NZTR Annual report a while back

Might have been a NZRB one

Here is the detail on pool per runner (This is a little old - for 2014/2015). 

starters	races	total pool  (single race)	avg pool	avg per runner
4	33	$1,123,351.00	$34,040.94	$8,510.23
5	131	$6,487,063.00	$49,519.56	$9,903.91
6	251	$15,761,478.00	$62,794.73	$10,465.79
7	450	$35,986,090.00	$79,969.09	$11,424.16
8	587	$55,517,087.00	$94,577.66	$11,822.21
9	675	$74,329,757.00	$110,118.16	$12,235.35
10	687	$79,447,735.00	$115,644.45	$11,564.44
11	555	$68,526,912.00	$123,471.91	$11,224.72
12	582	$78,999,736.00	$135,738.38	$11,311.53
13	420	$61,955,404.00	$147,512.87	$11,347.14
14	395	$63,307,475.00	$160,272.09	$11,448.01
15	145	$29,446,177.00	$203,077.08	$13,538.47
16	144	$30,102,679.00	$209,046.38	$13,065.40
17	17	$5,287,709.00	$311,041.71	$18,296.57
18	35	$13,897,685.00	$397,076.71	$22,059.82

 

This does not consider the individual races and race quality. Even though 15 - 18 are the highest per runner, I'd expect they are also inclusive of some of the larger betting races such as major cup meetings etc. I don't see much evidence of 14 starters being optimum.

[mardigras]

4 minutes ago, curious said:

OK. I had to read this post a few times to grasp it but I think I'm getting the issue now. What you are thinking you might want to use is not a normal distribution centred on the mean but a distribution centred on the midpoint of the range (excluding outliers)?

If I'm reading that right, yes, the midpoint of the distribution being the same, but a higher percentage of the distribution being on one side of that centre than what would occur for a normal distribution. I probably have to read your sentence a few times as well  I'll try a chart.

[curious]

ok so you are maybe saying a non-normal distribution but still centred on the mean with the tail on one side longer than that on the other?

Edited December 26, 2018 by curious

[mardigras]

1 minute ago, curious said:

ok so you are maybe saying a non-normal distribution but still centred on the mean with the tail on one side longer than that on the other?

Yep - because I have less confidence that the centre is the true centre distribution wise (if that makes sense). Sort of pulling the centre of the distribution slightly sideways, without changing the end points. For when my confidence is lower.

[curious]

22 minutes ago, mardigras said:

Yep - because I have less confidence that the centre is the true centre distribution wise (if that makes sense). Sort of pulling the centre of the distribution slightly sideways, without changing the end points. For when my confidence is lower.

ok, so not centred on the mean. A kinda slightly skewed distribution but centred on what?

[mardigras]

13 minutes ago, curious said:

ok, so not centred on the mean. A kinda slightly skewed distribution but centred on what?

Ultimately centred on a point that reflects the confidence - but still not centred as in a normal distribution, but skewed as you mention. A bit like pulling the centre slightly left or right depending on confidence. I don't even know if it would alter anything, but was keen to try it if I could work out how to do it.

Edited December 26, 2018 by mardigras

[FeelTheFear]

Still chewing through all this wonderful information. I will never reach the level of sophistication with regards to your analysis mardigras but trying to settle for something less complicated but still meaningful. 

Curious and mardigras, you are right. NZ races are difficult due to the reasons stated and the confidence level would be low. Would ascertaining a base time for each track by taking say the winner plus all other runners equal to or within 1 length of the winner help. Less accurate perhaps but may still give some confidence. 

I am thinking of trying to analyse the UK races. How do I get the historical data and in a form I can use? Are there any particular tracks that are poor for analysis and better left alone? I like the fact that they get their horses ready to run without the need to race them into fitness. 

 

[Newmarket]

Anyone able to help me with this, 

Just say i have a free $100 bonus bet to use at Beteasy,  i place a fixed bet on a horse at $10, for a return of $900. 

