T
tacker Agree with much of what you say. But reading articles written by official handicappers together with the odd video interview, encouraged me to do something that they cannot do in keeping the peace between trainers querying their horses marks. That is to produce predictions of future ratings of horses. In my day job I had to identify possible improvements, estimate the potential and figure out how much was worth achieving. So, it was natural to think this way when I started analysing 2yo performances nearly 60 years ago.
In that research, on the one hand I would look for ‘tells’ in how the horse actually ran, and on the other hand what it later achieved in future races (not necessarily nto). Raceform have a list of coded words and phrases to use in their comments in running reports – although some race readers use unlisted ones . Their ‘notebook’ comments are less structured but can add to the picture in my mind of how and why some action happened - as well as confuse? Analysing years of 1000s of races of horses in Timeform’s Black Books gave me extra vocabulary and a list of extra tell tale mid-race events. Early work was manual notes on paper; then cards. Methods had to be developed for quantifying collections of those words and phrases, that took many years of trail and error. Personal computers in the early 80s allowed automation and the first real practical applications – all calcs feed a routine that gives the win probabilities of all horses in a race with half Kelly staking. Its well calibrated in that 0.25 horses win around 25% of the time. Its less accurate at the two tail ends.
Its was this sort of approach that allowed me to conclude that Frankel was highly likely to be placed in the 2000 guineas after his first 2yo run. On analysing his 2nd run I was then pretty sure that he would be capable of achieving a rating that was well above the average winning rating for that race. Actually it was off my usual scale. I kept checking my calcs for a mistake. That was the result of a combination of evidence. But in some cases, a single clue can make a big difference between actual performance and potential improvement. An obvious example of that would be hanging left on a right hand bend, and vice versa. Lingfield 5f races come to mind with improvements of 20lbs + for some young 2yos trying to get round that tight bend. I tried recording videos and reviewing same when the first video recorders came out but too time consuming.
At last, I automated that bit of the method that was a big time saver. Everything is measured in lbs weight; could be in seconds, but at least the units are consistent across the board. This may all sound an odd thing to do, but its what problem solving practitioners in statistical science do all the time, just the appliance of science?
Apologies if I’ve already posted bits of this before, it’s just that tackers posting sent me off on one.
TheBluesBrother Figs confirm that potential ) On the Class Sheet you will see a column marked TBB theses are the Speed Figs that are
