Pricing — What Chance, at What Odds?

Finish the assessment before choosing the bet.

Theory: turn the questions into a distribution

Ability, readiness and conditions tell us what might happen. Pricing asks how often the different runners win across those possible races.

A rating gap does not have a universal conversion into probability. The number of opponents, uncertainty around their performances and the dependence created by a shared pace or surface all matter.

If a documented probability model supplies a baseline, use it as a starting point. If you have only ratings, a manual line is your own estimate and should be treated as a skill under development. Do not manufacture precision by converting a rank directly into a percentage.

One modelling approach is to describe distributions of possible performances, simulate races and count wins. Its answers still depend on how those distributions and shared conditions are modelled. A simulation cannot repair assumptions it was told to believe.

Work through a whole field

Here is an invented five-runner example. The probabilities are teaching assumptions, not an empirical conversion from rating points.

A begins as the strongest established runner. B is dependable; C has upside and limited experience. D and E have less demonstrated ability but could benefit if A does not reproduce its level.

You begin with 35%, 25%, 20%, 12% and 8%. On review, A's readiness is less secure than the initial assessment allowed, and today's likely setup gives D and E somewhat better routes. You settle provisionally on:

The particular revision is not uniquely correct. What matters is its stated reason, its effect on the whole field and whether you can later test the process.

Do not leave other runners unaccounted for. If you study only the main contenders in detail, reserve a defensible combined chance for the remainder.

A safety margin does not mean 110%

A probability line for mutually exclusive winners should total 100%, subject to the settlement assumptions of the exercise. Raising it to 110% increases the probabilities and lowers the prices you would accept. That makes you less cautious.

Handle uncertainty through the required price, the range of plausible probabilities and the amount of risk you take. If E's chance could reasonably be between 7% and 12%, the provisional 10% should not feel like a laboratory measurement.

The fair odds for 10% are 10.00; for 7%, about 14.29. An offer of 11.00 looks positive under the central estimate and negative under the lower one. You need a reason to trust the central estimate before calling the difference an edge.

Compare with an actual offer

For a simple one-unit win bet with no additional charges:

Expected net return = probability × decimal odds − 1.

D and E are candidates under these assumptions, not compulsory bets. Estimation error and costs can consume those apparent advantages.

There is no universal minimum overlay that makes an estimate safe. A small apparent edge built on uncertain assumptions may be unusable; a larger one may still reflect a bad model. Price discipline includes declining to act.

Fixed odds, exchanges and tote

With fixed odds, the accepted price is normally locked subject to the operator's settlement and deduction rules. Check terms and any later adjustments.

On an exchange, a quoted price is useful only if the required stake can actually be matched. Commission and market-level settlement affect the return. For an isolated bet with commission charged on its win profit, the effective winning return is 1 + (decimal odds − 1) × (1 − commission rate). A portfolio may need a different net-market calculation.

In a pari-mutuel or tote pool, stakes are pooled, the applicable deduction is removed and the remainder is distributed to winning tickets. The displayed odds are provisional. The price you see before submitting is not a guaranteed final return.

Use current liquidity and expected late movement when assessing a tote bet, while acknowledging that you cannot know the final pool in advance.

The margin is already a hurdle

Bookmaker overround and tote takeout are related costs but not identical quantities. A 20% takeout leaves 80% of the pool for returns; overcoming that proportional reduction requires a 25% uplift relative to the corresponding no-takeout return, since 1 / 0.80 = 1.25.

Do not then deduct the takeout a second time from an EV calculation using the actual quoted payout. Check what the odds already include. Separately account for commission or charges that are not included.

HKJC's published World Pool table lists 17.5% for Win and Place, 19.5% for Forecast/Exacta, 23% for Trio and 25% for Tierce/Trifecta. Offered pools and access vary by operator and jurisdiction; check the applicable rules. Larger liquidity does not automatically mean a lower fee or a profitable market.

Mr. World Pool, the form product used in this course, is distinct from HKJC's World Pool commingling service.

Use the market without surrendering the assessment

Writing an initial line before viewing prices helps reveal anchoring. Once the market appears, treat it as information too. A large disagreement may reveal an opportunity, but it may also reveal something you missed.

If you revise after seeing prices, save both versions and the reason. Do not retrospectively claim the market-informed line was your independent forecast.

Closing line value

For a locked-price win bet, comparing the accepted price with a comparable closing price can help assess execution. Taking 6.00 when the same selection closes at 5.50 is a simple odds-ratio improvement of about 9.1%.

That is a price comparison, not realised profit or proof of a correct probability. Closing prices can contain noise or different margins. Account for costs and compare like with like.

In tote, the earlier displayed odds were not locked. Record the indicated price at decision time and the final dividend, but do not present their difference as the same captured CLV available on a fixed-price bet. Expect the board to move late: in the big races we examined, 40–50% of all pool money arrived in the final minute before the off.

A long enough record of good execution can be encouraging. No fixed count of bets guarantees that positive CLV will turn into profit.

Practical: price before the reveal

Optional cases: investigate the apparent overlay

Quiz: price discipline

Up next: a good view of a horse and a good ticket are different decisions.