Two days later I find horse paying quite less on betfair, so i would like to lay at the $5-80 odds. What is the best way to work out odds so i have a return if horse wins or not. 

Honest question, i do not lay much on betfair. Ta

[curious]

2 hours ago, mardigras said:

Here is the detail on pool per runner (This is a little old - for 2014/2015). 

starters	races	total pool  (single race)	avg pool	avg per runner
4	33	$1,123,351.00	$34,040.94	$8,510.23
5	131	$6,487,063.00	$49,519.56	$9,903.91
6	251	$15,761,478.00	$62,794.73	$10,465.79
7	450	$35,986,090.00	$79,969.09	$11,424.16
8	587	$55,517,087.00	$94,577.66	$11,822.21
9	675	$74,329,757.00	$110,118.16	$12,235.35
10	687	$79,447,735.00	$115,644.45	$11,564.44
11	555	$68,526,912.00	$123,471.91	$11,224.72
12	582	$78,999,736.00	$135,738.38	$11,311.53
13	420	$61,955,404.00	$147,512.87	$11,347.14
14	395	$63,307,475.00	$160,272.09	$11,448.01
15	145	$29,446,177.00	$203,077.08	$13,538.47
16	144	$30,102,679.00	$209,046.38	$13,065.40
17	17	$5,287,709.00	$311,041.71	$18,296.57
18	35	$13,897,685.00	$397,076.71	$22,059.82

 

This does not consider the individual races and race quality. Even though 15 - 18 are the highest per runner, I'd expect they are also inclusive of some of the larger betting races such as major cup meetings etc. I don't see much evidence of 14 starters being optimum.

Yeahh, if you chuck the outliers there's probably no significant difference between 7 and 14 going by that.

[curious]

1 hour ago, mardigras said:

Ultimately centred on a point that reflects the confidence - but still not centred as in a normal distribution, but skewed as you mention. A bit like pulling the centre slightly left or right depending on confidence. I don't even know if it would alter anything, but was keen to try it if I could work out how to do it.

So what you really want to play with is a distribution where the centre shifts left or right depending on confidence but the range doesn't shift. So that would vary from event to event? I think I'm getting the idea but at the same time scratching my head as to how you would do that. I think you'd need to set some rules which you also can play with around how much you shift the centre relative to some sort of shift in confidence score which presumably you already produce. I think that could be done.

[hesi]

2 hours ago, mardigras said:

Here is the detail on pool per runner (This is a little old - for 2014/2015). 

starters	races	total pool  (single race)	avg pool	avg per runner
4	33	$1,123,351.00	$34,040.94	$8,510.23
5	131	$6,487,063.00	$49,519.56	$9,903.91
6	251	$15,761,478.00	$62,794.73	$10,465.79
7	450	$35,986,090.00	$79,969.09	$11,424.16
8	587	$55,517,087.00	$94,577.66	$11,822.21
9	675	$74,329,757.00	$110,118.16	$12,235.35
10	687	$79,447,735.00	$115,644.45	$11,564.44
11	555	$68,526,912.00	$123,471.91	$11,224.72
12	582	$78,999,736.00	$135,738.38	$11,311.53
13	420	$61,955,404.00	$147,512.87	$11,347.14
14	395	$63,307,475.00	$160,272.09	$11,448.01
15	145	$29,446,177.00	$203,077.08	$13,538.47
16	144	$30,102,679.00	$209,046.38	$13,065.40
17	17	$5,287,709.00	$311,041.71	$18,296.57
18	35	$13,897,685.00	$397,076.71	$22,059.82

 

This does not consider the individual races and race quality. Even though 15 - 18 are the highest per runner, I'd expect they are also inclusive of some of the larger betting races such as major cup meetings etc. I don't see much evidence of 14 starters being optimum.

