A Gp1 winner on G/F will always run a faster time than one that wins the same race on soft. That doesn't make the horse that ran the faster time superior. If I get time later I'll dig a hypothetical example out for you to crudely illustrate the principle behind it.
As promised;
We're dealing with a very hypothetical couple of cards here. Assume every race is a Gp1 race, all horses carry the same weight, and every race is a mile long to illustrate things, the list of times are set against a standard.
Card A
Horse 1a = +2.00
Horse 2a = +2.22
Horse 3a = +2.65
Horse 4a = +2.03
Horse 5a = +2.75
Horse 6a = +2.34
Card B
Horse 1b = -9.00
Horse 2b = -3.75
Horse 3b = -1.50
Horse 4b = -4.25
Horse 5b = -3.99
Horse 6b = -4.05
In terms of raw time Horse 5a is the fastest at +2.75. In reality though horse 3b is the fastest, despite running a slower time than all of the horses on card A.
This is all to do with with variance which are the conditions that the horses raced under. There's no hard and fast rule about how you calculate it (i routinely omit the slowest two races to minimise my chances of encountering a slow tactical race that might otherwise contaminate the calculation). In this instance horse 1b would be one such example.
With two slowest omitted you can then add the others together and divide by the number of races (4 in this case). What you get is a figure that indicates the degree to which the ground was either assisting or hindering a horse. The average for card A is +2.49 and card B -3.31.
That is to say the ground on card A was speeding horse sup by 2.49 secs per mile, and slowing them down by 3.31 secs on card B. (+2.49 is about Good to Firm, Good in places, whilst -3.31 is Soft).
With these variances now known you can assess the horses by either taking the time away in the case of a positive figure, or giving it back in the case of a negative. The fastest horse on card A ran +2.75 on ground that was assisting him to the tune of +2.49, which means that 2.49 secs of his performance was down to the ground, and +0.26 secs his ability. Although I use the American scale of 0.17secs per length, the BHB scale is 0.20 secs for a flat horse, and 0.25 secs for a jumps horse. That is the amount of time it takes a horse to cover its own length.
So you divide 0.26 secs by 0.20 which equals = 1.3. For ease I use 100 but it doesn't matter what figure you use so long as its the same reference point for each horse. In this case it's a positive rating, so you add 1.3 to 100 for a rating of 101.30. Horse 1a is a negative rating;
2.00 - 2.49 = -0.49 /0.20 = 2.45 - 100 = 97.55
Now as I said, 3b is the fastest horse. The ground was slwoing horses up by 3.31 secs, yet 3b only ran slow by 1.50 secs suggesting that 1.81 secs of the performance was down to ability.
3.31 - 1.50 = +1.81/ 0.20 = 9.05 + 100 = 109.05
Horse 1b would be rogue element as he was particularly slow which indicates that he probably possess a change of gear to win a Gp1 race off such a slow time.
3.31 - 9.00 = -5.69/ 0.20 = 28.45 - 100 = 71.55
Hope that helps? The variance calculation is the root of all speed ratings and it can be very useful at certain festival meetings as it allows you to monitor the ground in running and alter strategy accordingly. Some courses are notoriously misleading, and I 'd prefer to let the horse tell me what the ground is like rather than a jockey, trainer or clerk
Phil had had other horses with 'rebel' in the name eg Rebel Rebel [as "I'm a bit of a rebel see"] and was also born within the sound of Bow Bells
Might be part of the reason Cockney never got the credit he deserve i.e. not a very fashonable bloodline or connections.