Modern day angle players

[Ocala Mike]

Originally Posted by Capper Al

When you typically play, how many cards would you scan and about how long does it take you to scan them?

I typically limit myself to two tracks, and I can scan (say 20 races) for about 1/2 dozen of my angles in less than an hour.

[Capper Al]

Originally Posted by thaskalos

Since when are the horse's connections a "surprise"?

If a horse is going to turn around it most likely to come with the work of good connections. The horse up until it surprises should show very little.

[Capper Al]

I so like DeltaLover's term Scenario player, let's use it for what I called the modern angle player. Scenario player gets to the point that the handicapper is finding his angles via handicapping factors like class and speed.

[DeltaLover]

Originally Posted by Capper Al

I so like DeltaLover's term Scenario player, let's use it for what I called the modern angle player. Scenario player gets to the point that the handicapper is finding his angles via handicapping factors like class and speed.

The important thing is that a 'scenario' that does not significantly affect the chances of horse can still be very significant for betting purposes if the crowd is convinced for its fictitious importance... As an example, I can refer to the recent thread about the low vs high winning percent jockeys.

[DeltaLover]

What I find more interesting (and challenging from an implementation scope of view), is not to identify the individual 'angles' and compute the speed and pace figures (this is the easy and straight forward thing) but how to cluster the historical data based one them. In other words, how to build an algorithm to detect the degree of similarity between two races. given their profiles, as they are defined by the existing handicapping factors, the pace shape and the distribution of the various figures, I prefer to use.

Once this algorithm is perfected, it becomes easy to decide what kind of a horse profile has the potential to be an over or underlay. Following this approach, we know if we need to add or remove handicapping factors or to improve our figures, until we reach the desired levels of performance...

[traynor]

Originally Posted by thaskalos

IMO...it is wrong to criticise these "evaluation methods" as a group...as if they all closely resemble each other. Evaluation methods differ from each other as much as horseplayers do...and some are much more successful than others. I, myself, would need to see more proof before I am convinced that handicapping the connections is a better contender separator than "figure handicapping" and general handicapping knowledge are. The humans may prepare the horse, but the horse still does the running. Yes...in today's game we can't afford to ignore the trainer factor...but what you are advocating here seems to be ignoring the HORSE. And we can't afford to do that either.

There is a considerable difference between considering the horse as a factor, and considering the horse as the primary factor. I don't recall suggesting anyone ignore the horse--especially in the initial analysis stages of defining probable contenders in a given race.

[traynor]

Originally Posted by Capper Al

This is my current method. The thing is if one is hunting for a surprise factor for value then the connections are where it's at.

Indeed. The more bettors concentrate on the horses while ignoring the people, the better. It makes things easier.

[Capper Al]

Originally Posted by Ocala Mike

I typically limit myself to two tracks, and I can scan (say 20 races) for about 1/2 dozen of my angles in less than an hour.

That's pretty good. It's difficult to not get hypnotized while scanning.

[Capper Al]

Originally Posted by DeltaLover

What I find more interesting (and challenging from an implementation scope of view), is not to identify the individual 'angles' and compute the speed and pace figures (this is the easy and straight forward thing) but how to cluster the historical data based one them. In other words, how to build an algorithm to detect the degree of similarity between two races. given their profiles, as they are defined by the existing handicapping factors, the pace shape and the distribution of the various figures, I prefer to use.

Once this algorithm is perfected, it becomes easy to decide what kind of a horse profile has the potential to be an over or underlay. Following this approach, we know if we need to add or remove handicapping factors or to improve our figures, until we reach the desired levels of performance...

The late Danny Holmes book, Ten Steps to Winning, convinced me to handicap by race type treating a claimer different than an allowance different than a graded race.

[traynor]

Originally Posted by BlueChip@DRF

When no one else is paying attention.

When (almost) everyone else thinks "trainer studies" are simplistic number crunching and layering that have improved little since Greg Lawlor's and John Meyer's works way back.

I think it has a lot to do with fear of losing control. All those neat little rows and columns of numbers imply a degree of certainty that rarely exists in the analysis of human behavior. Those who fear losing control tend to shy away from those ambiguous, multi-layered, fuzzy areas in which the numbers are little more than convenient labels and indicators.

I also think that a more complex study of the human element is easier for those who wager primarily at one (or two or three or whatever) tracks, or those who make their more substantial wagers at a similarly limited group of tracks.

[Robert Goren]

Originally Posted by traynor

When (almost) everyone else thinks "trainer studies" are simplistic number crunching and layering that have improved little since Greg Lawlor's and John Meyer's works way back.

I think it has a lot to do with fear of losing control. All those neat little rows and columns of numbers imply a degree of certainty that rarely exists in the analysis of human behavior. Those who fear losing control tend to shy away from those ambiguous, multi-layered, fuzzy areas in which the numbers are little more than convenient labels and indicators.

I also think that a more complex study of the human element is easier for those who wager primarily at one (or two or three or whatever) tracks, or those who make their more substantial wagers at a similarly limited group of tracks.

Those who make substantial wagers only bet a few tracks because the tracks that have pools that can handle substantial is very limited. When you start to look at pool sizes and do the math on what it takes to move the odds, you get depressed very fast if you want to be a big spot play bettor.

[thaskalos]

Originally Posted by traynor

There is a considerable difference between considering the horse as a factor, and considering the horse as the primary factor. I don't recall suggesting anyone ignore the horse--especially in the initial analysis stages of defining probable contenders in a given race.

But you did suggest ignoring the horse during the critical contender-separation process...right? Or am I still misunderstanding you?

[traynor]

Originally Posted by thaskalos

But you did suggest ignoring the horse during the critical contender-separation process...right? Or am I still misunderstanding you?

Yes.

[traynor]

Originally Posted by Robert Goren

Those who make substantial wagers only bet a few tracks because the tracks that have pools that can handle substantial is very limited. When you start to look at pool sizes and do the math on what it takes to move the odds, you get depressed very fast if you want to be a big spot play bettor.

It depends on frequencies as much as pool sizes. High frequency (matches and wins) can do well with smaller wagers. Looking for that once a season "life changing win" is probably as self-desructive as betting every race.

The trick (if there is one) is finding a workable relationship between matches, wins, and ROI. You probably realize that most "ROI" figures are close to worthless--everyone saw the wins, figured what it took to catch them, and bet with both hands. I am WAY more interested in something with a decent match and win rate, and a (roughly) break even ROI.

Using "reverse models" that locate the most probable losers in the contention group (and eliminating them from wagers) can make it a lot easier to stay in the black. Including with patterns that "historically" show a low enough ROI to keep the rainbow chasers busy looking elsewhere.

[traynor]

Originally Posted by traynor

Speed, pace, class, and similar "evaluation methods" have a valid place in race analysis. That place is evaluating a group of entries to determine which are likely to be in or near the lead in the final yards.

Contender selection, eliminations, whatever else one wants to call it. I like to bet cavalry charge photo finishes.