Qp% and QT3

[Ken Massa]

Nice to see a lively discussion and interest in the QT3. I have also received no less than a dozen emails with positive remarks about the new item. Almost everyone realizes that this is not a pure analytical tool, but a signpost, much like the par.

The companion item to the QT3 is the "Qp%"
Qp% = Quirin speed point percentage

This rating is found in the HTR2001 velocity report and on the program screen. The number listed is a horse's percentage of the total of all Quirin speed points for the race.

The chart below is a rough guide to understanding.

31 +....... has complete control of the early running
26-30......looks in good shape to move to the lead if desired
20-25......solid part of the early picture-some pressure possible
14-19......may show something early-but will have company
00-13......not likely to be part of the front brigade

Two important elements are critical to understanding the Qp%. They are the QT3 and the field size. A larger field will dilute the points and the percentages. A 22 Qp% in a field of 12 is easily as impressive as a 32 Qp% in a field of 6.

Field size has no effect on the QT3. Yet, the lower the QT3, the more impact a high Qp% will have. For example, let's assume a race with 16 QT3; a runner in that race receives a Qp% of 28. This is the classic "Lone F" scenario as there is not much forward pressure expected.

Those are the obvious strategies and basic ideas. I expect we will learn much more from our users as they work with the concept and observe in day to day action. My feeling is that the Qp% will provide more interesting db information and spot play scenarios than the QT3.

A good query might be to test the lower end Qp% for elimination angle in dirt races. Or test those low rated Qp% for positive answers on grass. There has not been a great deal of research into Qp% and QT3 previous to now as no other vendor has offered it as an exportable function before. A pioneer effort into the depth of Quirin speed points - maybe Dr. Q himself will be reading here with interest!

[Glen]

Sounds good Ken. Can we also get rank of Qp% in the next export upgrade? Thanks.

[Carl]

Sounds "interesting".

(I like that word in that context. Means I haven't a clue yet, but hope I might have some day......)

[Rick]

Glen,
It look like it is already there as "nQpc" in the ALL_HX1 export.

[Glen]

Rick, is that the number of the Qpc percentage? or the rank? I thought it was the number of the percentage...

[Rick]

Your right Glen.

I guess my brain just ignored the word 'rank' in your post.

MikeDee has a module for doing rank, that is until you can get Ken to add it.

[Ken Massa]

Good question Glen -

The Qp% rank and the Qpoint ranking would be the same thing. The percentage is based on the points so if two horses in the same race had +6 Qp, they would have identical Qp% and thus the same ranking in both.

The problem with ranking the Qpoints is ties. Tons of ties! There is no rational way to break ties with the Qpoints. I have tested the Qp rank in the newsletter in the past and the results were listless in that context. But the higher the Qp, the higher the win rate for sure.

The ties mess up the testing if you export.
Suppose a field of 9. Three horses have +6, three with +5, three with +4 Qp.

This means the 6's will be ranked "1", the 5's ranked "4" and the 4's all ranked "7" due to ties in the data. The difference between the top and bottom here is not that significant in terms of Qp% but is big in terms of rank. Ties also distort the win% and ROI for obvious reasons of multiple plays in the same race when doing a query.

The Qp are best dealt with at the extremes where there are fewer ties. Do spot tests with horses +8, +7 Qp and those with 0 points too. Remember the zero is a legit value in the Qp data, the -1 is used to mark FTS, etc. in the export. When testing the other Qp values, add a factor such as Fr1 or C90 to the query so you don't get multiple plays.

If you don't find the secret to beating the races after all that, then I'll add the ranking for you in the next export!!

[Glen]

Rats Ken. You busted me...That was just another way of me asking if we could get Qpt Rank in there. Too smart for me. I'll have to catch you on an off day and try again in 20 years when things might start getting a little fuzzy...

I was just curious how Ev1 might do if it is top ranked Qpt horse in certain situations or how a Ev1 might do if it were ranked 3 or 4 in Qpts. Back to the salt mines for now...

[Carl]

Ken, what happens if you use ev1 ranking to break ties? As in highest ranked ev1 gets higher Qpt ranking.

