In-Play / Simon Rowlands / 06 February 2012 / 5

Rare, or "Black Swan", events make the headlines when they occur. Simon Rowlands considers the evidence of how and when they have cropped up for in-running punters...

One of the upsides of a spate of racing cancellations - such as we are having in Britain at present - is that it affords time to take stock and to refine betting strategies without much in the way of day-to-day distractions. Or that is what I am trying to convince myself at present.

As an example, I decided this might be a good opportunity to look at some data concerning in-play prices and the incidence of rare, or "Black Swan", events. I really should get out more (or "at all" would be nice given the snowdrift at my front door).

There is, understandably, a bit of a kerfuffle when the thoroughly unexpected happens.

Winners matched at [1000.0] in-running and losers matched at [1.01] in-running make headlines. Someone usually asks "how can you lay/back a horse at such a price without complete certainty?", while conveniently forgetting that the layer/backer picks up many more times than they leave their money behind.

From 2009 to 2011 inclusive (Flat and jumps), there were nearly 140,000 horses matched at [1000.0] in-play and just 108 of them won. At the same time, nearly 28,000 horses were matched at [1.01] in-play and 387 of them were turned over.

What should be stated up front is that rare events like [1000.0] winners happen too infrequently for it to be possible to come up with meaningful figures in many instances.

If something is expected to happen only once in 1000 attempts then you are going to need very many more than 1000 attempts to start to see a true pattern. Random variation will see to it that parts of the data stand out from the norm.

That said, betting is an exercise in acting upon information that is often incomplete. You are likely to miss opportunities (good and bad) if you always wait for copper-bottomed proof.

With all the above in mind, I consolidated the in-running data by course (treating Flat and jumps separately) and combined [1000.0] winners and [1.01] losers in order to see how often they occurred compared to how often they might be expected to have occurred. Here are a few of the findings:

The top jumps courses for Black Swan events were:
FOLKESTONE (rare events occurred 2.31 times expectation)
WORCESTER (2.29)
NEWCASTLE (2.14)

The top Flat courses were:
CATTERICK (1.60)
SANDOWN (1.58)
HAMILTON (1.42)

The least erratic jumps courses were:
ASCOT (0.00)
NEWTON ABBOT (0.16)
WINCANTON (0.19)

The least erratic Flat courses were:
PONTEFRACT (0.54)
RIPON (0.63)
CHESTER (0.68)

If the identities of some of those courses come as a surprise, it should be remembered that what is being measured is the in-play market's ability to get things "right" where rare events are concerned, and not the predictability of results and events at the course itself.

We may expect races to change in character at somewhere like Pontefract, but that already could be factored into the in-play prices themselves.

Ten Flat courses did not have a single [1000.0] winner in the period under review, with Goodwood (2109 horses matched) having the biggest sample.

Catterick (Flat) had 9 horses turned over at the minimum price of [1.01] from 358 horses matched at that price.

Chepstow had 5 horses beaten at the minimum price in chases from just 103 matched.
Meanwhile, Worcester had 5 horses win at [1000.0] in hurdles and bumpers from just 1,434 matched.

Ascot did not have a single [1000.0] winner or [1.01] loser over jumps from 135 races.

In order to drill down a bit more, and to lessen the influence of variation, I considered losers at less than [1.10] in-play at the four all-weather tracks, the courses at which there was, by some way, the most racing.

The fact is that there was little difference between them overall (Southwell led the way in in-running unpredictability, though not by much). There was, however, a tendency for fewer rare(ish) events - of losers as a proportion of horses matched at short odds - to occur at shorter distances than longer ones.

As mentioned, the above figures are prone to fluctuation. It will be interesting to see whether they are merely descriptive or will be borne out by events hereafter.

'.$sign_up['title'].'

'; } } ?>

(5)

1. Progressive27 | 07 February 2012

   Interesting piece,

   Can I just ask how punters can do their own research regarding 'in-running' statistics - is there a database somewhere? I've seen the horse racing results section on Betfair in which info can be collated manually by backtracking, but is there a research tool that speeds things up, as in your research Simon?

   Thanks.

2. Andy C | 07 February 2012

   Nice article Simon, good food for thought on a quiet day. Ascot's 0.00 might be partly explained by incredibly high IR liquidity. It's so thick there that high side 'spikes' are most unlikely to occur there as a consequence of 'thin' markets which you find created at most other tracks, hence meaning the only 1000 you get is about a faller.

   Similarly 1.01's don't happen as easily there as elsewhere as there is so much liquidity between 1.05 and 1.20 to churn through first that it's only when a race is genuinely over that the 1.01 money is then enveloped.

   I've tried to theorise the reasons for this occurring seemingly at Ascot and not elsewhere and there is only one logical explanation.

   It's status as the best IR track in the UK is due to viewing mediums. It's often on Terrestial & always on ATR, SIS, TurfTV and LV, there is no other track that is viewable in this way.

   Essentially it's in reach to basically everyone in some way, creating a liquidity bonanza with no regular equal.

   Goes to show what 'could be' for IR in an ideal world.

3. murtibing | 07 February 2012

   Hi
   Yes you can use
   www.fracsoft.com
   the adrian massey site was still up too last time I looked.
   Don't forget to tell me what you discover haha!
   Nick

4. Simon Rowlands | 07 February 2012

   Many thanks, Progressive27. I think the third comment answers your query.

   I have heard of fracsoft but cannot personally vouch for it.

   My own research came as a result of a specific data request from Timeform to Betfair. I can, however, say that Timeform hopes to take this sort of thing into account much more in future.

   Simon Rowlands

5. Simon Rowlands | 07 February 2012

   Many thanks, Andy C!

   I very much like your hypothesis regarding Ascot. It makes plenty of sense and underlines the kind of real-life factors that need to be taken into consideration when betting in-running at a course.

   People should not imagine that Ascot is somehow a no-go for rare events. There will be a 1000 winner there, and a 1.01 loser, some day. But it just might take a very long time for the reasons you suggest.

   Simkon

:

:

:

?

: