suggestions for updates - your opinions requested

[motorhead]

hello all

here are some ideas that i've accumulated for the robot

- ability to adjust purse size like on the modeler, as opposed to check boxes for 10K plus 0r less
if not, at least a check box for 'less than 25k' so 10k-24k can be separated

- sole remaining part of an entry

- 1M workout (watched the NYRA tv show with richard migliore and he said he likes to see this)

- trainer has re-claimed horse

- beaten favorite as a negative

- separate all distances less than 5F

- separate all quirin speed points

- longest winning streaks and losing streaks

- kline results like ml and tote

- eliminate selections produced by the desktop that are from races/cards that have been cancelled

- entered as 'main track only'


any/all input is welcomed

thanks

[km]

Vin, thanks for your ideas !

(1) good idea for Robot, purse range

(2) scratches take care of this, dont 'they?, you want to mark "e" or something for horses that are coupled? filter them separately? i'll pass on this, would not produce any impact IMO.

(3) 1mile workouts are too infrequent as a filter for Robot2, but they are definitely included in the formula for Wk rating and graded in the Clocker report and shown in PPs with a grade based on 365-day workout comparison

(4) a good one, reclaim

(5) you mean "eliminate Beaten Favorite"? I'll test it to see if it has any value, My recollection is that is a neutral factor with little up or down side. If I'm wroing, it can be added.

(6) lots of Qpoint in Robot2 now, don't want to disrupt current spot plays that use those filters and don't want to add a new set. Sorry, "no" on this one.

(7) don't understand "longsest win/loss streak" or how it could be used as a filter?? Take the longsest from each race? That would produce a multitude of ties. Chronic Loser is a losing streak of 11 or more. "Loves to Win" are horses that win 40% of their starts.

(8) Possible (KLine spread)

(9) just delete them on your own before you run the spot plays, i have no way of determining that a card has been cancelled from the data file.

(10) (M) horses in a special filter? interesting, i'll check it out. You can eliminate them because 100% are found in "OFF TURF" races which can be elminatedd in Robot2 now.

[motorhead]

ken

"longest streaks" wasnt intended to be a filter but a result

knowing the lengths of winning/losing streaks may be able to give a better handle on possible long term results

also

regarding 'last remaining part of entry"

how many times have we seen one half scratched and the other half run a great race?

(like the $28 winner at CT last night)

i just wanted to see if there was a way (and, hopefully, a benefit) to singling out those horses to see if there are profitable times to use them

could be a window into trainer's intent (and could be nothing)

v i n

[km]

You thinking outside the box for sure Vin, we appreciate the discussion.

re: lone coupled entry after scratches
Trainers are manipulative to their advantage. Will test in next newsletter and see if it bears fruit in broad sample. My conspiracy theory along these lines is that top trainers with two UNCOUPLED entrants in the same race, tend to show a much higher ROI with the longer odds of the two. This is difficult to prove or test as it would required specific trainers and race circumstances to uncover along with a comparison of the odds in the data sample.

re: losing streaks

If the horse is a maiden, we can infer the loss streak from the lifetime starts. I have tested chronic loser maiden streaks in the newsletter and the notation "Chronic Loser" is prominent in HTR. It is also applied to non-maidens, most of whom have lost 10 in a row. The Light Bulb also has notation "Hasn't Finished 1st or 2nd in a Long Time", applied to horses that show no wins or seconds in the 2 year money box.

However, regarding non-maidens, our data files only contain last 10 PP lines. Since i don't have access to older race lines in the Racefile, i can't count the losing the streak for a non maiden for you.

Winning streaks are far far less common and rarely exceed 3 in a row. They are so obvious from the PPs, that I don't think they merit a special notation, the odds alone will get our attention as horses that have won 3 in a row or more are universally heavily bet and always end up on top of the rankings unless it is a stakes.

[motorhead]

hi ken

thanks for the response

re: streaks

its not about the horse, its about the spot play

when i get the results of the play, lets say on 'learn tour/special", i'd like to see how long the worst losing streak(s) were for that spot play

longest winning streaks would be nice

but i would like the reality check of how bad the losing streaks were

if we can see how bad the losing streaks , we might have an idea of the long term success (or lack there of)

especially when we are not dealing with a large sample

[km]

re: Spot Play streaks

A tally of streaks during an entire 365-day spot play run by the Robot would not be accurate as the Robot doesn't process races in exact order of 'off time'. It processes by date going backward, but goes in alphabetical order by track on each day. So it cannot give a realistic streak assessment.

