Placegetting stats

[hermes]

I can't find much statistical info on placegetters (as opposed to winners) so I've compiled some figures from my small selection of old form guides. Sample is small but some trends are apparent and should hold good in a larger sample. Looking at 200 metro races, no 2yo, no hurdles or sprints:

AVERAGE PLACE PERCENTAGE

Average place percentage of placegetters = 48.78%

PLACEGETTERS WITHIN SIXTH LAST START

Last start winners = 32% of placegetters
Last start second = 14%
Last start third = 19%
Last start fourth = 16%
Last start fifth = 10%
Last start sixth = 9%

AVERAGE DAYS SINCE LAST RUN

31 days.

LAST START IMPROVERS

Last result was better than or equal to second last result = 69.6% of placegetters.

NINES

Best three of last four starts adds to nine or less. (eg. 3 x third = 9) A "Nines" horse has score of nine or less.

57.7% of placegetters = nines horse.

More stats as I compile them

Hope its helpful to someone

Cheers

Hermes

[quote]
On 2002-06-27 09:27, hermes wrote:

AVERAGE DAYS SINCE LAST RUN

31 days.

The problem with this is 1st up horses will distort the figure. Most will be within 7-21 days. Anyway most horses in general will fit into either 7-21 days or 1st up, so while this stat seems important it doesn't eliminate many horses.

ANYWAY KEEP DOING THE RESEARCH AND HOPEFULLY YOU'LL FIND PLENTY OF GOOD STATS TO HELP YOU.

[Placegetter]

On 2002-06-27 09:27, hermes wrote:
Looking at 200 metro races, no 2yo, no hurdles or sprints:


Can you please define a sprint? It seems unusual that you would leave these races out as most of the races in Australia are less than or equal to 1600m.

[hermes]

Sprint is wrong terminology. I eliminated races of less than 1200m.

[Equine Investor]

hermes..no placegetters finished worse than sixth last start ? Or did you only include placegetters which finished no worse than sixth last start. If so I wonder what percentage finished worse than sixth last start. As a total percentage.

You might need a bigger sample to gain a beeter indication. But this is well worth persuing. You're definately on the right track!

The other factor to consider is that there may be at least half the field that qualify as finishing no worse than say 4th. Then you got real trouble.

It's a great place to start though.

<font size=-1>[ This Message was edited by: Equine Investor on 2002-06-28 00:23 ]</font>

<font size=-1>[ This Message was edited by: Equine Investor on 2002-06-28 00:25 ]</font>

[hermes]

More results from the same sample + about 20 more races:

THE ZIP

The Zip ratings from the Sportsman give interesting results. In round figures:

17% of placegetters have a Zip Star* (Top zip rating)
16% of placegetters one point from top zip.
21% of placegetters two points from top zip.
11% three points from top zip.
16% four points of top zip
15% five or more points from top zip.

That is:

About two-thirds of placegetters are top zip or within three points of top zip.

About half of placegetters are within two of top zip.

About 85% of placegetters are within four of top zip.

In practical terms:

If you have within three of zip as a qualifying parameter you will net two thirds of placegetters. (Conversely, you will lose a third of your placegetters.)

If yopu have within two of zip as a qualifying parameter you will net half your placegetters (and lose half).

If you have within four of zip as a qualifying parameter you will catch 85% of the little buggers while sacrificing 15%.

Useful? I'm not sure yet...

Hermes

[hermes]

And here's a set of numbers:

Looking at a raw batch of 280 races (all races, any races)

In 192 cases one or more placegetter was saddlecloth 1, 2 or 3. (In only 88 races no placegetters in first three numbers.)

Of 840 placegetters (280 x 3) 284 were in top three saddlecloth numbers. 33%.

There were only 4 cases where all three placegetters were in top three, i.e. #1, #2 and #3. There were 80 cases where two placegetters were #1, #2 or #3.

Figures don't account for races with no third div. and other such factors. Just raw figures. But it seems in about 60% of races one or more placegetter is 1, 2 or 3. And a third of all placegetters carry 1, 2 or 3 - what you'd expect in a ten horse race, I suppose.

A general question about placegetters. They are a very diverse bunch with diverse characteristics. Is it the *third* placegetter that makes the stats such quicksand? Is there more terra firma looking at only first and second placegetters and chasing them?

If someone can answer that question for me it might save lots of wear and tear on the calculator.

Thanks

Hermes

You can't possibly catch all winners or placegetters, so just focus on a group which is above 50% and take the good dividends when they arrive.

[hermes]

Another one:

This time looking at those runners who have scored a first, a second and a third in its last six starts. A spread of places. Finished in all three places (as opposed to say all seconds and firsts).

The theory is that such a horse has proven its ability to be in the money on a consistent basis. Its a reliability factor.

Looking at a sample of 148 races, metro, 1200m or more, no hurdles, no 2yo:

1904 runners.
Of these only 156 had a 1, 2 and 3 in its last six starts.
Of these, 76 were placegetters.

Or, about 8% of all runners qualified. About 50% of those 8% placed.

In practical terms, seek out those runners with a 1, 2 and a 3. Half of 'em will be placegetters. So what? you say.

Stay tuned for more Hermes' useless placegetter stats....

Maybe seek out those which had 2 or more places in past 4 starts???

[hermes]

Equine Investor,

Thanks for drawing attention to a problem in my figures. I'd only counted those to sixth last start. This being because it nets some 87% of placegetters.

I've added some more races to the same sample. Fuller (and double-checked) TOTAL figures, rounded out are:

PLACEGETTERS LAST START

Last start winners = 29%
2nd = 13%
3rd = 16%
4th = 14%
5th = 8%
6th = 7%
7th = 3%
8th = 3%
9th = 4%
10+ = 3%

87% of placegetters finished sixth or better.
80% finished 5th or better.

Last start placegetters placed in their next race 58% of the time.

Needs a bigger sample obviously but the trends are there.

Cheers

Hermes

On Saturday,

Almost every placegetter had ran 5th or better twice in past 4 starts.

Problem is so did half a dozen other horses each race.

The most successful numbers based on a couple of weeks of data were
1,2,4.
3 & 6 were next best.

Most had a place % of 40% or higher.

Quick system

1. 2 or more top 5 finishes in past 4 runs
2. No 1,2 or 4.
3. Place % of 40%+
4. Place div of $2+

[hermes]

Thanks for the advice freddy. You're right. Can't pick 'em all. But I was asking about statistical batches. The trouble is that the category "placegetter" is very broad. It includes a diverse range of horses that have few characteristics in common. I'm wondering if maybe a batch of 1st and 2nds only would be more than one third more stable, if you see what I mean? Maybe it is far easier to find meaningful trends in the 1st-2nds category because it is the third runner that upsets the trends???

I'Ii try out placegetters last two starts and see.

[hermes]

Quick system is worth a try Freddy. I'm having no trouble inventing systems that pursue the lines of least resistance in the stats but they run into trouble at the market. In most systems I'm trialling or have trialled it is hard to find the horses at $2 or better, or hard to find any volume of them. There's a couple every Sat. sure, but then you need discipline and most of the fun is gone...

[hermes]

Freddy,

Two or more places in last 4 starts.

Sample- 176 races
Total placegetters - 528
Qualifiers - 300
Non-qualifiers - 228

About 57% of placegetters in this sample have two or more places in last four starts.

Very similiar to "Nines", see my post above. Nines is an old-fashioned shortcut for extracting winners from the form guide. Add together the best three of last four starts. If total is nine or less, qualify. I know someone who uses "Nines" for trifecta selection with some success. At least, he collected last Sat. and I didn't...

Hermes

[hermes]

Freddy wrote:

"Almost every placegetter had ran 5th or better twice in past 4 starts. Problem is so did half a dozen other horses each race. "

That's the problem I'm finding with placegetter stats. As I say, they are a very diverse bunch, placegetters, with few characteristics to distinguish them from the pack. Exactly the problem.

I found a solution, but can't say what it is.

[Equine Investor]

hermes, perhaps even though you are looking for place % etc, when you can't separate say 5 or 6 runners, then look at their average prizemoney per race which should give an indication of class, and also their win %.

This should sort out the wheat from the chaff and not all will be short odds. You may notice in the formguides that wins are not counted as places, so if you are backing for a place you should include wins in your place %.

Also your stats on saddlecloths 1,2,3 seem to be right and throws up some great place dividends, but your stats would be a lot higher if you only included handicaps and not WFA and Set weights. See what I am driving at? The handicapper usually gives the best horses the biggest weights, and this does not apply to set weights and WFA etc.

[quote]
On 2002-06-30 18:11, Equine Investor wrote:
You may notice in the formguides that wins are not counted as places, so if you are backing for a place you should include wins in your place %.

Unless you are using Irish formguides this is not correct. Wins are included in place %.

[Maya]

Freddy,

Have you checked out the Wizard formguide lately? Win% are seperate from place%.
eg. 1 win, 2 placings from 5 starts is written as 20% win and 40% place - and no, it's not Irish.

[Equine Investor]

Exactly, unless you purchase a publication like the Sportsman, many daily formguides don't include wins in the place %

I don't buy the Wizard. Every other one has win % included as place %. It's not logical otherwise.

[hermes]

Here's some interesting stats on placegetters that should be of interest to place betters and those hunting trifectas. I know the stats have all been done before and I'm reinventing the wheel here, but you can learn a lot from doing a few stats yourself - it is a wheel worth reinventing.

What I did was examine a stack of races looking at placegetters vis-a-vis the MIDWEIGHT. The midweight is simply the middle weight in the race. If top weight = 56kg and bottom weight = 50 kg. Midweight = 53kg.

How many placegetters were above or below the midweight?

