The Rating — A Common Measure

Start with ability; leave room for today's horse to be different.

Theory: what the number measures

A horse can finish second in a strong race and produce a better performance than a winner elsewhere. Finishing position tells you who beat whom. A performance rating tries to express the merit of the run on a scale that can be compared with other runs.

MWP uses a common scale across the racing it covers. Higher is better. A performance assessed at 120 is intended to represent the same level wherever it occurred. That is a statement about the measurement, not a promise that the horse can run 120 on any surface or at any distance.

Three numbers are easy to confuse. The historical rating describes a past performance. The Expected Rating is a model's estimate of what the horse may produce today. An official handicap mark belongs to the local racing authority and helps determine eligibility or weights under its rules. Similar-looking numbers need not share a scale or a purpose.

The rating is most useful when you ask a specific question: has this horse shown a level that makes it competitive against today's opponents?

Why the clock needs context

Raw times change with distance, going, wind, track configuration and how the race was run. A slowly run race may produce an ordinary final time even when its leading horses are very good. A fast surface can produce striking times without a field of exceptional horses.

Time-based figures attempt to correct for these effects. They are not useless, and a projection method is not infallible. The practical question is which measurements remain reliable when the clock and the form disagree.

MWP's approach uses the prior records of established runners to help calibrate the race. The horses act as reference points, and the margins between them help place the performances on a common scale. Weight and distance are handled within the relevant calculations rather than ignored.

Illustrative example: Horse A has usually run around 110; Horse B around 113. B beats A by a margin worth roughly three rating points under the example's conditions. A reading near 113 for B and 110 for A fits their records. A newcomer finishing two points behind A would land near 108 on that interpretation.

That is an explanation of the principle, not a rule that established horses must always reproduce their old figures. If A was unwell or B improved, the reference changes. A larger set of reliable comparisons helps; an inexperienced field leaves more uncertainty. Some races deserve review rather than a confident number.

Know what has already been adjusted

A performance figure is not a blank starting point to which you add every possible advantage. Read the definition of the field you are using.

MWP's core performance figures do not automatically turn every difficult trip into a fully corrected measure of ability. Ground loss and trouble may still need interpretation. But some displayed projections or user adjustments can already incorporate a factor. Today's carried weight also belongs in today's comparison.

If a calculation has already accounted for a weight difference, do not add it again by hand. If you have already raised a horse's expected performance for a wide trip, the same trip is not another independent reason to raise its probability later.

Keep a short adjustment record: original estimate, changed estimate, reason. It is much easier to spot double counting when the steps are visible.

Read a range, not a greatest hit

Consider two illustrative horses:

B has the higher peak. A has more evidence of repeatedly running around 111. Whether B is more likely to beat A today depends on what produced 121 and whether those circumstances are relevant again.

Do not discard the peak because it is inconvenient. Do not promote it to the default because it makes the horse look exciting. Ask whether the run was credible, whether the horse had unusual help, and whether the subsequent performances have informative explanations.

Representative ability is a judgement about a plausible performance range. A horse with a narrow range and a horse with a wide range can have similar average expectations but different chances in a particular field.

New tops, improvement and the bounce story

A new top is a performance above the horse's previous best. It may reflect development, a better-suited task, improved fitness or favourable circumstances. It may also contain measurement error.

A lower figure next time is not automatically proof that the top exhausted the horse. Extreme observations often move back towards a more typical level even without a physical cause. This is regression towards the mean.

A physical recovery question is still legitimate, especially after a demanding effort. Examine the spacing of runs, prior responses, age and the conditions. The error is turning a sensible question into a universal instruction to oppose every new top.

An improving horse can repeat or exceed a peak. A mature horse can run below a recent peak and still win. The bet depends on today's distribution of outcomes and the offered price.

Two measured details sharpen the question. In our data a modest new top, three to seven points above the old level, was the strongest next-race group, while a jump of eight points or more was the one that usually fell back – so size the improvement before deciding which reading applies. And a collapse, a figure far below the horse's level, tended to persist: a third to a half of such horses ran poorly again next time, whether they returned in a fortnight or after months off. Rest did not rehabilitate a collapse.

Using the scale without memorising a class table

Begin by comparing horses within the race. How large is the gap between their credible performances? Are the strongest figures recent and relevant? Does an apparently weak runner have little exposed form rather than established low ability?

Class averages can provide orientation, but they depend on the period, population and definition of the rating. They are not admission requirements. A Class 3 label in one country is not a conversion formula for a Class 3 label somewhere else.

For orientation, here is what the scale looks like in three of the populations this course uses. The figures are the average MWP rating of runners and of winners by race class, measured on our data in June 2026 for races from 2021 onward. They describe populations, not entry requirements: classes overlap, and the winner typically sits about ten points above the average runner in the same class, because the winner is usually the best horse having its best day.

Hong Kong

Britain, all surfaces

Australia

Note the numbering. In Britain and Hong Kong, Class 1 is the top of the handicap tree and the figures fall as the class number rises. Australia numbers its classes the other way round: Class 1 is the bottom, horses that have won a single race, and the ratings climb with the number. Group and Listed races sit above the lot in every jurisdiction.

The most useful habit is to read several runs, inspect the context, and write down a plausible expectation before moving to the next lens.

Practical: read the pattern

Quiz: measurement and expectation

Optional practice: challenge the easy pattern label

Up next: build a complete race opinion without counting the same evidence twice.