Some dangerous questions there Goober and I fear it might involve a lengthy response by way of explanation
However, I'll try and keep it short but can obviously only speak for myself.
The first one regarding the revision of ratings is quite easy. You just deny the original existed and hope to hell no one remembers.
If they do, you blame someone else or an arithmetic error.
If I've read the question correctly, the issue of the one of the good horse smashing a standard on a card of moderate ones isn't too hard. As you say you can use medians which is what I believe Dave Bellingham does. I'm not totally convinced that these aren't prone to even greater osciliations myself, especially in jumps races where you get a number of eased down performances which are more likely to be found in the midfield finishers, or small fields.
Personally I use a class par method where by each race is allocated some time back depending on its class. It's designed to level the playing field and isn't too far removed from a golf handicap in principal. The class par equates to a number of seconds against a standard unit measurement (I work at a mile) that a horse is given back to its time to standard. This is adjusted and used to set going correction (or track variance in American) which all speed ratings are derived from. The class pars are as I understand it, is based on averages of thousands of winning performances put up in different classes of race. Again, as I understand it, the par time isn't necessarily that important, so long as the ratios between the class divisions are the same, as you'll still generate the same rank order.
It is this class par assignation that has principally been responsible for my observation of so called grade 1 staying hurdlers being questionable, as they never hit the grade 1 par, yet their 2 mile contemporaires do. The par itself is indexed to 100, with a score in excess representing the better performance.
It is therefore quite possible that the relatively best performance on a card might be put up in handicap, rather than the grade 1, as the par makes a concession to the class of race thus by way of compensation. Once variance is established though, you can assess any horse off any class par, provided you treat all the horses under investigation to the same treatment and have a level of consistancy thus. Such an analysis for instance allows you to assess a novice off an open grade 1 par.
It's not fool proof of course, but it tends to help you spot a stand out performance early, especially in the event of 2yo's, novices and juveniles, and especially if introduced on a minor track where they will be given a lowly class par which they will normally shatter. The table I use (or should that be nicked) differentiates between minor and major tracks in view of the likely calibre of animal that such tracks attract, and every now and then you will encounter the emergence of a top class animal on a minor track. Their contribution to variance can have the effect of raising the bar however, so it is often better to use the second horse home to set variance in this event. Indeed, Twist Magic was first spotted at Fakenham thus, and backed at 40 odd to one for the Arkle.
As regards setting standards, I don't do my own, and am lazy enough to use the RP's. The standard can as I understand it be affected by a series of standout performances, and this is more likely to occur in a disruptive fashion on a minor track. There are certainly a few standards that I'm far from certain are correct, but so long as each horse is subjected to the same yardstick, the same rank order should emerge.
It's probably more helpful to let someone who does calculate their own standards explain it though as I'm sure they can do a better job than I
:suspect:
