Handicapping Tracks Differently Matter?

[InFront]

Does anyone really think that handicapping tracks, distances, class and surfaces differently really matter? In otherwords should you use a different handicapping approach and/or racing factors for a 6F dirt Claiming at PHA compared to a 8F turf Allowance at FL compared to a 6.5F dirt Maiden at SUF? And then to make things more complicated the way you handicap a 8F dirt today at SUF may be different how you handicap that same distance same track next week?

I been reading over some of the more popular software websites on how they use the ability to not only find which factors, methods, ratings, etc. work but which ones work for each specific track, class, surface, distance. And then adjust them frequently to catch a Bias trend that may be happening.

The problem with all these theories if you slice up results and data enough you will of course find "something" that worked. And if you use smaller recent samples of races you can easily find even more things that have been recently working at each track. But what does all this have to do with future races? What picked winners yesterday at sprints at Track Z may have not much effect for todays races. Then we have the stuff such as trainer angles or trainer/jockey combo that when this trainer did this or that combo got together it showed a profit in the past.

I agree that maybe such things as Biases exist on a daily basis like closers did exceptionally well or PP#2 was hot. But can we as handicappers rely on such short term stuff or even long term things when broken down between tracks, distances, class and surface???

Or is it better in trying to find ONE THING OVERALL that works well over a very large sample of races regardless of track, distance, surface or class? At least if that can be done you would think that this methodology would have a better chance at working in FUTURE RACES than what I mentioned above. What do others think?

[JustRalph]

you're spreading too large a net. Those things you mention in your first paragraph are invaluable individually speaking, for a smaller sample. Nobody plays every track..........well........almost nobody.

If you are a player who plays a certain circuit.........the tools in your first paragraph are the difference between winning and losing if you ask me.

A single piece of software trying to employ those methods over a large group of tracks is a different animal all together.

If you employ those methods on certain circuit, it would be entirely different than trying to do it over a couple of different racing circuits. That is where the value comes in.

[Pell Mell]

I usually download every track running and using my own standard methods I see what I come up with and here's what I've found, of course this only applies to my method.

All the small northern tracks like HST,GLD, PM, FE, etc. leave me totally baffled. On a very rare occasion I find a play at MNR and lately, since they seem to have a lot more horses I'm doing well at PEN.

I can't buy a winner at FE and have given up on that place yet I do great at WO, in fact I had 4 winners at double digit odds in two days there last week.
I do great at the LA and TX tracks and stink in SO. Cal. Hard to find but I do get some good plays at BM and EM.

Can't do a thing at GP or Calder but have success at Tampa. I do real well at the KY tracks except for ELLIS. NJ tracks suck big time and as far as I'm concerned so does NY.

My point is, a lot of these tracks are "sister" tracks yet the racing is entirely different for my money. It doesn't make a difference to me as to distance, surface, or whatever since I use form cycles and trainer angles to make my selections. I'm not sure but I think the reason I have problems with some of the smaller tracks is that the connections don't care that much about winning as long as they can get a check.:confused:

[alysheba88]

yes, yes and yes

[Dave Schwartz]

There are so many different possible "race configurations" out there that should not be ignored as to make each race seem very unique.

Most handicappers (as well as software) see the racing world from a single point of view. That POV may be pace or class or even what is perceived as a comprehensive approach. Nevertheless, we are always leaving something out.

Ask yourself this question:

"How should races be segmented for study?"
or
"What are the most important elements of the segmentation process?"

I believe that most players would agree that surface and approx. distance are probably the two most significant elements. That is, turf races should be handicapped differently than dirt races, and sprints differently than routes.

Thus far we have four permutations:

dirt sprints
dirt routes
turf sprints
turf routes

What about race type? Are maidens different than non-maidens? Of course they are. Okay, so now we have:

2 surfaces
x 2 distances (general)
x 2 race types (general)

and we are up to 8 different handicapping systems.

If you go on from here, you will find more elements to be considered. Is there a difference between a 5fur and a 7fur race? Sure. Well, then more precision is needed than just "sprint and route."

Age? Are 2yr olds different than older horses? Sure. High class races different than low? Yup. MSW different than MCL? Yes. Starter allowances different than Graded Stakes? Of course.

And what about differences from track-to-track (or circuit, if you prefer)? How about Poly track? What about off tracks, time of the year and day of the week?

My point is that there are so many different... configurations... that one cannot practically handle them all. Yet to not handle them all causes one to be... leaving something out.

So, in answer to your question, I would say, "Yes. You need to handicap tracks differently. And then some."


Regards,
Dave Schwartz

[InFront]

Ya but you gotta ask yourself why does my standard handicapping formula work at this track, not that track, this distance, not that distance. Sure if you break down ANY METHOD enough you'll find it works somewhere or at some distance or at some class but may be kidding yourself based on those PAST RACES. So you start playing where it works then find out many races later it didn't work there anymore but now starting to work at this track or distance.

This kind of "over backfitting" can make you flip and flop like crazy. I think that's one of the biggest weaknesses of databases. We can see things and a push of a button then slice and dice that data to force something to work somewhere. And since this is based on all PAST RACES WHERE THE RESULTS WERE ALREADY KNOWN doesn't mean it will continue to work there.

I did similar stuff in the past where I would find a method that works decently over a large sample of ALL races ALL tracks. When I say work I mean showed small losses. Then broke it down to find where the real profits were coming from by track and distance and class. And found huge PAST profits. But then forward tested it at just those profitable catagories on NEW RACES and it flopped terrrible. Why? Cause I chopped up my data so much I "forced" it to show me something that looked good on paper but in reality was "misleading". I think some serious database users on this forum understand my point here.

[Pell Mell]

In my case, since I'm going after what I think are "live" horses as far as the connections are concerned, what difference does it make to me whether it's a sprint, route, turf or dirt. I leave it up to the trainer as to whether or not the horse is placed properly. When one is trying to "Handicap" a race or race, isn't it just a process of trying to determine which is the better horse according to speed, pace, class, dist. etc.? If this is what you do then, IMHO, after almost 60 yrs at this, you are DOOMED!

[alysheba88]

Ya but you gotta ask yourself why does my standard handicapping formula work at this track, not that track, this distance, not that distance. Sure if you break down ANY METHOD enough you'll find it works somewhere or at some distance or at some class but may be kidding yourself based on those PAST RACES. So you start playing where it works then find out many races later it didn't work there anymore but now starting to work at this track or distance.

This kind of "over backfitting" can make you flip and flop like crazy. I think that's one of the biggest weaknesses of databases. We can see things and a push of a button then slice and dice that data to force something to work somewhere. And since this is based on all PAST RACES WHERE THE RESULTS WERE ALREADY KNOWN doesn't mean it will continue to work there.

I did similar stuff in the past where I would find a method that works decently over a large sample of ALL races ALL tracks. When I say work I mean showed small losses. Then broke it down to find where the real profits were coming from by track and distance and class. And found huge PAST profits. But then forward tested it at just those profitable catagories on NEW RACES and it flopped terrrible. Why? Cause I chopped up my data so much I "forced" it to show me something that looked good on paper but in reality was "misleading". I think some serious database users on this forum understand my point here.

For a variety of reasons I have given up on trying to be a master of all tracks/distances/race types.

I have consistently won at Aqueduct and certain types of races there, so I focus 75% of my play at Aqueduct. I will dabble at other tracks during the year for entertainment, but get serious when its Aqueduct. Again there are a few reasons for this

[ryesteve]

The problem with all these theories if you slice up results and data enough you will of course find "something" that worked.
You're 100% correct; for this reason, I'm also skeptical of the viability of chasing "what's winning now". At the same time, different tracks certainly do play differently. My approach is somewhat of a compromise between the two ends of the spectrum. It's a global approach where I ignore the tracks at which it consistently doesn't work.

[Capper Al]

yes, yes and yes

I agree. It seems to be all about setting up a winning model nowadays. Some software has it broken down very specific, for example, to 3yo Fillies at AP running 1 mile on dirt. It's tough on the small guy trying to beat the machines.

[Tom]

It still all boils to one horse goes out and the others try to catch him.

[ryesteve]

It still all boils to one horse goes out and the others try to catch him.Yup... but it still helps to know which are the tracks where they always do, vs. those where they often don't.

[JustRalph]

Great thread, I can see myself over the last five years in about half the posts..........