_____________________________________________

  Data for 391 UK Horse Races Donated by 
  Raymond Batkin , 
  April 8th, 2002 ***
  ______________________________________________

  Hi Doug
 
  Further to your message:
 
  Please find, attached, FlatFile.txt which is a
  file holding horses that have run over 7.5 to 8.5 
  furlongs (110 yards either side of a mile).
 
  This file was produced by a program of my own (HORSES)
  written in Delphi and using a Paradox table (raw data).
 
  HORSES has places specific conditions on the filtered
  data (that ends up in FlatFile.txt) and this data covers
 
  1. Only races run on GOOD going
  2. Only the first 5 finishers
  3. Only horses finishing within 3 lengths of the winner
  4. Only horses that are in the first 6 of the betting
  5. Races between 7.5 and 8.5 furlongs
  6. Handicap races only
 
  The fields that are downloaded are as follows:
 
  1.  HORSE    horse name
  2.  RACETYPE race type (i.e. handicap)
  3.  DIST     distance of race (furlongs)
  4.  GOING    going conditions (GOOD in this case)
  5.  VALUE    prize money (GB£) to winner
  6.  GRADE    grade of highest rated horse in race
  7.  CLSRF    highest performance rating produced in this race
  8.  PLACE    finishing position
  9.  WGHT     weight carried (in lbs)
  10. WTPSN    position of horse in weights table
               (1 = top weighted, 4 = 4th top weighted etc.)
  11. WTDIFF   weight difference compared to last race 
               (i.e WGHT - WGHT_1) 
               so if WGHT=130 and WGHT_1=123, WTDIFF=7.0
  12. BTN      distance behind winner 
               (if PLACE=1 then BTN=0)
  13. DPRICE   decimal equivalent of starting price 
               (i.e. if horse was 9/2 then DPRICE=4.5)
  14. BETPSN   position in betting 
               (BETPSN=1 means horse was favourite)
  15. JC       the official Jockey Club rating that 
               the horse ran off in this race
  16. to 41.   items 3 - 15 repeated twice to cover the
               previous 2 races of the horse.
               hence DIST_1 and DIST_2 refer to the distance
               run in previous races etc. 
 
  This particular file has 391 horses in the sample so
  with the shareware I 'trim'it (randomly) to 250 records.
  Clearly in the conditions set down by the HORSE program,
  long priced winners (outsiders) are missed - you can see
  this in some cases where the first horse listed in a
  particular race has PLACE=2 which effectively means the
  winner was OUTSIDE the first six in the betting. 
  I hope you get the drift.....
 
  HORSES is very flexible. I can select what I like so I 
  could say 'first 50 in the betting' which would then trap
  long priced winners (there are rarely more than 30 horses
  in a race on the Flat - and this is an exception).
  My reasoning here is that approx 80 percent of winners
  come from the first 5/6 in the betting. In handicaps the
  prices are often better than 5/1 for the favourite -
  on the big handicaps like The Stewards Cup over 6f at
  Goodwood they can be 10/1 even - not for nothing is the
  'Glorious' Goodwood meeting called 'the bookie's benefit'!
 
  So if we set the target field to PLACE then we get a tree
  setting out all the conditions for PLACE =1, 2 and 3 etc.
  I just wondered if you had ever considered allowing TWO 
  (or even 3)  fields to make up the target.
  In the BEAGLE program it was possible to say, for example,
  (PLACE=1 & WTPOSN=1) which meant 'locate all top weighted
  winners' - but Beagle was an extremely finicky program to
  analyse the rules with and misconceptions were possible
  whereas VisiRex provides clear indications every time.
 
  However the (PLACE=1 & WTPOSN=1) case is a good one to
  look at because my betting is geared to looking for horses
  that are DROPPING in class which means they will carry 
  MORE weight - often they are top weighted. Handicaps offer
  the best value as a rule but one needs to try and get into
  the trainer's mind as he will probably run a horse in the
  wrong sort of race over the wrong distance and on a track 
  that is not suitable and on going that the horse has not 
  given evidence of form. You probably know all this but in
  the UK it is much more complicated as the tracks are all 
  so different...it probably explains why very few non-UK 
  trained horses ever win the Derby - the 12 furlongs at
  Epsom climbs 110 feet in the first 5 furlongs - takes a 
  U-turn and then involves a downhill/cambered stretch with
  a short climb to the finish. Not for nothing are the Derby
  Trials run at Lingfield (a similar layout) and horses that
  run at Brighton often run well at Epsom and vice versa.
 
  I think I have covered most points here. Let me know if 
  there are any queries. If there is a way of merging two 
  fields to make the target expression I'd be interested to
  know..as it opens up the data better.
 
  Regards
 
  Ray Batkin