The Last 100 report or the 8-week report may be the best bet for this type of assessment.

[motorhead]

i have used it

good feature

btw

of the things on my list of suggestions, being able to adjust purse size is number one for me

several of my spot plays are "less than $10K" and the number of plays each year is declining

(using the epr on races over $10K+ doesnt get the same results)

v i n

[km]

Consider it done on the Purse Value filter, Vin. Next update for Robot2 in January.

I like your ideas on some of the other filters, especially reclaim.

[njcurveball]

From personal experience, I have seen days where I have 35 spot plays and 5 of them win. My average with price horses is around 16%, so it was a pretty average day.

However, the day went like this, 3 wins to start, 2 wins to end, and 30 losses in between.

I have found through personal experience that streaks can never be counted on no matter what the perceived win percentage is. When I was playing Dr. Z Bets my win percentage was well over 60%, yet I could lose 7 or 8 in a row on any given night. On "paper" that is nearly impossible.

If only Math were as relevant to betting horses as knowing the trainer's name. This would be a much easier game.

Jim

[Candybag]

NJ:

Amen on the win/lose streaks. We need to look no further than the roulette wheel and that ingenious board that shows that last 10 or so numbers that have hit. Certainly if black has hit the last 7 straight, then red is due. Vegas loves that board and the people that follow "due" logic.

I know Motorhead's thought process is deeper than "due" logic, it is worth saying that win/lose streaks are precarious territory even with the deep thought process on display by Motorhead.

If I may borrow a famous stock market quote and apply it to handicapping and streaks.....the horses can stay illogical longer than I can stay solvent. :eek:

Great thread folks!

[DanG]

Over large samples this simple excel formula does remarkably well predicting potential run outs imo. I use it to set the starting bank for each spot play and it’s held up ok for what its worth. (In cell B2 just enter your projected win% in decimal form and cell B10 will show the theoretical run out.)

[Inglewood Flamingo]

Seems like this would be simple enough to implement and makes a lot of sense. I've been tracking it manually and it seems to increase the win percentage.

[km]

I.F. can that be tested with the current Robot2 setup?

* RAZOR SHARP

* LAST WORKOUT (at today's) TRACK

It would probably catch 90%-95% of what you seek. Sure there will be a few that worked offsite for their Razor prior to another workout over the track, but that will be rare.

I could look into it and see if applying the specific "Razor worked over today's track" has an impact over the above. Thanks for the idea.

[Candybag]

Ken:

I know that you have discussed not adding negatives as a positive on the Robot 2 reports. On the negative Robot 2 screen, please consider the opposite of allowing the subtraction of what normally would be a positive from the negative screen.

For example, late kick, bid but hung, hot trainer, first time surface can be positives that have been very negative on some of my spot plays. Either those positives are over bet or not compatible to the spot play emphasis at hand. I do not know how to eliminate these positives with the current format.

I note some positives that are allowed to be subtracted on the current negative screen, morning line favs, second favs, last out winners, etc.

Please consider allowing the user to subtract bid but hung, hot trainer, first time surface, late kick and any other that you deem helpful from the negative screen.

Thank you for your continuing consideration!

[km]

Candy, here my answer is in a nutshell=

1) Providing for the yin/yang (filter/elimination. pos/neg) of every item would require a complete overhaul of the current Robot. It is too much for me to program into the current version.

Yes, the Robot has a limited scope, but that's a tradeoff for a tool that can be used 'right out of the box' with no training and provides powerful data report generation. The other tradeoff with using Robot2 is the speed, it's far slower than a database query, but again, that is mitigated by the reports (Learn All, ROI Dump, etc.)

(we do provide 100% of the Robot data in the Export files to use in db query for any conceivable permutation of the data.)

2) The majority of the time when something assumed positive turns up negative, or vice-versa, there is an underlying reason for it that can be found with additional testing. My experience is that the something on the LEARN MORE screen (range of data) turns up the culprit, such as the odds range.

This is one of the problems with the 'ROI Dump' report. It's great for immediate eyeballing of good and bad, but loses some context that might explain it. For example, I often see "Bad Jockey" with a positive ROI. Ok, now I would like to use "Bad Jockeys" in my spot plays to make money :confused: I don't think so. A glance at the Tote Odds in LEARN MORE shows why the "bad Jockeys" make money, they are always > 5/1 odds, and that's the real reason why the spot play is profitable, the overlay price, not the jockey.

