Why 74?

[njcurveball]

In my data mining, I found an interesting pattern, perhaps?

The #1 HTR rating as has been written about, is certainly worth it's weight in gold. Ken has done what many young computer Jedi's train with Yoda to accomplish. One single rating that can prove profitable across tracks, distances, surfaces, seasons, etc.

Thanks for your work Obi Ken!

Now my data is limited, but in testing the #1 HTR rating with ML > 5, I have seen great things. Since the K rating is a good estimate of a horses chances, the two of them together perform well.

However, when the K rating drops below 74, there are only 3 winners out of 107 plays. Although two bombs almost made this profitable, the win % got anemic.

With the rank of 9 and K > 73, the horses are 26 for 361, which is a good 7.2% for major longshots.

But with K < 74, the win% is 2.2% for the same K rating.

OK, back to the mines. ;-)

[hurrikane]

my nj data friend.

how did you arrive at 74?

And it is surprising how many longshots the htr# will come up with...esp inside the top 5 K

[njcurveball]

74 was the cutoff for a respectable win% when I was testing. THE #1 HTR horse also seems to hinge on whether the horse raced on todays surface in their last race.

I would suggest anyone that is down in the "data mines", try slicing with different groups, one for horses who ran on the same surface last out and those switching surfaces. Some factors perform well in one case, but not the other.

The export tool that we have in HTR thanks to Ken is basically the "key to the mint" that was headlined in so many direct mailings in the 70s.

And the great thing about this game is that with so many handicappers worried about winning, they forget all about profitability.

I would agree that prices on logical favorites have took a hit with computer handicapping. But it seems horses in the $10 to $30 range have stayed pretty consistent from what I have seen.


Jim

[AwolAtHTR]

refer NJCurveBall

2060401-20060407 Races= 970 PL-5 Track= ALL MLO 5.0-99.9 HTR=1

Factor Plays Winners Win W+P ITM WROI PROI SROI $AvgM LongW High I.V. 04/18/2006
---------------------------------------------------------------------------------------------------------
Play All 00289 00053 18% 32% 42% 1.27 1.05 0.86 $13.8 17% $73 1.58

================================
20060408-20060413 Races= 796 PL-5 Track= ALL MLO 5.0-99.9 HTR=1

Factor Plays Winners Win W+P ITM WROI PROI SROI $AvgM LongW High I.V. 04/18/2006
---------------------------------------------------------------------------------------------------------
Play All 00241 00034 14% 27% 39% 0.98 0.80 0.77 $13.8 21% $67 1.21



well NjCurveBall,
you spiked my interest when you said '..#1 HTR rating with ML > 5, I have seen great things....'

I prepare my archives on a weekly basis. So, when I try a spot play, I will check two or three weeks before going to my Access data base which has archived months and years.

with ONLY these conditions, we find that YES, using all tracks there are over two hundred '..bettable..' races for each of the first two weeks of April 2006. However, first week makes GOOD money and the second week is a loser.

maybe HTR works best the first week of every month!!

that is, I just tried spot play by Victor and found this same result.

now, I really need to get back to my cave.

[njcurveball]

If I am reading your test correctly, the 2nd week has a return of 98 cents for every dollar bet. That isn't all that bad.

What happens if you take out all the horses with a K rating less than 74?

I think you are definitely fishing in the right place when you find an 800 race sample that almost breaks even using only 2 factors (HTR and ML) over all the races in the database.

That is just amazing, dont you think? This Ken guy is good. ;-)

Jim

[km]

Sure appreciate the nice compliments NJ !

I've run tests on the (K) trying to isolate a number like "74" that is the break off for a good contender selection. Can't beat the plain ranking most of the time. K < 74 corresponds to a typical K rank of 7 or 8, 9 and that is the pain level statistcally.

Mentioned this in the newsletter this month in the article about NJ's 1000 losers - the K rank "7" seems to give a firm break in the stats for winner/loser.

Good stats NJ and AWOL, appreciate it, keep em coming

[hurrikane]

thanks folks

I use the htr number in a big spot play I have had for a couple of years. Surprising how well that number holds up. Seems to have gotten better with the advent of the K ranking.

[hracingplyr]

and how does your spot play work?

[dehere]

So, the cut-off was 74 right? Just curious as the next to the last leg in tomorrows "NHC Qualify" event (5th at HOL) has a 20/1 ML horse with #1 HTR and K-rate of 78. I haven't looked at anything else about the race but that combination kinda intrigued me after the thoughts mentioned in this thread.

[tomcat]

Today at Aqu. #1 in the 5th race is definately a no no with the very low K and the ML30-1. However he has a competive pdif of -2, an ! and a # and a @ somewhere. Remind me Ken, of what all that is about and which factors carry the most weight.

[km]

# Jockey = change from last

! Cla (TLC), Msw drop to Mcl (first time in a maiden claimer after running in all Msw previous)

@ Lost last by 20 or more (any reason)

Competitive Pscan based on the fact that the horse has faced much better in those Msw despite finishing far back, and that the other horses are really slow too.

Enjoy the maiden claimers in NY while AQU winds down. BEL and SAR do not run cheap maiden claimers - maybe an M75 now and then.

[hurrikane]

Those cheap mdns and the early speed on turf have given me some very sweet p4s.

I'm going to miss them