Horseracing Header Image

King George Winner Harbinger Rated 140 - The Reasons Why

Features RSS / David Johnson / 27 July 2010 / 18 Comments Free £25 Bet View Market

Harbinger stretches right away to record a record-margin 11 length win.

Harbinger put up a scintillating display to win the King George VI and Queen Elizabeth Stakes by 11 lengths at Ascot over the weekend to earn a Timeform rating of 140. One of the handicappers behind that rating, David Johnson explains the process of how that figure was reached.

"A 140 rating puts Harbinger in the top dozen performers that Timeform has rated, alongside Dancing Brave, Shergar, Vaguely Noble and Sea The Stars, and behind only Sea-Bird (145), Brigadier Gerard and Tudor Minstrel (144), Abernant, Ribot and Windy City(142) and Mill Reef (141)"

As much as many have tried, there is no denying that Harbinger's 11-length win in Saturday's King George VI and Queen Elizabeth Stakes (Sponsored by Betfair) marks him down as an outstanding colt. Rated 131+ prior to Saturday, his revised rating stands at 140, marking him out by far the best horse in the world currently in training, 6 lb ahead of Rip Van Winkle (134), whose rating comes from his 2009 form.

Timeform fully recognises the historical significance placed on their ratings within the industry, and a rating of 140 certainly wasn't arrived at lightly and without due diligence. We continue to hear of 'yardsticks' when people are discussing handicap ratings but, assigning a horse as having run its race and basing the entire assessment of the race on that premise, isn't the way we handicap races at Timeform. References like 'Youmzain is just about as solid a yardstick as can be found' is surely bogus, as even that horse's record in just the King George - beaten four lengths by Dylan Thomas in 2007, nine and a half lengths by Duke of Marmalade in 2008 and now fourteen and a quarter lengths by Harbinger - will testify.

Rather than try to guess which of Harbinger's rivals 'ran their race' or is 'the one to rate the race around', the first thing we do is produce a 'standard' for the race that objectively takes into account the difference at the weights of the runners and makes an adjustment for the likely strength of the form through the field size. The standard for this year's King George compared to the last 5 runnings of the race is 138/135/138/140/138 (most recent race first).

However, it could be argued that this year's King George was rather different in make up to those of recent seasons as it is the first year since Alamshar won in 2003 that there has been a meaningful representation from the classic crop. A standard compared to the years of 2004 to 2000 would read 138/140/140/139/143. Either way, whether looking at the race through a five-year or ten-year perspective, a rating of 140 for Harbinger isn't hard to justify.

Race standards are an excellent guide to rating horses with a limited amount of form to go on, but races like the King George are contested by horses we know quite a bit about, and we can use their previous form to calculate a different standard for the race (more on this can be read about in greater detail here). Using this methodology to rate the King George would suggest a rating of 142 for Harbinger.

The third key aspect we have for helping us to compile as accurate a rating as possible for any individual race is the overall race time. Form is probably a more key component than time in the ratings that are printed in our racecards and Black Books these days, but the time of a race is still very important in giving support to how solid a form rating is. You'd be far happier to expect a horse that finished fifth in a Lingfield maiden with a form rating of 72 and time figure of 70 to be able to reproduce that form rating than a horse with a similar form rating but a time figure of just 27. The timefigure for Harbinger returned at 'only' 135, is still one of the best over the last 20 years, and it is because the timefigure is unable to fully back up the form rating that we are happier to rate Harbinger 140 for the time being rather 142.

Using the overall race time in isolation though sometimes has its flaws and it is a pity that a flirtation with sectional times being produced and placed in the public domain has ultimately come to nothing in British racing. However, that doesn't mean that you can't produce your own and I am grateful to Simon Rowlands for sharing these. An analysis of the time recorded for the last 3f of each contender compared to the overall race time they recorded suggests that Harbinger was the only runner in the King George to run evenly.

