Modeling

[Giahorse]

Originally Posted by Ted Craven

Gia,

Just to establish terminology - let's use 'model' when we mean a Decision Model (or Brohamer Model) which is constructed using the paceline from the PPs you used to represent the horse (e.g. on the RDSS Analysis screens).

And let's use the term 'Profile' to mean info drawn from the Result Chart of the winner of a race (you didn't have this info before the race).

I've seen Bill's Profile/Model - it is a hybrid of the 2.

Your question - if you organize your Analysis screens on a consistent, recurring basis so that there are 5 horses in the Primary section (and the rest are in the Secondary- or Non-Contenders section - OR, are eliminated altogether, which I would not recommend), then the Export to Excel will show rankings for various factors based on a consistent 5 horses (and not 14, or 7, or 11, or varying field sizes, etc).

If you downgraded the Winner to Secondary (or eliminated it) - you should still include it in your model. You need that race to include in your total count in your Model, otherwise your model inaccurately over-represents the impact value of factors. You may have 8 of 10 winners at 6f at Sartin Downs top 5 in 'Factor OatsPeasBeans' or 80% but if you didn't include 3 winners who were ranked worse than 5th, the measurement is really 8 of 13 (62%).

The new Model Template provided (MasterTemplate2), calculates % for a given rank factor in Ranks 1 - 4, Top 2 and Ranks 5 or worse. You need to include your horses who finished 5th or worse to give these models meaning.

The rank of 5th or worse may mean that that winner truly ranks like that on a given factor, OR often it means that you misjudged an appropriate paceline, OR that the race was rather inscrutable (FTS, Foreign horse with no rankings, lightly raced horse trying something new - just 'popped up' today), etc.

Above all, you should try to develop a consistent set of line selection methods, so your models of the past are meaningful for future races. Over time, you can analyse your '5th and worse' ranks to see if they tell you something - such as, paceline inconsistencies, certain types of races, times of the year, age groups, race conditions, etc.

HTH,

Ted

Thanks Ted

I understand the model part, what about a profile, how is that done, do we just take the result chart data and do what with it, ?

[Bill V.]

The information you get from the result charts
is used for modeling position and beaten lengths at the points of calls.

At most tracks in the USA you will see that horses in position 1 2 and 3
at the first fraction and second call will win at a very high percentage

It is also helpful to know how far back a horse can be at the calls by beaten lengths. Equally important (for me) is to model if a horse can win at each distance based on this information. I also keep track of the visual running styles of the winner, I model many other factors from the result charts but that information is just supplemental, and really is used out of curiosity

I use the result chart, POR and beaten lengths to calculate my own
variation of the Phase 1 readouts plus F1

Bill

[Segwin]

Thanks Ted.

Both you and Bill are very patient and understanding.

Good stuff.

[Ted Craven]

Originally Posted by Giahorse

Thanks Ted

I understand the model part, what about a profile, how is that done, do we just take the result chart data and do what with it, ?

You can create a hybrid Decision Model / Track Profile / Betting Record spreadsheet using the RDSS Export and Master templates as a starting point, then expanding them on your own.

Use the existing Modeling tool to populate a spreadsheet with the Winner's line for each race - this gives you the race id, conditions, age, sex etc and all the factors for the line you chose for the race winner. It also gives you empty cells to record the mutuels (which will be populated automatically again in the next release).

Then - using currently empty cells to the right of the currently populated ones (leave, say, 50 columns blank for RDSS to use in the future ) add in the columns for data you want to extract for the winner from Result Charts: position, beaten lengths as the various call points, and calculations you want to make from the Result Chart times and beaten lengths. You can record your wagers, payouts, bankroll balance and any trip notes or wager decision notes as well. You can 'Hide' spreadsheet columns you don't care about to bring the distant columns into a viewable area, or use macros to distribute the info over several tabs in the spreadsheet.

FWIW, automated Model and Profile keeping (from Result Charts) and various queries on that data, is absolutely going to be built into RDSS sometime next year. Record keeping is too onerous for many people - that's why they don't do it and thus they don't have access to the kind of info that a Decision Model or Profile can give them.

Meanwhile, you can 'roll your own' like Bill V does, and like many others do by this multi-step process.

Ted

[dirty moose]

If you you look at the "bris" PP's, at the end of each race it will should you the "track model". I'm on my phone so I can't attach a photo. Maybe someone else can show you.

The pp's there are also free if you make a bet at the specific track with twin spires.

[Giahorse]

Thanks,

The bris pp profile is ok but gives limited info, thanks to everyone that has commented.

[dirty moose]

Originally Posted by Giahorse

Thanks,

The bris pp profile is ok but gives limited info, thanks to everyone that has commented.

Yea I agree with you, but it does all the work for you =)
If you were to take the bris info and some of the info off a result chart, you could hybrid your own and make it just the way you like it.

[clore1030]

I'm still "rolling my own" Ted, just as I have for years. I just take the rankings off the BL/BL screen and input them manually. Boxes below are then updated to determine the weight of each factor that I track.

I prefer this as it keeps me aware, I'm reading between the lines as I input. An old habit from when I used to analyze TV programs based on Nielsen ratings.

I trim every field down to five contenders, most of it based on how a horse ranks through the various screens. Because I don't track races with too many unknowns (distance, surface, days off), or below 20K claimers, I'm dealing with a better grade of animal and it's rare that I can't get the eventual winner in my top 5 - it hovers around 10-12% of the time that I don't.

I look for situations where the top ranking is 40% of the time minimum, all this is tracked at the various distances.

For example, at AQU, I have CPR, FW and BL top rankers winning at a 50-55% clip at 7 furlongs. I just wait for the right odds as too many of these go off at 4/5.

I look for 5/2 or better and I can be enormously patient and also quite satisfied. Case in point - during the last Belmont meet, my first two wagers about a week deep into the meet gave me a 56 and a 25 dollar winner on the same day. I sat out the rest of the meeting although I handicapped every day.