Going at Ayr has changed to Good (GS places) and Fitzgerald reckoned this morning that it would be Good by race time. Silver By Nature and Rose Of The Moon both non-runners . Certainly more like the ground that Aurora's Encore will relish and he could well run a big one this afternoon despite it being only 14 days since Aintree. Like the GN, counter-intuitively the stats suggest that being “fresh” is not an advantage in the Scot Nat. AE’s close 2nd last year was an identical 14 days after his last run and win (though well short of 4.5m) and Hello Bud (2009) and Joes Edge (2005) both won the SN having won their previous races just 9 and 7 days before respectively. So, expect a big run from the National hero……….but no “Rummy”. On good ground, the topweight did win the SN twice in the 1990s and was runner-up to Hello Bud by 0.5L in 2009 but in the last 9 renewals on (officially) Good ground (excl GF), horses carrying <10.12 account for all winners and 7/10 of those placed <4L of them. All 4 of mine below 10.12. The Off is at 3.50 (ch 4).
Looks like Cool Operator was 6th and Rebeccas Choice 7th (approx 15L behind the winner) look like the only other finishers - pretty brutal pace. Monsieur Cadou never going and PUd with a lap to go, as did Captain Americo, Rig De Beauchene and then a host of others over the last mile and a half. Sorry guys.
Another novice winner (5 of last 14 SN winners). I think that's the basic problem (common to both Irish & Scottish Nats) - that relatively inexperienced horses can win them. Even in the Welsh, a horse older than 9 hasn't won for 20 years. Quite simply, younger horses (novices in particular) typically have relatively fewer stats to work with (i.e. they're "unexposed" from a stats-perspective) compared to older horses. That's why the GN model isn't replicable for other races and any trend-analysis has to be low conviction. That's my excuse and I'm sticking with it!
Another novice winner (5 of last 14 SN winners). I think that's the basic problem (common to both Irish & Scottish Nats) - that relatively inexperienced horses can win them. Even in the Welsh, a horse older than 9 hasn't won for 20 years. Quite simply, younger horses (novices in particular) typically have relatively fewer stats to work with (i.e. they're "unexposed" from a stats-perspective) compared to older horses. That's why the GN model isn't replicable for other races and any trend-analysis has to be low conviction. That's my excuse and I'm sticking with it!
Now I know the meaning of your moniker Peanuts. ;-)
Folks - many apologies indeed for the rank selections at Ayr…….they were utter crap but, by virtue of my post-GN stat-analysis and model-tweaking, I shall endeavour to make recompense next April.
In case anyone is interested in how I intend to do it, this is the 2nd of 3 Parts to my post-GN stat-analysis. It completes the explanation of the tweaks to the model, mainly in respect of the poor performance of runners with 11.00+. The 3rd Part (to come shortly) will be a statistical analysis of the actual (contrasting with the apparent) effects over the years of the handicapper’s compression of the GN weights.
GN ANALYSIS - PART 2: Final Tweaks To The Model
My initial reaction to the 2013 race was that it was a “pre-compression”-type result, with only 1 of the first 11 home carrying 11.00+ (15 runners carried 11.00+ of which 4 finished: half the completion rate of those carrying <11.00). However, as explained in Part 3 to come, there are some erroneous assumptions about the pre-compression era and the compression of the handicap may not have had the effects that are commonly supposed. My conclusion now is that the 2013 GN result is not some aberration from the “new normal” but confirmation that it is just as hard as ever (in certain cases it may even be harder!) to carry 11.00+ to victory in the GN.
Before the race I had suggested that the withdrawals of Tidal Bay (dropped 9lbs from his current mark) and Alberta’s Run (-8lbs) and consequent 4lb weight-rise for the field had half-neutered the compression and, from the model’s perspective, could materially affect the result. To recap, this was because:
1. All 3 remaining runners favoured by the handicapper (Imperial Commander [-4lbs], What A Friend [-4lbs] and Weird Al [-8lbs]) had had interrupted campaigns and/or missed preps or simply poor seasons (no amount of favourable relative weight treatment would be enough to compensate); and
2. The 4lb rise in weights changed the task for several former GN heroes from tough-assignment to mission-impossible (especially Seabass, Sunnyhillboy, Ballabriggs and Big Fella Thanks).
These horses together accounted for 7 of the top 9 on the card and all carried 11.04+. All of them under-performed even the model’s moderate (at best) rating of their prospects (some significantly).
It led me to trial a tweak to the model, which increased (by a stepped scale according to weight to be carried, starting at 11.00+ with a step-up at 11.06+) the standard penalties applied across a range of criteria, including age, runs in 3m chases, form over 3.5m+, days since last run, quality of prep, season’s form, “spring ground” form and, if applicable, weight increase on prior GN. Not only did this tweak reconcile the poor runs this year of most horses with 11.00+ but, in back-testing the tweak against all GNs since 1988, I was amazed at how robust this change to the algorithm is, even when applied to the recent “compression GNs” when horses with 11.00+ won and/or filled 3 of the first 5 places (all GNs 2008~12) and even when there was no weight-rise after the handicapper had framed them (2010~12).