I presume that is NZ data over more than a season

Is that from NZTR/RB

I'm sure I saw in one of their reports that 14 was optimal

Will try and find it later on

[mardigras]

2 hours ago, hesi said:

I presume that is NZ data over more than a season

Is that from NZTR/RB

I'm sure I saw in one of their reports that 14 was optimal

Will try and find it later on

That's VIC TAB data (on NZ races) for approx 23 months - January 2014 through to November 2015. It is only pools that are single race pools (i.e no double/treble/quaddie pools etc). VIC TAB data/NZRB data essentially the same thing since commingling (outside of exchange rate fluctuations).

It hasn't come from NZRB - it is from me. But its veracity is unquestionable in my view.

[mardigras]

2 hours ago, curious said:

So what you really want to play with is a distribution where the centre shifts left or right depending on confidence but the range doesn't shift. So that would vary from event to event? I think I'm getting the idea but at the same time scratching my head as to how you would do that. I think you'd need to set some rules which you also can play with around how much you shift the centre relative to some sort of shift in confidence score which presumably you already produce. I think that could be done.

Exactly. Not 100% sure why I want to do it - since I don't know how to do it. But its just something I've thought about and felt it may be worth a look.

[mardigras]

2 hours ago, Newmarket said:

Anyone able to help me with this, 

Just say i have a free $100 bonus bet to use at Beteasy,  i place a fixed bet on a horse at $10, for a return of $900. 

Two days later I find horse paying quite less on betfair, so i would like to lay at the $5-80 odds. What is the best way to work out odds so i have a return if horse wins or not. 

Honest question, i do not lay much on betfair. Ta

If you are looking to win the same amount irrespective of the result, then this is the formula for the amount to lay.

Lay Stake = (Back Odds) * (Back Stake) / (Lay Odds - Commission Rate)

So for your example - using say a commission rate of 5% (0.05)

Lay Stake = (10 * 100)/(5.80-0.05)

Lay Stake = $173.91

Return = +$65 net of any commission whether wins or loses.

If wins, you get 900 - ($173.91 * 4.80) = 900 - 834.72 = $65

If loses, you get -100 + ($173.91 - $8.70) = $65

if you are looking to win more if it wins, but lose nothing if it loses, then that is a different formula but is simply laying the stake where (stake less commission = back bet staked) so 100/0.95 = $105.26 to give you $100 back if it loses and a greater amount if it happens to win.

If the commission rate applicable to you/the race is different, you can change it to suit.

[Newmarket]

14 minutes ago, mardigras said:

If you are looking to win the same amount irrespective of the result, then this is the formula for the amount to lay.

Lay Stake = (Back Odds) * (Back Stake) / (Lay Odds - Commission Rate)

So for your example - using say a commission rate of 5% (0.05)

Lay Stake = (10 * 100)/(5.80-0.05)

Lay Stake = $173.91

Return = +$65 net of any commission whether wins or loses.

If wins, you get 900 - ($173.91 * 4.80) = 900 - 834.72 = $65

If loses, you get -100 + ($173.91 - $8.70) = $65

if you are looking to win more if it wins, but lose nothing if it loses, then that is a different formula but is simply laying the stake where (stake less commission = back bet staked) so 100/0.95 = $105.26 to give you $100 back if it loses and a greater amount if it happens to win.

If the commission rate applicable to you/the race is different, you can change it to suit.

Ok, Thanks. 

[curious]

3 hours ago, hesi said:

I presume that is NZ data over more than a season

Is that from NZTR/RB

I'm sure I saw in one of their reports that 14 was optimal

Will try and find it later on

Might have been from a delusion in the NZTR business plan about optimal field size?

This is from the old 2010-12 one:

2.5 Field Sizes
Due to punter preference, NZTR must ensure strategies to encourage consistent field sizes of 12-14 starters.

[hesi]

I think there were bar charts and more data, not just a statement.

But you are probably right, after seeing some of the stuff in NZTR/RB Business plans....delusional

Andrew Brown and Racing Ahead being the most so

[mardigras]

20 hours ago, FeelTheFear said:

I am thinking of trying to analyse the UK races. How do I get the historical data and in a form I can use? Are there any particular tracks that are poor for analysis and better left alone? I like the fact that they get their horses ready to run without the need to race them into fitness. 

