But seriously, I guess this discussion boils down to what one thinks of as a speed figure. If you are really die hard old school I guess you would think that a speed figure should measure speed, and nothing else, and that every yard one runs matters just as much and that it is the time from A to B that counts. Then you can look at the results, times etc, maybe you have a par time chart or something \"shrewd\" like that and you assign a variant and you get maybe 80 % of totally worthless speed figures because they don\'t say anything about the relative performance of the horses, and they will also be extremely fragile to different pace scenarios, wind differences, errors in timing, and changing track speeds, things you can\'t control or be confident about. This is a kind of \"top down\" speed figure making that maybe was the logical approach 50 years ago, when everything \"modern\" was \"scientific\", normative and top-down. The projection method or regression analysis etc is the \"bottom up\"-approach, the opposite, where you use the horses to project the most likely scenario and frankly you could to this just fine without any variant at all (as long as your database is up and running and in the right ballpark, to get there is a different story). I think TGJB and Thoro-Graph does the right thing in general when they try to keep it some place in between, you use the horses, you try to generate a picture of the day as a whole, you (prefer to!) tie races together, you try to create a picture of how the track was playing out and you go from there. The more datapoints, the lesser the chance to get a race completely wrong, or to produce totally meaningless figures that can\'t be used to anything.
I think it\'s sad that one would still debate the premises of whether track changes speed, of course they do and it\'s obvious for anyone that try to make figures with an open mind. This is sad because these kinds of discussions hinder progress, because you for one Miff have raised many very valid points that really should be where the differences lie between the different figure makers, and what the discussions should be about. What values should one assign a path of lost ground and a pound of weight, and are there \"thresholds\" or whatever where this doesnt matter as much or matter more? What should one do with different pace scenarios both regarding to adjusted figures (if one can \"add\" or \"subtract\" to a final figure because of unfavorable/favorable pace scenarios), and to the values of ground loss (could one find a \"solution\" to the problem of slow pace on ground loss which would create inflated final figures?)? How should one use projection when you have runaway winners on the slop? I really think these are the issues where figure makers should differentiate oneself, and not on the crazy assumptions that tracks don\'t change speed or on a methodology that try to make \"best fits\" of a variant across a whole raceday, it\'s simply to much going on for that approach to ever make sense. The only reason those approaches worked 50 years ago was because before that everyone was betting on grey horses with number 7, or on horses of a different \"class\". Then those numbers represented an edge, but today they can\'t compete. What do you make of traditional speed ratings and the results they can show for in turf racing, doesn\'t that tell a story about flawed methods? Those guys essentially \"gave up\" on turf racing !
