Analytics for the impossible

What's the hardest thing to be found in horse race handicapping?

That's like asking somebody to count the stars on a clear night.

It's all hard.

Even acknowledging a sure thing is hard. Remember the one that hit the rail up ten lengths in the stretch? The one the rider fell off? The one that stepped in a hole? No matter how far you're ahead, you never quit watching. A big lead is two parts joy, one part fear.

As you bet and learn, you try to affix rules to the most difficult situations that reoccur. Most rules or guidelines follow this comment: Never again. Most rules have to do with the one thing that you can control at the horse races: your money.

Here's one such rule. Always bet a long shot to win, anything carrying odds of 10-1 or more.

Horse race handicappers are often right about certain long shots without collecting. That's like losing your winning scratch-off lottery ticket. I once had a 40-1 horse in an exacta box. It won and I lost.

Never again.

Here's the hardest situation to handicap: the impossible.

With most sports, it's the unlikely that breaks your heart. With horse racing, it can be the miraculous.

What do you do about the dog that that suddenly wins a race, or the decent horse that runs last?

You see lines like this all the time, with each number representing the horse's most recent finishes:

10, 9, 5, 7, 8, 1.

1, 2, 3, 2, 1, 12.

Has the surprise winner suddenly improved? Was the sudden lousy race the result of injury? Or was it just a bad day?

The impossible is routine at the horse races; it can be so frustrating, a handicapper is apt to try anything, even analytics.

Analytics are available for every inch and nuance of a horse race. Analytics are numbers, percentages, charts and graphs, trends. They're intended to replace thought. Plug in some stats, pull out a winner. Whereas the field of sports analytics reeks of higher intelligence, in a way it's like having a brain that never forgets. Facts without color is a boring history class. Pure memory is not intelligence. Reasoning and interpretation are what make memory come alive.

I tried to come up with some artificial intelligence to help create a rule about what to do with a horse after an impossible result.

I watched dozens of races that included winners that couldn't have been bet to win before the race, and losers that looked like cinches to emerge victorious. Here's what I learned. Most upsets are spawned in the starting gate. Break bad, run bad, that's usually the way it goes at the cheaper levels. Sometimes breaking bad is an illusion, as is the ability to close late, which is actually the act of running a little faster than all those slowing down. In one impossible race, a 4-5 horse was simply out-broken by four others with good gate past performances, didn't like being way back there, and loped around and ran last to a 20-1 horse. Big upsets are often first out of the gate. Horses that ran poorly because of an injury show a break in routine.

Horses that perform unexpectedly usually migrate toward the middle of the pack next time out.

When you watch races in search of a new analytic, without betting, bad rides jump off the screen like circus clowns. So the only inexplicable performance worth a bet is the horse that ran terribly for no reasons except that it's an animal, and there was a joker on its back.

Victories after miracle wins are rare as an opera concert after the races.

The only conclusion that can be drawn from an analysis of the impossible is that the same freakish event seldom happens twice running. So the best way to deal with the impossible past performance is by employing what may seem like the impossible in horse race handicapping, passing on the race and having a beer.