There was some discussion on the SmarterSig email forum today about the merit of horses running in races they had won or perhaps run well in the previous year. There was anecdotal evidence among the discussion that NH racing was particularly susceptible to these kind of trainer patterns with NH trainers being more creatures of habit perhaps. Its not too difficult to take a look at this although being 100% sure of matching races is a little tricky as race names can change from year to year. One thing we can do is match on the track, month, race distance and in the case of NH racing the race type eg Hurdle or Chase.
Applying this to NH racing yielded the following results to BFSP before commission. These results were without deletion (we will come to that in a while), in other words, horse winning a certain race were checked not just in the following year but in subsequent years as well. The test data was for 2009 to 2013
Bets = 1771 PL = +49.05 ROI = + 2.76%
Restricting just the front of the market at BFSP < 10.0
Bets = 1127 PL = +81.72 ROI = +7.25%
Now if we do a similar check for Flat racing
Bets = 6228 PL = -185.4 ROI = -2.97%
Bets = 3690 PL = -135.82 ROI = -3.6%
Now returning to NH racing and removing a horse from the list the first time it runs in the ‘corresponding race’ again, we have
Bets = 1548 PL = +115.3 ROI = +7.44%
Bets = 1057 PL = +96.4 ROI = +9.1%
We have 10 qualifiers running on Boxing Day.
Kempton 3.10 Silviniano Conti
Kempton 2.35 Faugheen
Font 2.5 Beau Lake
Hunt 3.0 Tophorn
Mark 12.25 Kings Road
Mark 2.45 Chestnut Ben
Sedg 3.0 Discoverie
Winc 3.25 Minella Definelely
Winc 3.25 Quite By Chance (not this race but won at track in December)
Market Rasen 2.10 Soudain
It would seem that in the NH game Christmas does not come but once per year but potentially every day of the year.
Merry Christmas every one and a prosperous betting new year.
FootNote – Having produced the qualifiers for Boxing day it highlighted a fundamental flaw in my code. I had forgot to check that any horse running over same course etc in a given month was not simply duplicating a win from earlier in the month from the same year …. DOH!.
This is a pretty important mistake as well because the same year qualifiers earned the lion share of the profit for NH racing as can be see below
Wins from previous year –
Bets 1133 PL + 42.4 ROI +3.7%
Now wins from same month same year –
Bets 466 PL + 57.2 ROI +12.2%
The flat showed a similar bias towards same year repeaters rather than previous year repeaters but neither were as promising as NH racing. The same year repeat winners basically broke even whilst the previous year winners lost 7.6p in the pound (-7.6% ROI)