Expressing the horses' finishing speeds as a percentage of their average overall speeds comes up with figures of: Harbinger 99.7%; Cape Blanco 95.5%; Youmzain 95.4%; Daryakana 95.4%; Workforce 93.7%; and Confront 91.4%, where a figure of around 100% would be optimum. This seems to confirm that the overall margin of victory was exaggerated somewhat by the beaten rivals running an uneven pace.

Poundage allowances for margins beaten need to be sensitive to the conditions under which they occur, and the pounds-per-length we used on Saturday - on good to firm ground and with the horses running at fast speeds - was greater than would usually be the case at a mile and a half. However, the evidence of the sectionals is that the beaten horses need to be rated a minimum of 5 lb better than the result. In practice, this is why Harbinger's master rating is 140, having recorded a performance figure of 140 in the race, Cape Blanco is 125/120 and the rest are further below form than simply allowed for by their closing sectionals, with Youmzain 123/113 and Daryakana 119/110.

In terms of a historical significance, a 140 rating puts Harbinger in the top dozen performers that Timeform has rated, alongside Dancing Brave, Shergar, Vaguely Noble and Sea The Stars, and behind only Sea-Bird(145), Brigadier Gerard and Tudor Minstrel(144), Abernant, Ribot and Windy City(142) and Mill Reef(141). Here's hoping he stays fit and well, as he has the form to carry all before him at middle distances for the rest of the year.

Tags: Handicappers' Corner, Harbinger, Timeform

Comments (18)

  1. James Knight | 27 July 2010

    Interesting article and a good explanation of how you have arrived at the number.

    I can't help but feel that "Race Standardisation" is always going to overrate wide margin winners.

    If you see an 11 length winner of a Group 1, (particularly at what I would consider an "odd" track like Ascot where form often seems unreliable), one's first reaction is that the other serious competitors have almost certainly run way below form. Allocating the winner's rating based on past standards for the race under these circumstances is surely fraught with danger?

    I agree with you that the time of Saturday's race adds weight to the argument for a high rating and have no doubt that Harbinger is a very good horse indeed. However, if the time says 135 - I would be sceptical of awarding any form rating in advance of that.

    If I were a betting man, I would say it is about 20-1 that Harbinger ever runs to a rating of 140 again in his career.

    Very much hope I am wrong, though!

  2. Wilson Lister | 27 July 2010

    “An analysis of the time recorded for the last 3f of each contender compared to the overall race time they recorded suggests that Harbinger was the only runner in the King George to run evenly”
    ”This seems to confirm that the overall margin of victory was exaggerated somewhat by the beaten rivals running an uneven pace”

    Interesting article but it seems to me like you are adding 2 and 2 together and coming up with 5.

    The beaten horses weren’t beaten because they ran their races unevenly – it was because the timefigure achieved by Harbinger was 135 and the form performance was around 140/2 and he was the only runner, hardly surprisingly, who could sustain the same effort for the last 3f (99.7%, your figures) that he did for the preceding 9f. You aren’t seriously suggesting that Youmzain, who was a length around Harbinger on the home turn, or the Derby winner Workforce, who was a length and a quarter/half ahead, were beaten because they ran their races at an uneven pace? Agr ed, the distances between the horses passing the line were exaggerated but that was because the horses behind Harbinger (with the exception of pacemaker Confront) were ridden to win the race and not to run their races optimumly. That they didn’t win, or come close to replicating Harbinger’s 99.7%, is only because, in probably all cases (Workforce might still prove us wrong) is that their abiliities didn’t allow it.

    Rather as some supposed ‘time experts ’ fail to include weight when making their calculations, prior performance ratings shouldn’t be ignored when taking into account when analysing these sorts of figures. That you have Cape Blanco running to a race rating of 120 while maintaining his master rating at 125 suggests to me that, in the context of the King George, he ran his race pretty much as effeciently as prior expectations for him had it; and if you didn’t believe that – that he nearly replicated his master rating while running his race ‘ineffeciently’ - surely you would have attached a + to his master rating.

  3. David Johnson | 28 July 2010

    Thanks for the responses guys.