A. Three placegetters above the midweight = 17% of the sample.
B. Two placegetters above, and one below = 34% of sample.
C. Two below and one above = 25%
D. Three below = 6%

(If a placegetter was on the midweight I counted it above or below depending on the other two. Found no cases of three placegetters on the midweight.)

It was a small sample but trends are there in even small samples. There are four types of possible results: A, B, C and D.

I reckon the proportions will eventually pan out at about: 40/30/20/10. - with A = 20, B = 40, C = 30 and D = 10.

These stats match others that tell us, for example, that there is a distinct bias among placegetters for the lower saddlecloth numbers (topweights).

Similarly, I'm told (by a trifecta enthusiast) that trifectas are rarely filled by the three market favourites. More often, two of the three market favourites, and one other. Much the same pattern.

This could have many practical applications. For example, I rate horses in a race according to how likely I think they are to run a place. If, however, I have three selections above the midweight in my ratings, the rule should be: have a closer look. Only about 20% of races have three placegetters above the midweight. Is this race really one of them? Or is there a horse below the midweight that I've overlooked? If there is, chances are it will pay well (but it will be correspondingly more difficult to pick).

I haven't checked but I imagine prices follow much the same pattern. (Only talking about handicaps here.)

Three above the midweight = trifectas that pay the least.
Two above and one below = pay better
Two below, one above = pay better still
Three below = pay best.

In general I'm beginning to think that the long-term success of placebetting (and trifectas) depends on how many of the below midweight placegetters you can pick. So my next task is to try to find common characteristics of that allusive below midweight placegetter.

As always, constructive suggestions welcome..

Still at it.

Hermes

[hermes]

STATS APPLIED

All systems of selection must involve some mode or mechanism for eliminating non-prospects in a race. I'm interested in picking placegetters. All systems of picking placegetters must at some stage eliminate those horses you think have no chance of running a drum. (In fact, if its a starter, it has a chance, but you know what i mean.)

Using the stats I've been working on (and others kindly supplied by punters in this forum and elsewhere) I've been designing some coarse filters. (I'm building a system from the ground up - that's what I'm like. I'm fantatical about this. I want a 100% strike rate!). In my system (still under construction) I begin (step 1.) by identifying a fairly broad group of qualifiers. This is the first filter. Eliminate the rubbish.

But its harder than you think. I want a filter that eliminates the rubbish without taking out too many placegetters in the process. I've tried numerous combinations of factors, testing them out on my sample of old races. You tend to either lose too many placegetters or not enough of the rubbish. But I have found some good combinations. Here's a couple:

*Top 8 average place percentages.
*Last start finished 1st to 6th.

This will usually reduce a field to 7 or less qualifiers and net about 80-85% of placegetters. Gets a race down to manageable proportions.

*Above 33% place average
*Within 4 points of the top Zip rating (in the Sportsman)

Depends on the calibre of the race. It will net over 75% of the placegetters but often it won't reduce the field much. Occasionally, very good results where only a few horses are within 4 of Zip and you've still got sometimes up to 80% of the placegetters. (That's two and a bit horses!)

Another good one:

*Top seven average prize winners.
*Top seven average place percentage.
*Last start finished 1-6.

Good results. Reduces the field and nets a reliable number of the placegetting horses.

(I might add that I'm not all for stats. I'm looking for coarse filters as guides. My approach is to reduce the field step by step, filter by filter, but the ultimate decision is not a mechanical one. I'm increasingly convinced that you also have to know something about horses....)

Cheers

Hermes

[hermes]

SOME ZIP STATS

The zip system used in the Sportsman offers an alternative handicap assessment of runners. I compared the zip rating to the official weight, looking to see whether it was above, below or the same as. A surprising result but only in a small sample.

Looking at all runners the zip is overwhelmingly below the handicap - 74% of runners. In 15% of runners the zip rated over, and in only 10% of cases zip and handicapper were in agreement.

But among placegetters you get quite different stats:

Placegetters where the zip was above handicap = 33%
Placegetters below = 48%
The same = 18%

So, in the mass of runners you get the (rough) proportions: 70/20/10. In placegetters: 50/30/20.

Conclusion:

Where the zip is below the handicap, it counter-indicates a placegetter. Zip below handicap is more characteristic of all runners than it is of those runners who place. But
where the zip is above or agees with handicap, a pointer to a placegetter.

Not a strong pointer, but a pointer all the same. Keep a look out for horses with zip over or the same as the handicap weight.

A strong trend in a small sample (187 all races), no hurdles.

Another possible path to placegetters???

Hermes

[hermes]

The average zip rating of placegetters, by the way, is about 52, but 51 and 53 are by far the most common zip numbers in my sample.The numbers fall right off below 50. Virtually no placegetters with zips below 50 - couldn't find a single one below 48. Which is useful to know because there will often be two or three runners below 50 which you can therefore safely eliminate. Very few placegetters with zips over 57. The range is 57-50. A tighter range of 56-51 would still net the vast majority of placegetters.