RDSS BLBL Question
 

Should the BLBL score be independent of the other horses scores?

I've been playing around with a few races trying to get a better instinctive feel for some of the ratings and came across this odd situation...AQU 3rd today 1/22/17...

I've had Mark T with ranks 2 3 3 3 3 3 3 in the primary factors but with different combinations of horses (each with primary LS of 20), I've get BL Scores of 18.3, 19.3, and 20.3.

Does this make sense?

Well, no - ALL paceline dependent factors in RDSS are dependent on how a given horse compares (ranks) against the other horses in the analysis set.

Since BL is a weighed composite of rankings of the 7 Primary factors it comprises, let's take just 1 Factor: TPR (or CPR) rank: which is itself a sum of EPR+ LPR, then that sum ranked. If one horse has a TPR rank (check the TPR Tab) - say a 3, and another horse is ranked 2 (i.e. its TPR is higher than the the 3rd ranked horse), then - that rank 2 gets more weight in the BLBL calculation than the rank 3. Continue that process for all 7 Factors.

So, you can see, even without tinkering with which paceline you choose for a horse, horses' weighted BLBL numbers will be relative to each other. If a horse scratches, the BL may well change (the scratched horse with rank 1s and 2s will cause lower ranked horses to inherit those higher ranks and adjust their BLBL score). If you demote a horse to the Secondary group or Non-Contender group (or remove them entirely from the Analysis set - NOT RECOMMENDED) - same effect on the weighted ranking process. Etc.

BLBL is dependent on 1) which line you choose for the horse (or let the software choose) and 2) which other horses' ranks a given horse is being compared to.

Ted

I understand that, but I'm seeing some (what I consider to be strange behavior)...for example, a horse is ranked 2 3 3 3 3 3 3 for the 7 primary factors - I would think that as long as the horse remains with those ranks, the score should stay the same, but, as I change the horses that are secondary candidates, the score of the horse, whose ranks stay unchanged at 2 3 3 3 3 3 3, changes.

Also, I thought I understood that secondary contenders are ignored for the scoring of primary contenders, however, I've seen a case where scratching of a secondary contender changes the score of the primary contender.

Seems like most of the cases that aren't making sense to me are related to ties in rankings.

Perhaps some details? For example (using default chosen pacelines):

Attachment 42087

FWIW - Rx (from Testing V2.1)

Attachment 42088

Could you use your own chosen pacelines lines and Contender groupings to illustrate your question? Often much more illuminating than generalities.

Ted

I've managed to lose the choices that were causing the confusing I referenced above, but here's another example from the same race...

If you look at American Mink, in the both examples the ranks are 1 4 3 1 4 4 4. I can understand when the BL Score is 18.0, but it doesn't make sense to me that adding another contender would improve his BL Score. I don't see where those 2.25 extra points are coming from.

This is particularly "troubling" when you look at the BL Odds which have gone from 2.5 to 1.8.

Thoughts?

My
 

Question is, why are you using a turf line for a dirt race?

Bob

I'm not handicapping the race, I'm trying to get a better understanding of the of the readouts and the values.

Raceman
 

There might be a issue with your settings ,
and Raceman,Bob maybe on to something withhis question about using the 6.0 turf line. Maybe there are too many adjustments using such a non coparable pace line ? , I doubt that is the issue but lets see.

I use the full 50% DYV and I use Validator mode

Using your lines with my settings.

Attachment 42096


Now when I move the 2 from a secondary contender up to a main contender
the blbl looks like this

Attachment 42097

Now just for curiosity, I changed horse 3 to a dirt route -line 5

Attachment 42098

Now moving the 2 up

Attachment 42099

Bill,

Using your settings, I feel like the following lines are inconsistent with other calculations. These don't bother me as much as the logic inconsistency from my last post, but they still seem inconsistent.

Also, regardless of adjustments, turf lines, etc., I thought BLBL comes from the seven primary factors. Since we can see what the primary factor values are and what their ranks are, even if there were problems with too many adjustments, the problems should show up in creating the primary factors, not in the BLBL which should be using the already calculated primary factors as inputs.

