If Madison has improved, for whatever reason? (cheekpieces etc) then I think it might well be dangerous to rely too much on his previous performances as a pointer, as the horse might need reassessing for performance patterns in light of the improvement and a new set of parameters drawing up based around when this improvement seems to have occurred onwards. I'd have no problem with that logic.
I'm still struggling to sustain the idea that the 0-135 chase was a quick pace though (no I didn't watch it, I was laughing at England trying to an impression of a rugby team). However, if you break the raw times down into mile aggregates the evidence wouldn't weigh on the side of your hypothesis.
Hohlelthelonely = -7.75
Tartak = -8.56
Madison du Berlais = -9.23
Pancake = -10.26
Pause & Clause = -12.02
Strawberry = -12.93
Bally Conn = -13.33
Now it's normally the case when split going is given that the hurdles course rides half or a full description slower than a chase course (although Kempton hasn't been without its well documented problems in this area to do with true distance). Indeed this appears tohave been the case yesterday with the chase course given as Good to Soft, soft in places and the Hurdles course as Soft, Good to Soft in places. The RP haven't given a hurdles going description out incidnetally, so I've had to use the Sporting Life. If anything though, you'd expect the chasers to be benefit, yet two of the hurdle races ,in what should have been weaker affairs than a 0-135 (one was a 0-115 handicap, and the other a novices race) also ran faster. My own going allowances (which would be coming under pressure admittedly for the smaller sample) work out thus;
Chase = -4.54
Hurdles = -4.32
In this case I wouldn't be too concerned about using times across the card to set a variance figure, (though obviously I do split them from time to time).
Now you could attribute the faster times than Strawberry to the respective class of the horses involved as otherwise you'd be comparing apples with pears. But surely this is why we use a class par? They may not be ideal, but they're better than anything else available to us, although I'm still not convinced they're as good for jumps as they are the flat. When I've given them the appropriate time back at the mile aggregate for the class of race then I get the following;
Hohlelthelonely = -1.15
Tartak = -1.36
Pancake = -3.06
Pause & Clause = -4.32
Madison du Berlais = -6.03
Bally Conn = -6.73
Strawberry = -7.33
It's this that would make me nervous about using the evidence put up by Strawberry as despite what you think you might have seen from the armchair, your eyes can easily decieve you. Indeed, there was some experiement done about gauging speed that involved jockeys being told to ride horses to a certain speed. What it proved is the jockeys weren't terribly good at knowing how fast they were going, and couldn't estimate pace to the extent they thougt they could (they eventually stopped co-operating with the experiement). I can't remember where I read it? but it sounds screwy enough to be soemthign Nick Mordin might have dredged up. What tended to happen is that jockeys took their indications from the signals the horse was sending out, rather than any ability they had with their eyes to estimate against visual reference points. If they can't do it therefore, what chance do we have watching from television (even though a lot of us would like to think we can). What I suspect we can do is reasonably accurately gague extremes of pace, but once we start hitting the middle area of a spread of speeds, we struggle to differentiate between 2 or 3 mph here or there, which is why we use stopwatches etc.
In fact it was mordin, because I remember him going on to implore us to drive a car at 35 miles an hour on an open piece of road with no other vehicles around for reference, and only look at the speedometer once we thought we'd achieved this speed. Then increase it 37 or 38 mph and do the same. He contended that most of us would be more experienced in terms of miles driven then jockeys would be at miles raced, and yet would struggle to gague the speed we were travelling at.
I don't know, I've never tried it
There might very well be a case for Madison, but at the moment I'd say it lies in things like stats, some formlines (although others contradict these) and the notion of an inform improver. I'm not sure the evidence is there on the clock yet though.