I remember you were using jockey rankings to break some ties in the "old" program, and think you were happy because that put the "better" horse ahead 60% of the time or so. Wonder if ev1 ranking could do the same for Qpts ranking.

Glen, Mike will do you a calculated field if you want a number ranked (I think). He has that item in his storehouse of goodies, on the shelf even.

[Bernie]

I like Carl's tie breaker idea a lot. However, one of the nice things about speed points is that they are not paceline dependent, ie, they take into account multiple pacelines, while EV1 depends on which paceline is selected. Maybe a better tie breaker would be the early speed rating from Impact, nESP, that also looks across mutiple pacelines. What ya think?

[MikeDee]

Take a look at the number of ties you find in a race. You will find multiple ties sometimes 3 and 4 runners have the same % of qpts. Tie brakeing a tie once in while is one thing but breaking multiple ties in the same race..... in effect you are ranking the tie breaker.

[Ken Massa]

Mike's point is right on regarding the tie breaking.

While jockey win% is a proven tie breaker when two horses are tied for the same ranking (a 60/40 success rate) what do we do with 4 or 5 ties among the Quirin points? Use Fr1 to break em? Might work - never tested it. Will do it and put the results in January newsletter.

Ties are the dirty little secret to phony win% in some handicapping software. To inflate the win rate they create multiple #1 ranks in velocity type data by engineering a tie when the figures were within a certain figure, say 0.20 fps. This may sound rational in terms of separation, but what really happens is that there are 1.7 horses per race with #1 ranking and that turns a 28% win rate into 41% and that looks much better for advertising!

[Carl]

Originally posted by Ken Massa
Mike's point is right on regarding the tie breaking.

While jockey win% is a proven tie breaker when two horses are tied for the same ranking (a 60/40 success rate) what do we do with 4 or 5 ties among the Quirin points? Use Fr1 to break em? Might work - never tested it. Will do it and put the results in January newsletter.

Ties are the dirty little secret to phony win% in some handicapping software. To inflate the win rate they create multiple #1 ranks in velocity type data by engineering a tie when the figures were within a certain figure, say 0.20 fps. This may sound rational in terms of separation, but what really happens is that there are 1.7 horses per race with #1 ranking and that turns a 28% win rate into 41% and that looks much better for advertising!

Hehe to the second paragraph Ken. Sort of the software version of Little Jack's Green Card pick three horses to win in each race, SOMEBODY has to get there. "Little Jack, Little Jack, six winners again yesterday....."

Look forward to your testing of tiebraking factors, I liked Bernie's suggestion of ESPr1 better than mine even, as it wouldn't paceline sensitive. Although ev1 pl1 might be good too, as it moves the current form factor up.

Agree with Mike about "ranking the tie brakers" to some extent, but nothing wrong with a multiple factor number IF IT WORKS.

What would be the measure of "success" here? Would it be ROI/win percentage OR ability to get he lead at the first call?

[MikeDee]

I was wondering if there would be any value doing the other way i.e. taking a htr factor like f1 and weighting if with qpts. Although as I think about it this could just water down the fr1's as lower ranked fr1s' could jump ahead of the rfr1-1 runner if they had a big qpt number.

[hurrikane]

I'm wondering if the lower qp% fr1s perform less than expected anyway thinking they may not get a clear lead or do not typically run to thier fr1 rating.

[Carl]

My first guess would be the opposite (albeit by default, the rest of the horses would wait on them), but it is a good thing to test.

[pizzaman]

A review of gulfstream, churchill and philadelphia park results did not show consistent trends of QP vs QT3.

The study used the HTR-dos calculation of QP. The data was tabulated into several groups by distance, class and surface. Results are included in attached file.

This study did show that QP vs QT3 produced expected trends at Philadelphia Park for dirt sprints. However, these trends did not hold up at Philadelphia for dirt routes, or at Gulfstream or Churchill.

This data does show significant trends in the value of QP at Gulfstream, regardless of QT3. For instance, the results for high QP at gulfstream turf may well be worth further study.

[hurrikane]

Carl...my contrarian friend...good insight...worth checking both ends and maybe eliminate the middle.

Pizzaman....interesting...can you send over a large supreme while I look into this further. :D