Example 1:
Bailiv 3 1 3 4 2 2 2 = 20.5, should higher - compare to
Example 2:
Bailiv 4 1 3 4 2 2 2 = 20.3, usually shifting from a 4 to a 3 is worth more than 0.25

Example 3:
Mark T 1 3 3 2 3 3 3 = 21.0, should be closer to
Example 4:
Mark T 2 3 3 2 3 3 3 = 20.0, shifting from a 1 to a 2 is usually only worth 0.25

Here's another situation - same race...Mark T has the same ranks in both setups, but the BL Score is different...

Hi, I'll look at this further tomorrow. Could you post a screenshot of your Configure Tab settings, just so I can produce the same numbers as you.

Thanks,

Ted

This is the configuration I had for the last post I made in response to Bill's...I think the settings were different before that. I'll keep them here for the time being.

I don't use RDSS presently so take my view on this from that standpoint.
What I see is when you bring TWO DO up from the secondary contenders, it brings up a horse who is ranked 3rd in EPR and 2nd in TT. Notice that affects the readings of BALLIN lowing its rank to #4 and also affectin MO TOWN and MARK T's rankings as well.

That happens again within TT as TWO DO is tying in ranking with MARK T.

This has caused the 3 horses ranked above AMERICAN MINK to have have larger BL scores and not changing AMERICAN MINKS but AMERICAN MINK has closed the gap on the horses ranked above it making its odds ranking to be lowered.

I am not real good at writing what I am seeing but I hope maybe this might help.

Ray

Hi 'dread',

Just for the record, since Dr Sartin never published the 'formula' for weighting the Primary Factors to come up with BLBL score, neither have I. However ... and anyone doing this for 2 minutes will create their own answers:

1) clear Analysis then add 1 paceline: all ranks are #1 and the BL score is 24.5. Now you know the weighted value of rank #1
2) add another paceline: now some ranks are #1 and some #2. Knowing the points value of a rank #1, now you know the weighted value of a rank #2
3) FYI, BL Scores ending in .3 really mean .25 - they're just rounded up for compactness of presentation
4) and so on ...

In your example with 4 horses in the Primary Group, now you know how #3 American Mink got BL Score = 18.0

Take note of the Primary Line Score (PLS) of the #1 and #3 (20 and 21 respectively).

Now, when you add the #2 as the 5th horse to the Primary Group, because it has a couple of good #2 ranks, it bleeds a couple of factors' ranks from the #1 horse, who as a result now has a Primary Line Score (PLS) of 21 - same as the #3 had before and still has.

Now for the nuance. When 2 horses have the same PLS, they get the same BL Score. Regardless of the composition of the ranks in the Primary factor array, if they sum to the same PLS, they're equal in our assessment of an Oddsline. Conceptually, one horse may have longer ML Odds and 'look' dodgier, but if it's a fairly chosen paceline (i.e. it could be duplicated today given the horse's current form and competition) and is otherwise a reasonable Contender - well, that's how you get longer odds horses as Contenders on the BL Oddsline.

Now that the #3 and #1 horses have the same BL score (and same BL Odds line = 1.8-1), since the BL ties are broken by Total Energy, the #3 shows above the #1 - but they're really equal on the Oddsline. We say they're in the same 'tier'.

That's the answer to why the BL Score and BL Odds change when adding other horses. I didn't invent it - Doc Sartin did, and it has seemed to hold up remarkably well over the years given proper paceline selection and Contender qualification.

I am making no comment on pacelines chosen or other qualification of the horses - I just used your pacelines and horses to illustrate the mechanics. I know you're just trying to drill down and get a handle on the Oddsline mechanism, possibly to know if you can trust it. Fair enough!

The next step would be to understand VDC, since it modulates BLBL and incorporates deceleration. Then, if you are interested, consider checking out the Rx factor in the testing version RDSS 2.1 (just ask). To BLBL/VDC, it adds Class, Consistency, Early vs Late pressure, and an outside non-correlated Oddsline (either ML or BRISnet ProfitLine) into a further weighted Oddsline - the Rx Line.

Hope that helps.

Ted

Ted,

Thanks - that was pretty much what I figured - that specific situation was a rare enough and strange enough behavior, that I wanted to be sure it was working as intended.