The Market Knows Plenty — But It Doesn't Know Everything🏇⤵️👇

There is a wonderful little phrase in betting that gets wheeled out whenever a well-backed favourite bolts up.

The market knew.

Apparently, the market always knows after the race.

Beforehand, of course, things are rather less straightforward.

That's why the Kalshi calibration charts caught my eye. They have nothing directly to do with horse racing, but they tell us something enormously important about betting markets, probabilities and the one thing every punter is really searching for.

The difference between the price and the chance.

What does 30% actually mean?

Strip away all the statistical language and calibration asks a beautifully simple question.

If you say something has a 30% chance of happening 1,000 times, does it actually happen roughly 300 times?

If it does, you're well calibrated.

If it happens 500 times, your 30% estimate was much too low.

If it happens only 100 times, you were much too optimistic.

The Kalshi charts compare the market's predicted probability with what subsequently happened across enormous numbers of markets.

The dotted diagonal represents perfection.

A 20% prediction happens 20% of the time.

A 50% prediction happens 50% of the time.

An 80% prediction happens 80% of the time.

And the fascinating thing is that as the event gets closer, the market gets better and better.

Three months beforehand there are noticeable wobbles.

One month out, things improve.

A week beforehand, the line tightens.

With one day remaining, the market is following that perfect-calibration line like a greyhound chasing the mechanical hare.

That's not magic.

It's information.

The market gets smarter as the clock ticks

Three months before an event, plenty remains unknown.

Things happen. Circumstances change. New information arrives.

By the final day, much of that uncertainty has disappeared and thousands of individual opinions have been distilled into one number:

the price.

Horse racing isn't identical, but anyone who has watched a betting market develop will recognise the process immediately.

Take a Saturday handicap.

When prices first appear, we might still be wondering about the weather, going, draw implications, jockey bookings, non-runners and how the market itself will interpret recent form.

Then the information starts arriving.

The rain comes.

Two horses are withdrawn.

The going changes.

A trainer's horses run well in the first two races.

One horse is smashed from 12/1 into 6/1.

Another drifts from 4/1 to 7/1 despite being tipped all over the place.

By five minutes before the off, the market knows considerably more than it did the previous evening.

Which is why beating the closing market consistently is bloody difficult.

But — and this is the important bit — difficult isn't the same as impossible.

Calibration isn't the same thing as efficiency

This distinction matters enormously.

Imagine every 4/1 horse in Britain collectively wins approximately 20% of the time.

Wonderful.

The market looks beautifully calibrated.

But that doesn't mean every 4/1 horse genuinely has a 20% chance.

Hidden inside that enormous sample could be horses whose true chances are:

28%.

And others whose true chances are:

12%.

Average everything together and you can still finish remarkably close to 20%.

That's where handicapping enters the conversation.

We're not trying to demonstrate that bookmakers are idiots.

They're not.

We're not trying to prove that betting markets don't understand probability.

They clearly do.

We're trying to find the relatively small number of occasions where this horse, in this race, under these conditions, at this particular price has been assessed incorrectly.

That's a very different challenge.

Stop picking horses. Start pricing them.

This is why I've increasingly come to believe that the traditional question —

"Who wins?"

— is the wrong question.

The better question is:

"What price should every horse be?"

Suppose I make a horse:

4/1 — 20%

The market offers:

8/1 — 11.1%

Now we have something worth investigating.

I'm not saying the horse will win.

In fact, according to my own assessment it loses four races out of five.

That's perfectly compatible with it being an excellent bet.

That's one of the hardest concepts in betting to truly accept.

A good bet can lose.

A terrible bet can win.

The result doesn't retrospectively change the quality of the wager.

Back my genuine 4/1 chance at 8/1 repeatedly and I'll happily endure plenty of losers.

Back genuine 8/1 chances at 4/1 and I might enjoy a lovely afternoon occasionally, but eventually somebody else will be buying the drinks with my money.

The market can also tell us we're wrong

There's another lesson in those Kalshi charts that punters shouldn't ignore.

If markets become more accurate as the event approaches, then price movement contains information.

Suppose I've made one 4/1.

Early bookmakers go 8/1.

Lovely.

Then it becomes:

8/1.

7/1.

6/1.

5/1.

9/2.

The market has gradually travelled towards my assessment.

That's interesting in itself.

But imagine the opposite happens.

I've made it 4/1 and the market opens 5/1.

Then:

6/1.

8/1.

10/1.

12/1.

There are two possible reactions.

The first is the classic punter reaction:

"Fantastic. Even more value!"

Sometimes that's correct.

Sometimes it's the financial equivalent of watching smoke pour from the kitchen and deciding dinner must be nearly ready.

A major market disagreement should make us reconsider our assumptions.

Not automatically abandon them.

Reconsider them.

Have we missed something?

Going?

Pace?

Fitness?

Stable form?

Draw?

A rival we've underestimated?

Or is the market genuinely wrong?

That's where judgement enters.

Your tissue isn't a prediction. It's an audit trail.

This is also why making a genuine 100% tissue is so useful.

It forces discipline.

Instead of saying:

"I quite fancy Horse A."

you have to say:

"Horse A has approximately a 17% chance of winning."

That translates to roughly 5/1.

Now we have something measurable.

If the market is 7/2, leave it alone.

If the market is 8/1, we're interested.

And, crucially, after hundreds of races we can test ourselves.

If we repeatedly assign horses a 20% chance, do approximately 20% actually win?

If our 10% horses win only 5% of the time, we've got a problem.

If our 30% horses win 45% of the time, we've got another problem.

Even if we've been making money.

Because a properly constructed tissue should ultimately behave like those calibration charts.

The percentages should mean something.

The uncomfortable truth about betting

The Kalshi graphs don't tell me that markets are unbeatable.

They tell me something much more useful.

Markets deserve respect.

They're enormous information-processing machines.

Thousands of opinions, models, biases, pieces of information and pounds eventually collide and produce a price.

Most of the time that price is pretty sensible.

Which means we're unlikely to beat it by simply finding the horse with the prettiest recent form figures.

We need something the market has underestimated.

Pace.

Trip.

Ground.

Sectionals.

A hidden run.

An inefficient ride.

A misleading finishing position.

Trainer intent.

A horse improving faster than the handicapper can react.

Or simply a probability we've assessed differently.

That's where the edge lives.

Usually not in screaming that the favourite can't win.

Usually not in discovering a 50/1 certainty.

But in the quieter discrepancies.

We make it 5/1.

They make it 8/1.

And then we find out whether our 5/1 really means 5/1.

Because ultimately that's the difference between having an opinion and having a betting method.

The market doesn't know everything.

But it knows plenty.

And the serious punter's job isn't to fight it on every race.

It's to wait patiently for those occasions when the market whispers:

8/1.

And your work says:

Not a chance. That's a 5/1 shot.

That's when things get interesting.