So this tweak is now confirmed and its logic is simple: despite compression, it is still so demanding to carry 11.00+ in the GN that there must be virtually NO chink in the stats-armour of such runners if they are to have a chance of being in the shake-up, and none if they are to win.
The tweak also corrected the model’s underestimation of the the effect of the weight-rise on those that were brought into the 11.00~11.03 range. That was (IMO) the critical factor in T43 not quite getting home (the tweak reconciles his virtual dead-heat with Cappa Bleu) and Join Together (an 8 year-old with poor spring-ground form) running OK but well below expectation.
Finally, together with other small adjustments, the “11.00+ tweak” reconciles the starkly different runs of two horses with very similar profiles but different weight burdens: Oscar Time (4th 20L with 10.11) and Ballabriggs (PU with 11.04). Both 12 year-olds, respectively the close runner-up and winner of the 2011 GN but with no win since and a 10L near-miss over 3m the best of their season’s runs, the difference in absolute weight burden would seem to have been highly significant; in addition to Oscar Time clearly relishing a return to Aintree and decent ground (he seems to be one of those horses, like Mon Mome, that had excellent form on heavy in his early days but, since injury, simply doesn’t want to run on the testing ground he faced all season).
So that is the model tweaked for another year. It’s now in shape to run the slide-rule over 80 or so entries and their weights next February and, hopefully, see us right.
Part 3 to come (stat-analysis of the effects handicap compression).
I only read this morning Phil Smith's blog of last week, bleating about Harvey Smith's understandable beef that Aurora's Encore had been raised 11lbs after his GN victory, to a mark 5lbs higher than his close 2nd in the Scot Nat a year ago. Harvey Smith made the point that Smith is unduly harsh on GN winners, evidenced by the fact that the last GN winner to win any other race was Bindaree ( 2002 GN). I happen to agree with Harvey.
In his reply (below) Phil Smith makes the point that he was right to raise AE 11lbs since this equated to the 9L margin over Cappa Bleu (due to go up 2lbs for his good run at Ascot). If he had agreed with Harvey, CB would only have had a 3lb pull for a 9L defeat. But that argument only makes sense if you equate 1lb to 1 length, which is palpable nonsense over 4.5m.
It is commonly accepted that over 5f 1lb = 0.33L and only over 2m does 1lb = 1L. Over greater distance, logically every 1lb of weight should have a larger effect upon performance (probably exponentially so). Willie Mullins has estimated it at 1lb = 2~3L for the GN. If one takes Mullins 2-3L guestimate, that suggests that Harvey Smith is spot on and AE should have had a 5lb rise.
My own analysis supports Mullins’ 2~3L estimate but only at the 10.00~10.07 weight range (which is the relevant range here because AE carried 10.03). In my view, the equation also depends significantly upon (and grows exponentially with) the absolute weight that a horse is carrying over a marathon trip such as 4.5m. To take an extreme example, if you loaded 15 stone on any horse and asked it to run at 15~15.5 secs per furlong they probably would not be able to go as far as 3m let alone 4.5m. Therefore, at the margin, every 1lb added at and above 11.00 will be more of a burden that at 10.00). Based upon my stats, in the GN, roughly, 1lb of extra weight is at least 4-5L at 11.00.
I shall be returning shortly to the matter of Phil Smith in my analysis of the effects of compression of the GN weights but I do have doubts as to whether this bloke really is up to the job…….apart from which his arrogance (later in his blog, he tells Sue and Harvey Smith they should run AE in the Bet365 at Sandown, not the Scottish National) is totally unbecoming of a BHA Head of Handicapping…………IMHO!
QUOTE
National winners and their runs in handicaps
BHA head of handicapping's blog - by Phil Smith
LAST week I explained how I came to the decision of raising Auroras Encore from 137 to 148 after his win in the John Smith's Grand National. It seemed a fairly logical decision to me, but it seems it has upset Harvey Smith who thought he should have been raised by only 5lb not 11lb.
As I explained Cappa Bleu was due to go up by 2lb before the race so that if I had indeed only raised Auroras Encore by the 5lb suggested by Harvey then Cappa Bleu would only have had a 3lb pull for a nine-length defeat. I think most impartial observers would feel that in that case Auroras Encore would have had a better than equal chance if the first two in the Grand National were to meet next time…..
COMPRESSION OF THE GRAND NATIONAL WEIGHTS:…………….separating fact from fiction.
When betting in the GN, particularly if one has a stats-based approach, it’s important to try to determine what effect the handicapper’s compression of the GN handicap (at his discretion) tends to have on the prospects for certain types of runner, if not the whole field.