Not sure how you go about getting historical data around form. Maybe RacingPost subscription - although I'm not sure what historical data they have available.

Not so sure about poor tracks, but the tracks are far more varied than NZ/Oz tracks. Tracks with big undulations and long uphill or down hill sections. Take a track like Epsom for example. A maiden might typically run around 55-56 seconds for 5f (downhill) - compared to running 62+ seconds for 5f at Pontefract - which has an uphill straight. I am not sure if any horse has broken 60 seconds at Pontefract for 5f!

[FeelTheFear]

3 hours ago, mardigras said:

Not sure how you go about getting historical data around form. Maybe RacingPost subscription - although I'm not sure what historical data they have available.

Not so sure about poor tracks, but the tracks are far more varied than NZ/Oz tracks. Tracks with big undulations and long uphill or down hill sections. Take a track like Epsom for example. A maiden might typically run around 55-56 seconds for 5f (downhill) - compared to running 62+ seconds for 5f at Pontefract - which has an uphill straight. I am not sure if any horse has broken 60 seconds at Pontefract for 5f!

cheers mate. Sounds like I have years of learning ?

I did all the Group, C1 and C2 form for Hong Kong until recently as they have all the times/sectionals etc. My data spans 54 columns. It allows me to give a rating based upon times run versus par times for the track/distance/class. Also has all sectionals that I convert to a % for each so I can see where the horses energy was used. Class was defined by Group racing rating 100 and par times for C1 and C2 compared and converted to lengths and taken from 100. A track variant for each day was done by comparing race times against their par times, averaging to produce a + or - figure added to the raw rating. Hong Kong update their data regularly to capture the past 3 years racing. I used to do everything manually.  I fairly recently learnt how to use formulae to do the individual calculations, much faster now. 

I stopped doing it in Sep. because I don't have a connection with Hong Kong. NZ racing I have that. I do all the breeding, have done for over 30 years. Perhaps I should go back to the Hong Kong database and catch up. Haven't had a chance to test it yet. 

[mardigras]

On 28/12/2018 at 11:03 AM, FeelTheFear said:

cheers mate. Sounds like I have years of learning ?

I did all the Group, C1 and C2 form for Hong Kong until recently as they have all the times/sectionals etc. My data spans 54 columns. It allows me to give a rating based upon times run versus par times for the track/distance/class. Also has all sectionals that I convert to a % for each so I can see where the horses energy was used. Class was defined by Group racing rating 100 and par times for C1 and C2 compared and converted to lengths and taken from 100. A track variant for each day was done by comparing race times against their par times, averaging to produce a + or - figure added to the raw rating. Hong Kong update their data regularly to capture the past 3 years racing. I used to do everything manually.  I fairly recently learnt how to use formulae to do the individual calculations, much faster now. 

I stopped doing it in Sep. because I don't have a connection with Hong Kong. NZ racing I have that. I do all the breeding, have done for over 30 years. Perhaps I should go back to the Hong Kong database and catch up. Haven't had a chance to test it yet. 

 

FTF, lot's of info there. Have you considered installing some database software on your computer. And then do a bit of teach yourself SQL. There are many free choices around for databases. Just allows analysing the information you have in an easier way.

Options like MySQL or Microsoft SQL Express can be installed and used free.

Learning SQL. Even just enough to load the data into tables. Then some simple querying to select data with group by and where clauses. 

If wanting to get more advanced and take up less disk space, you could look to normalise your data. I.e. have different types of data stored in separate tables with identifiers that relate the data together. Separating horses from meetings from races from runners and horse prior starts etc. Just means you don't have to store the same data repeatedly.

I understand these things take time. But the main issue with things like spreadsheets, when you want to change something minor, it can become more time consuming compared to changing the same thing in a database query.