    James - In practice, the weighting of race standards (as have been explained on Simon Rowlands blog recently) act in a way to prevent wide margin winners being given 'full' credit, certainly in the same way that the yardstick approach does. To use the Hawk Wing Lockinge as an example the standard for the race based on the previous 5 runnings (latest first) was 146/142/143/144/137. Had the race been 'rated around' Where Or When (124), the figure for Hawk Wing could have been as high as 149!

    I think its a dangerous approach to assume that everything else in the race has run way below form just because there was a general pre-conceived idea about Harbinger's merit.

    I think it is exactly races like this that it is a good idea to trust the methods that have stood the test of time through tens of thousands of races over the last few decades. Ofcourse though the rating obviously remains open to revision as future evidence of his merit becomes available.

    Wilson - I don't think we are too far apart from singing from the same hymnsheet. I agree it is trying to match the efficiency of Harbinger's efforts which led to the beaten horses finshing tired. Like you say, had they been ridden to run an optimal race rather than try and win it, they would have finished closer.

  4. James Knight | 28 July 2010

    Fair enough.

    I'm prepared to accept that Harbinger is an improver and that this was a clear career best - it just takes quite a leap of faith to think he has suddenly improved to this extent and is now pretty much the best horse of my lifetime. (He wasn't even considered the best horse in his own yard on Saturday morning)

    Watching the race again, I'd be very surprised if Cape Blanco, for example, has run within 2lb of his Dante form and 3lb of his Irish Derby run and that is what Timeform are claiming with their numbers.

    Anyway, as I'm sure you are aware, these arguments tend to go round in circles, so probably best if we agree to differ.

    Let's just hope the beast can stay sound and gets another couple of chances to show us exactly how good he is.


  5. Simon Rowlands | 29 July 2010

    It will be interesting to see whether the bookmakers - and the punters on here, whose views they tend to mimic - price Harbinger up like a horse who improved a couple of pounds at Ascot or 7 lb at Ascot the next time he runs.


  6. Wilson Lister | 29 July 2010

    Interesting reply - particularly the first bit.....

    Had some of the beaten horses been ridden 'optimumly' in a race run at that pace they would have been a fair way behind Harbinger on the home turn and the way they were ridden would then have been critisised by those that had backed them to win for not having been close enough on the home turn.

    Who really knows what the expected optimum effeciency level is in a race run at that extreme pace - after all it's not something that happens every day - perhaps it's 95% and it's Harbinger who ran the race ineffeciently...which is why he is 140ish and the others aren't

  7. Jose93 | 29 July 2010

    As Harbinger never started to "gap" the field until 2f out, what was his finishing speed from 2f out in relation to the other 10f?

  8. James Knight | 30 July 2010

    Bookmakers were noticeably keener than the punters on here to lay the last supposed track record breaking "superstar", Workforce, and I suspect it will be the same story when Harbinger next shows up....

  9. Simon Rowlands | 30 July 2010

    ...but I suspect they still won't price him up as if he is "only" a 135-rated horse.

  10. Charlie | 31 July 2010

    Good article

    Keep up the good work David, Simon.

  11. Simon Rowlands | 31 July 2010


    I made the finishing speed %s for final 2f as follows:

    HARBINGER 24.4 sec (100.3% fin speed)
    CAPE BLANCO 26.4 sec (93.8%)
    YOUMZAIN 26.4 sec (94.2%)
    DARYAKANA 26.5 sec (93.8%)
    WORKFORCE 27.3 sec (91.3%)
    CONFRONT 28.3 sec (88.7%)

    The more sectional information the better, but these would result in similar adjustments to the overall time/result in my book.

    There are a number of reasons why 3f out is preferred to 2f out at this trip, however, including that the camerawork makes it easier to take more accurate figures at that juncture.

    Thanks Charlie. I think it makes a welcome change to have a handicapper like David explaining his reasoning articulately and without recourse to platitudes.

    You don't have to agree with that reasoning, but I hope people will understand that it has been thought through.