[Candybag]

Ken:

Thank you for the prompt response. Robot 2 is very user friendly, access is beyond my abilities. Perhaps one day a HTR member will hold an Access workshop regionally if there was enough interest.

Thank you and good luck with the BC.

[njcurveball]

Perhaps one day a HTR member will hold an Access workshop regionally if there was enough interest.



If I had a best friend who wanted to learn Access I would not put them into a Workshop "cold". I work with Oracle tools and databases and still need documentation to get around in Access as I do not use it on a day to day basis.

Like anything else the more you use it, the better you will get. But there is a learning curve that involves reading and some trial and error. It is also good to work on some databases that are just "throw away" in the beginning.

For example you may start out with a simple table which has data for your DVDs. You add columns such as Title, Genre, Actors, running time, year made, etc. Then you create forms to add the data, create reports and queries, etc.

Look at this like Golf. You are going to have to put a lot of practice time to learn all the clubs, what they do, how to use them, etc. But after a while you know exactly when to use each club and what to do with it.

Yes, it will take time, but in time it is all well worth it.

Jim

[Candybag]

NJ:

Thank you for the overview. It sounds like I will be using Robot 2 as Access seems a little too complex for my limited abilities.

I did see an elephant using Access the other day, but his laptop had more memory and no mouse! :)

[njcurveball]

I did see an elephant using Access the other day, but his laptop had more memory and no mouse! :)

Zing! Then there was the one about the lady who bought Olympic Condoms but kept giving her husband the Silver instead of the Gold, because he needed to <everyone groan> come second. :D

[zimal2]

Just caught this thread and am interested in the discussion about streaks which seems to be about horse or spot play. In Europe (France I believe) the limited info always includes how long it has been since a trainer last won. Since trainers do seem to win in streaks at times that's an intriguing stat.

[DanG]

Assuming a majority agree… :)

With many of the ratings also in HX4, would it be possible to get most of the HXD export in a decimal format? PAC, PER etc it would be helpful not to dig into HX5, HXV etc to get the decimal version.

[km]

Zim, you can get the trainer's streak and last win by selecting the trainer name in Robot2 and running Last-100 report (assumes you have a full slate of data for the last 3 months or more)

Dan, should be no problem to promote HXD to decimal, let me look into why i didn't do it in the first place, perhaps to keep the Differentials manageable.

[tomcat]

Howabout TRN ML>= 8? as in TRN button.:)

[km]

Easily done right now in Robot2 Tom. Select the trainer from the Connections filters and then set the ML Range Filter : 8.0 - 99.0.

[waytogo]

This would be an addition to the whole program and may be too difficult to add.

I have had a torrid time over the past few weeks and checked my daily paper trail. This may just be an anomaly, but the tracks lately seem to have wet one day then fast then the rain comes again etc.

Most of my really bad results have come on "fast" tracks where they were Muddy or Sloppy the day before. In other words the track was still drying out.

Of course if there is no previous days racing we cannot check this, so three codes would be required:

W: Prev Day was Wet
F: Prev Day was Fast
N: No Racing Prev Day.

For some this may seem a tad silly, but for those of us who have visited the track on a regular basis know that a drying track has not usually regained it "fast" qualities such as early speed and could be a quantifiable reason for a lot of failures.

[km]

I think the (N) would be the most interesting of those, wtg

The others would probably have mixed and indeterminable results.

There is nothing i can do with our Racefiles or Robot2 to help you though, you'll have to set it up in database.

There is one possible way to get at the (N) days in Robot2. Use the first day of the racing week. For example BEL, select Wednesday only, they never run on Tuesday and rarely on Monday.

[waytogo]

I think the (N) would be the most interesting of those, wtg

The others would probably have mixed and indeterminable results.

There is nothing i can do with our Racefiles or Robot2 to help you though, you'll have to set it up in database.



The only real interest for me there was the drying out tracks (prev day wet). This would certainly be track ( location ) dependent. At the Mountain in Dec with likely temps under 40 degrees it would dry out a lot slower than Retama, Tx with likely daytime temps of 80.

Thanks for the reply.

For those interested in access then I would suggest a make table query/module of the HX4 table for research purposed then change it values to an append table with the same primary index as the HX4.