First of all, yes, I place far more weight on proven horses who run in tight figure ranges.

The point of the expo presentation is that any combination of factors can cause a change in track speed-- if they harrow it between races, if they don't (which happens at some tracks), if they water it at a pace faster than it evaporates (gets wetter), at a pace slower or not at all or intermittently (gets drier, or drier then faster when watered again-- or maybe not, depending on soil characteristics and moisture content at the start of the day).

You can use any working hypothesis at all-- that it stays the same, or that it doesn't. But then you test it out by looking at the figures you end up assigning using that hypothesis-- and it is wrong to ignore that step because 1) the scientific evidence says so (see the expo presentation) , 2) the basic premise of doing figures this way is in fact that past figure histories are a good guide to what they will run-- if that's not true not only can we scrap the entire method of making figures, but it would make no sense to use them to bet, and 3) I say so. At this point I'm pretty sure I've done more track days than Ragozin, Friedman and Andy combined ( about 10 tracks a day for 22 years), and it is absolutely clear from working with the data that the data base comes up much tighter when you do it this way. Andy, by the way, came to the same conclusion independently. -- the only one who insists upon a totally dogmatic approach based on assumptions completely unfounded in science is Ragozin. And if you don't want to take my word for it, you still have 1 and 2.

And this brings us to the question of tightness, or "pairing up". There are a lot of guys who have made their own figures who post here (I'm only aware of one guy who uses Ragozin who made his own, and he only switched because he's pissed at me for throwing him off the board), and every one who has made figures knows that the tighter your range-- the more horses run back to previous numbers-- the better, in terms of evaluating your accuracy. Again, the relationships within a race are frozen-- so if you have a way of doing the race that gives lots of horses figures in the range they usually run, it is far more likely to be correct than doing it in such a way that 5 horses, who were only 20% to run a number much better or much worse than usual, do so-- work out the parlay on them all doing it at the same time.

Which in turn goes to a comment that Friedman made at the expo, which I only caught when I recently watched the DVD. He said in effect that it was wrong to "give them what you want" because over the course of time, all random distributions are possible, and will appear. Aside from the fact that you can only give one horse per race "what you want" (the relationships being prescribed, see above paragraph), he's right-- but only BEFORE the race. After the race, when we sit down to do the figures, there is only one "distribution" possible, and we can only move the WHOLE thing up and down (faster/slower), looking at the various possible scenarios trying to figure out what happened. In practice, once you have a tight data base, the right one usually jumps out at you for about half the races in the card on your first pass, and the others have only two scenarios, occasionally three. Once you have done the obvious races, though, you have some idea of the "shape" (LF called it 'texture") of the day, and it helps in working through the other scenarios.

I had math (among other things) forced down my throat when I was a kid, and although I skipped a couple of grades I eventually rebelled and dropped out of school, so my formal math background is limited. I have been told what we do is regression analysis, and if you want to hear a great description of something similar in a completely different context, check out the last act of Tom Stoppard's "Arcadia". When I heard it the hair on my neck stood up.



TGJB