  12. Simon Rowlands | 03 August 2010

    It seems I was wrong: the early bookmaker prices on Harbinger's next race - the Juddmonte International at York on August 17 - imply a rating of 135 if you reverse the ratings-to-odds process that I will describe this week and take a least-squares outcome.

    Then again, that is a 135-rated horse at 10f, a trip shorter than any he has run at in the last 15 months...


  13. Paul Smith | 03 August 2010

    Surely the bookmakers have priced the race up using less complex methodology than ratings-to-odds processes and least-squares outcomes, for all that they sound very sexy?

  14. Simon Rowlands | 03 August 2010

    Sexy, you say? Whatever turns you on.

    Personally, I am interested in establishing whether those who query things like ratings are being logically consistent in doing so.

    If that involves using a bit of simple maths and being open about it, rather than being glib or irrational, then so be it. Each to their own.


  15. Adam | 04 August 2010

    The rating that really needs explaining is the 132 from the Hardwicke.

    This makes Duncans run 125? equal to Sea the Stars' Derby win? Just shy of his Arc win?

    Then building on from this it would seem Harbinger has been elevated to an unrealistic level in the King George.

    I think a little bit of common sense has gone out the window here?

  16. James Knight | 04 August 2010

    I think it is perfectly reasonable to query a one-off rating - particularly such a high-one, and I haven't really seen any evidence of logical inconsistency or glibness from the doubters.

    Maybe he is the best horse of our lifetimes - or it could just be the case that a bunch of horses turned up at Ascot and ran disappointingly behind him...

    FWIW no-one seems to want to touch him with a bargepole for the International at 11-10, although that could be down to the possibility of him not showing up if the ground is really fast.

    If he does show and Rip and Byword line up against him, I know where my money will be going...

  17. Simon Rowlands | 04 August 2010

    I, too, think it is perfectly reasonable to query a one-off rating, though what "highness" has to do with it in this instance I do not know. It is arguably a LOW rating judged on established handicapping methodology. That point has been done to death.

    And I, too, have not seen evidence of logical inconsistency on this particular blog, for all that it has been in evidence elsewhere.

    An example of logical inconsistency might be someone talking down a rating and then putting up a price that is clearly far less negative in its assessment. That is not the case here.

    That there is a relationship between ratings and odds, and in reverse between odds and ratings, is self-evident and easily proved.

    That does not mean that a ratings compiler always forms an odds line from his ratings, or that an odds compiler uses ratings to come to his conclusions. Just that one is implied in the other.

    I find the suggestion that pointing this out, and pointing out how the rating implied by the odds could be verified in this instance, is "very sexy" to be glib. But perhaps it was intended as a compliment.

    It is the nature of horseracing that in much the same way that nothing was proved beyond doubt by Harbinger's win at Ascot, nothing will be proved beyond doubt by his win or his defeat at York, or wherever else he turns up. What a wonderful game!


  18. Simon Rowlands | 07 August 2010

    It has just been reported by Racing Post that Harbinger has sustained an injury on the gallops and may never run again.


Post a comment

Free £20 Bet + Up to £1,000 Cashback

Join Today
How to claim your £20 Free Bet + £1,000 Cashback offer
  1. Open your account (3 mins)
  2. Make a deposit into your account and place a bet on your selection (minimum £20)
  3. Should your selection lose we'll refund your bet + get cashback on your betting for your first 30 days up to £1,000
  4.   £20 Free Bet + £1,000 Cashback, Join Today

Writers' Tips

Follow our daily tipsters

  • Check back later for more tips

© Betfair 2007–12 | Contact Betting.Betfair team on:

Proud to back    

Betfair UK | Australia | Online sázení | Betfair Danmark | Wetten | στοιχήματα | Apuestas | Fogadas | Ireland | Scommesse | Norge | Онлайн ставки | Kladjenje | Vedonlyönti | Apostas | Zakłady | Vadhållning | 网上投注 | Betfair Corporate | Betting Education