PDA

View Full Version : Believe half of what you read, then test that half!


Carl
28th July 2001, 11:00.32 AM
This is a sort of continuation of the "Request for Info" tread.

The story today for newbies:
1) Mike posted a quote from Gordon Pine's Newsletter:

"At Colonial Downs, early pace is important. The horse that figures to be in front at the second call wins 28% more than it should given its odds (has an A/E of 1.28), and generates a 1.09 ROI (9 cents profit for each dollar bet). (These stats are derived from a handicapping program I designed.) It’s also a track favorable to longshots: Over-achievers have an A/E of 1.20 and an ROI of 1.02. Pretty good predictors. But take these same predictors to another meet: at Arlington Park, the top second call horse has a .81 A/E and a .69 ROI. Over-achievers have a .92 A/E and a .79 ROI. So, at Arlington, one predictor earns 40 cents less per dollar and the other earns 23 cents less per dollar than at Colonial."

2) Carl glibbly replied:
"I am sure Gordon is right about what has happened at Colonial and Arlington. "

3) Carl should know better. About the only "sure" here is Carl should check his facts before being "sure" of anything. Duh.

4) Anyway, Carl did run the first 22 days of July rEP1 for both CNL and AP, and got TOTALLY different results than Gordon reported. (Yeah, I know, probably different time period, different data provider, etc.) Anyway, by way of my records anyone who followed rEP1 at CNL over the last three weeks is now broker by a considerable amount.

ALL_MSA Misc Query ROI Report
CNL----rEP1 July 1-July 22nd, 2001 pl1.
Total Bets-------------------129
Total Amount Bet-------------258.00
Wins-------------------------20
Pct.-------------------------15.50%
Amount Won-------------------147.80
Profit/Loss------------------(-110.20)
Pct Profit Loss--------------(-42.71%)
Avg. Payout------------------7.39

And AP, while slightly better, is no hell either.

ALL_MSA Misc Query ROI Report
AP rEP1 July 1-July 22nd 2001 pl1.
Total Bets-----------------------152
Total Amount Bet-----------------304.00
Wins-----------------------------28
Pct.-----------------------------18.42%
Amount Won-----------------------243.20
Profit/Loss---------------------(-60.80)
Pct Profit Loss-----------------(-20.00%)
Avg. Payout---------------------8.69

Anyway, I will do a seven day report (July 23-30th) on Monday, see if they reverse or do whatever.
------------------------------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------------------------

Which leads me to point two, which is a shameless plug for HTR, but in a strange sort of way..

1) When I first started here about 18 months ago, Ken had posted the results of a 1100 race turf study, whereby he found that 90% of the winners had two characteristics. I remember them as
A. Within 4 of the par numbers of the class entered in and
B. One of the top 5 LV horses.

2) I didn't know how to do db at the time, but I checked 100 or so turf races manually and found these numbers to be vastly overinflated, and posted so on the old board. A few of the other poster who had db's checked, and agreed with me.

3) I e-mailed Ken a few times, all hot and bothered, accusing him of giving out "false information".

4) Ken was amazing. Believe it or not, he thanked me for checking, told me that I did the right thing, and that I "should never believe anything I read without checking it first". Not quite the response I expected! (Well, anyway, it was not the response I would have given to someone who just accused me of "falsely posting incorrect data".)

5) Upshoot of it was, Ken had taken his numbers from a lot of summer "big time turf tracks", I was looking at Tampa Bay/Turf Paridise type tracks. This explained at least part of the difference.

So, sometimes we all post stuff that turns out to be untrue (yes me too). The interesting part for me is how we react when someone points it out. Some of us thank the person, some get huffy. I like buying my data from a store where the owner thanks me.....

later,
Carl

hurrikane
28th July 2001, 01:29.34 PM
Thanks for the post Carl, I was very suspect of Gordons numbers myself as just from playing the track I have not found that to be true. As you say though..different data..different days.

As a little dig...so what you are saying is that the results you and Ken were getting were different because of the tracks? Hmm...that would seem to make things more track specific.. :p

Of course, with the ROI you have been playing lately I'm certainly not questioning your thinkings..just want to understand it.

tbrown
28th July 2001, 03:58.25 PM
I guess I would just assume anything that involved a pace factor like EV or Lv would be more likely to be track specific. Even at Belmont, I see two different things depending on inner or main and by distance.

MikeDee
28th July 2001, 07:54.11 PM
In addition to the fact that one can be looking at different time frames, one must keep in mind how the pace line is being determined. Assuming Carl is PL1 and I run the same data PL5, I would not be surprised if we did get different results. All you need is a few different picks and some decent prices and a loser becomes a winner.

Ken Massa
29th July 2001, 03:43.14 AM
Thank you Carl, for those very kind words.

HTR remains the only handicapping vendor that allows 100% of it's proprietary material to be accurately and thoroughly tested by users without bias. Because we are an unlimited download service, it wouldn't take long for subscribers to find out if this stuff was b.s. and worthless.

The turf race test results that Carl mentioned from a past newsletter, have never been duplicated by me or anyone else since. They were flawed and the results were not indicative of reality. They were probably coded improperly or based too heavily on SoCal/NYRA tracks in my database.

As a benefit of Carl's challenge - I am more careful about test results that look too good to be true. When a positive ROI is found, it is either re-tested or sent to other users for independent verification before being printed in the newsletter. Often, the information also has a disclaimer such as "test sample small" or "results skewed by two huge longshots".

When people are betting real money on the words we write - make damn sure it is the truth and well researched before putting it out.

Carl
30th July 2001, 09:50.14 AM
In any case, here are the numbers I have for rEP1 for this last week at AP and CNL. AP got worst, CNL got good, even made money.


ALL_MSA Misc Query ROI Report AP July 23-29th rEP1 pl1.
Total Bets------------------50
Total Amount Bet------------100.00
Wins------------------------6
Pct.------------------------12.00%
Amount Won------------------43.60
Profit/Loss-----------------(-56.40)
Pct Profit Loss-------------(-56.40%)
Avg. Payout-----------------7.27

ALL_MSA Misc Query ROI Report CLN July23-29th rEP1.
Total Bets----------------------51
Total Amount Bet----------------102.00
Wins----------------------------10
Pct.----------------------------19.61%
Amount Won----------------------118.00
Profit/Loss---------------------16.00
Pct Profit Loss-----------------15.69%
Avg. Payout---------------------11.80

later,
Carl