65 more races =

88: 2-3-4
rank 1-3 = 36 82%
rank 4 = 4 9%
rank 5+ = 3 7%

88:5+
rank 1-3 = 10 48%
rank 4 = 0
rank 5+ = 11 52%

Swing for the fences ONLY when races are 88:5+??
Or in races 88:2-3-4 where a top 4 horse has a rank 1 or 2
and is high odds.

65 more races =

Races with at least 1 rank 1: 44 68%
(There can/will be multiple horses per race that have
at 1 rank 1 in their factors.)
Races with no rank 1's and at least 1 rank 2: 14 22%
Races with NO rank 1's or 2's: 7 11%

If a horse has no rank 1's or 2's, there is a 89% chance
that they will lose.

Horses may rank low, but still have a single 1 e.g.
1-7 and rank 5+ and still be able to win. Do not
overlook these horses.

This has amazing possibilities!

Can't wait for you to teach this!

The best part is that it it’s built on using AI odds instead of ML or the tote.

It's not AI.

It is done using factors from HSH and mathematical jujutsu.

I've been studying the strength of this Report in identifying low odds horses to single, as Brian demonstrated last week. I've been trying to define what the Gap in Negative Reynolds numbers should be to consider a Low Odds horse as a single. Right now it's all feel and eye test, but as an example I like to see a race with a single L that has a Neg Rey in the 4th column and a gap to the next group of horses that looks something like 24 to 300, or 8 to 180, or 40 to 168 and so on.

Yesterday I identified 29 races from the 100+ races that I've played live in past two weeks that fit that criteria. By "eye test" alone I see 15 winners in 29 races. When I eliminate 4 races that were 88:7 or 88:8 the sample improves to 15 hits in 25 races and a dollar net of $2.90. Steady diet of winners $4.00 to $7.00. with a few $3.60s and a couple $8.00 and $8.60s. When I layer in the place contenders from my existing handicapping- which is a combination of BH A+/A's and Neal H's with low object odds- those same 25 races yield 18 exacta hits that are very "gettable". Tiny sample, of course, but my gut tells me this is not a fluke. Simultaneously, I've been studying how to isolate the BALOs in similar fashion.

Haven't reached any conclusions on that BALO piece yet, but it appears to me that a recipe of Brian's HSH jujitsu AND the AI is quite powerful. I will continue tracking both singles and BALOs in search of the right recipe and am looking forward to seeing the new L/H column in action in our next meetings!

My comment refers to using the 62:x and 88:x figures to segment races and build separate handicapping and betting strategies. Those percentages are based on the AI line/probabilities.

Good work once again from Dave and Brian. Look forward to expanded explanation!

Mostly Brian.

I did a study on Brian's "RSE" object. This study uses 2025 races with 0 FTS. I used the PickMaster in HSH to see how the six factors in the RSE object performed in these races. For the first run, I also added public choice and morning line to give some perspective. Here are the results after the first run:


Field Rank Starts Pays Pay% $Net IV PIV ExW PExW
1 rPubCh 1-1 2,746 1,041 37.9 $1.73 2.84 1.04 367.1 997.5
174 rPSR 1-1 2,913 888 30.5 $1.61 2.29 1.00 388.1 886.8
176 rML 1-1 2,697 878 32.6 $1.64 2.44 1.01 360.3 872.8
82 rFTB2L3 1-1 3,008 777 25.8 $1.64 1.87 1.01 396.2 731.6
1625 rtSC 1-1 3,306 701 21.2 $1.56 1.60 0.99 439.0 711.6
1632 rtClass 1-1 3,600 699 19.4 $1.57 1.46 1.00 478.3 700.1
1633 rtAPV 1-1 2,728 548 20.1 $1.70 1.50 1.01 364.3 542.9

I chose PSR R1 because it had the highest number of pays (other than public choice). I used it to make a rule. When making a rule using the PickMaster, you eliminate the races in which the #1 ranked horse in PSR won. Then you run the PickMaster again to see which factor has the most pays/winners in its first rank in races in which the PSR #1 ranked horse didn't win.

After the second run, I made a rule with rtClass since it had the highest number of "Pays". Even higher than public choice!

Field Rank Starts Pays Pay% $Net IV PIV ExW PExW
1632 rtClass 1-1 2,616 378 14.4 $1.56 1.10 1.00 344.0 377.9

Then I made a rule with FTB2L3. FYI: Public Choice had 274 "Pays".

Field Rank Starts Pays Pay% $Net IV PIV ExW PExW
82 rFTB2L3 1-1 1,303 229 17.6 $1.65 1.37 1.06 167.2 216.3

Next rule:

Field Rank Starts Pays Pay% $Net IV PIV ExW PExW
1625 rtSC 1-1 1,262 163 12.9 $1.45 1.00 0.95 163.5 171.0

I made a final rule with rtSC.

Field Rank Starts Pays Pay% $Net IV PIV ExW PExW
1633 rtAPV 1-1 932 114 12.2 $1.83 0.95 1.02 120.3 111.8

Then I was left with:

Field Rank Starts Pays Pay% $Net IV PIV ExW PExW
78 rFTLR 1-1 689 91 13.2 $1.64 1.04 0.93 87.7 98.1


What's the takeaway here? In this 2,683-race sample, horses that ranked first in either of these five factors won 1,863 races (69%). That means that 31% of the time, the winner did not rank first in any of these factors.