Are strike rates over rated?
 

When using strikerates to award bouses or penalties or whatever do we put too much credence in them? for example, A stat I saw stated that tab no 1 had a strike rate of 15.88 % and no 10 a strike rate of 3.97 Does that mean we give tab no 1 a4 times more chance of winning? What we should remember is that tab no 1 will be competing on average against9 or 10 others in each race it is involved in whereas tab no 10 will probably compete against 13 or 14 others on average ( only guessing here for the sake of trying to make a point) so that raw strike rates on their own can give too much of an advantage to some horses. This theory of course can relate to barriers, win and place percentages, jockey and trainer stats or any other factor to do with racing. Maybe some people with data with the number of starters in in each race can give some stats feedback on my theory. My overall point being that Tab no 1 will find it easier to win against 9 or 10 than no 10 will against 13 or 14 yet their raw strike rate will will not show this. Hope this made sense.

Dr Ron,
I have attached a spreadsheet that will help you understand how to assign statistical Weights to Variables/Factors based on Observed and Expected results and to a confidence level
You can compare the results of each Variable/Factor tested to a fixed standard and then assign a Statistically Significant Weight to each performance IN ORDER TO DETERMINE IF THE POINT SYSTEM PERFORMED BETTER THAN RANDOM CHANCE?
In the attached SPREADSHEET simply enter the EXPECTED PERCENTAGE, THE OBSERVED PERCENTAGE, and THE TOTAL NUMBER OF OBSERVATIONS. The SPREADSHEET will compute the POINTS to assign to that FACTOR. In order to be statistically significant the POINTS COMPUTED must be more POSITIVE or more NEGATIVE than the CRITICAL SCORES at the bottom of the page.

You would go thru each Variable /Factor ranking enter the data for each ranking record the result Point score and ROE score the ROE records if it is statistically significant ie greater than 90% etc..

Most Basic point score systems would use the above method

Thanks for the spreadsheet woof, will conduct some experiments with it.

Would still like some stats re my TAB no example please guys.

Hi DRRON,

Yes, I believe too much emphasis is placed on OVERALL strike rate, just as too much emphasis is placed on career prizemoney and API.
All these figures are only relevant to today's class and distance.

For example: if TAB 1 has a strike rate of 15.88%, why can't I just back anything over $6.30 and show a profit. The reason is that the stat is influenced by other factors as you have already stated.

Some figures for TAB #1

Overall S/R 17.45%

3 runners 0%
4 runners 46.67%
5 runners 28.67%
6 runners 30.00%
7 runners 26.85%
8 runners 22.63%
9 runners 18.86%
10 runners 17.52%
11 runners 17.32%
12 runners 13.88%
13 runners 14.88%
14 runners 12.71%

I'll stop there to illustrate a point.

To compare the strike rate of TAB# 10 to TAB #1, you must have equal starters.

So we have...
TAB #1 10 runners = 17.52%
TAB #10 10 runners = 5.78%

But that's just the tip of the iceberg... there is weight added to the TAB number sometimes and other times not, also the gap in weights would be different in a field of 10 as opposed to a field of 21.

Not to mention form!

TAB #1 is three times as likely to win as TAB #10 - all other things being equal, but they're not. :(

The answer to the question lies in weight, fitness and class.
More class horses are TAB 1 than TAB 10, and even semi fit horses carrying more weight will beat a fit less classy horse carrying less weight, more times than not. Hence the skew in the figures.

If they were Robohorse (pinched from Duritz I think), then you could back them according to strike rate of TAB #.

I for some time studied a lot of statistics and they can tend to give you a lot of confidence that what you are doing is right.

But what has to be remembered is that each race is an individual contest on its own. ie: in a given race No.1 may be an absolute moral but this would be due to other combined factors. The same may apply to No.14 on any other occasion.

I find stats can help to fine down the chances in a race but you then have to look at whether the stats are applying to a horse that actually has a chance in that race on that day.

What you will also find with any stat is that they cluster there winners. You will not find a stat that has a strike rate of 30% hitting three winners every ten runners. They cluster and that is what makes them hard to follow.

Good Punting

Thanks for the replys guys, as your figures demonstrate chrome, the advantage for 1 over 10 goes from a raw strike rate of about 4 times to an adjusted one of three times when the number of starters is taken into account.

Please Explain Woof
 

Mate, in spite of what people may think I'm no dumb-******** but I can't make much of the spread sheet you sent. I've got a keen interest in statistics and I'm pretty sure the spreadsheet would be very useful if I could make head or tail of your explanation. Even putting figures in there hasn't helped. What exactly do the percentages refer to, what figure is the points?
Number of observations I think I've got nailed :-)

KV

Hi Kennyvictor,

Any point system based on ranking runners needs to be Statistically sound, in order to be STATISTICALLY SIGNIFICANT and offer a 99% CONFIDENCE LEVEL, The WEIGHT NUMBER in the attached spreadsheet must be more positive than Plus 2.33 or more NEGATIVE THAN MINUS 2.33.
The attached SPREADSHEET WEIGHT COLUMN is designed to provide points that place all items on a comparable basis. That is the beauty of this Spreadsheet. It compares the results of each Factor tested to a fixed standard and then assigns a Statistically Significant Weight to each runners performance.

An example would be to use the number of Wins from each Barrier. Imagine the average number of starters is 8 in each race, the Expected % would be 12.5%, we would then record the actual number of Wins for each Barrier and divide that number into the number of Observations this would become our Observed % in the spreadsheet.

Now if you were creating a POINT SYSTEM with several Handicapping Factors, you would be justified in giving points to each ranking within each Factor Based on whether it exceeded the +/- 2.33 Confidence level.

Once you have a Point Score for each ranking for each Factor, you will find maybe only 3or4 will be STATISTICALLY SIGNIFICANT and offer a 99% CONFIDENCE LEVEL.

Now for Todays Race you rank each runner for each Factor and using your Point Score Table you have previously developed you can rank each runner based on Points awarded from the Point score Table in a Statistically sound method.

Woof43!
Can't deny your knowledge in all things statistics.
But can you recall who ran second in the 2004 Melbourne Cup?
No one wants to lose money punting,but surely there is more to horse racving than making a profit.
p.s. I don't know who ran 2nd.Could have been Grey Scale but I think it was number 1 from overseas.

When was Grey Scale racing or did you mean Grey Song. I think it was Vinnie Roe.

Thanks for that Woof, much clearer now.

Melbourne Cup? Where do they hold that then?

KV

Yeah,right on both counts.
Grey Scale ran during the late seventies,early eighties.
Memories I still have.Money has long since departed.
But don't forget.

Money has no more intrinsic value than notches cut into a piece of wood.

Try buying a beer with a piece of wood. Chances are the barman will hit you with it. :-)