Compression started in 2002 and such is its apparent success that it is now the received wisdom that to carry 11.00+ is no longer the impossible burden that it once was. No GN winner carried 11.00+ from 1989 to 2004 inclusive but all 4 winners from 2009 to 2012 carried 11.00+. So, it’s obvious that compression has succeeded in its mission of making the GN easier for horses carrying 11.00+ and/or assisting the best-rated horses in the field to go well……..…………………...isn’t it?
Well, the 4 wins are in the record book but, if you think that it is easier to carry 11.00+ now and/or that the horses favoured by the handicapper are reaping the benefits, you might be surprised by the following stats, which compare 11 pre-compression GNs (from the last 11.00+ winner in 1988 to Bobbyjo’s win from a full stone out-of-the-handicap in 1999) and the 12 “compression GNs” (from 2002 to 2013):
• Proportionate to their representation, in the pre-compression period horses with 11.00+ were more than x2 as successful at filling one of the 1st~5th places in the GN as those with 11.00+ in “compression GNs”.
• Pre-compression, horses with 11.00+ had a 58% completion rate (41% of horses with <11.00 completed). In the “compression GNs”, horses with 11.00+ have a 32% completion rate (38% for horses with <11.00); i.e. while completion rates for <11.00 are broadly unchanged, there has been a significant fall in the rate of completion by those with 11.00+ in the compression GNs.
Yes, but that’s surely because compression has encouraged more entries from higher up the handicap and, by definition, dragged more horses into the 11.00+ camp, many of them with unsuitable credentials for the GN. That is all true but consider these stats:
• Horses from the top 5 on the race-card (i.e. those horses likely to be most favoured by the compression) filled 31% of the 1st~5th finishing places in the 11 pre-compression GNs. In the 12 “compression GNs” they have filled only 15% of the 1st~5th places.
• Spreading the net wider, horses from the top 10 on the race-card filled 42% of the 1st~5th finishing places in the pre-compression GNs, compared to 20% in the “compression GNs”
The figures and ratios change only slightly if one adjusts for the specific (often lower) field-size of the pre-compression GNs (average 36.5 vs 39.9): the only change being that the top 10 pre-compression rate falls to 40% from 42%.
So, not only are runners with 11.00+ now faring worse collectively with weight-compression but runners that are likely to be most favoured by the compression (defined crudely either as the top 5 or top 10 on the race-card) are only half as successful at making the frame as were their counterparts pre-compression.
A similar ratio of outperformance applies for wins aswell:
• Pre-compression, the top 5 on the race-card contributed 2 wins in 11 GNs. This compares with 1 win in 12 “compression GNs” (and that solitary win was by a Nose last year)
• The top 10 on the race-card contributed 7 wins in 11 pre-compression GNs (6 when adjusted for low field-size), compared to 3 wins in 12 “compression GNs”.
• Moreover, only 1 of those 3 compression wins (Neptune Collonges’ Nose) was by a horse whose weight was dropped by the handicapper from its prevailing mark (2lbs in his case), though it is just as relevant to note that his GN prep (after the GN weights were framed) had been so impressive that his real mark was due to rise by 5lbs; so he was in fact 7lbs well-in in the GN. Both of the other “top 10” wins were by horses (Don’t Push It in 2010 and Ballabriggs in 2011) not specifically advantaged by the handicapper, though there were runners in both races whose weights had been dropped by 4lbs+ from their current marks (none of them completed).
As they say, “chew on that”.
I shall be doing exactly that and doing some more data-crunching over the months to come. I have some working theories as to these counter-intuitive stats (some you can guess at if you’ve read my earlier post-GN analysis) but I’d like to verify these statistically. By the time we start to look at the GN 2014 entries and weights, I aim to post a more detailed explanation of the sometimes weird and wonderful side-effects of compression of the GN weights……….IMHO, the BHA’s equivalent to the Bank of England’s Quantitative Easing programme…..…….to both of which “the law unintended consequences” appears to be apposite.
Sorry Peanuts, what is handicap compression ? Is it say a horse should be handicapped at ten stone but is actually given higher rating of ten stone four so that there is less gap between them and the top weight?
It's the handicapper using his discretion to adjust the marks used to frame the weights for runners in the GN away from their prevailing Official Ratings. The GN is open to horse of all ratings (though it now requires a minimum of OR 120) but because top-weights have such a poor record in the race (Rummy winning it with topweight in 1977 was the last to do so and the first since the 1930s) by 2000 the BHA was concerned that horses with high ratings tended not to enter. As a result you had smaller than maximum fields and a lot (up to 81% !) of the field running from out-of-the-handicap (i.e. carrying the minimum 10-00 but this representing more weight (sometimes as much as 2 stone more) than their OR). In 1999 Bobbyjo won the GN running from a full stone out-of-the-handicap (he had a rating of just 128). Handicapper Phil Smith said: "Looking back over the history of the race, we realised that the highly weighted horses had a moderate record, so we thought something needed to be done to try to not overburden the better horses." So the BHA gave the handicapper discretion to "compress the handicap". The GN is the only race in which he has such discretion. So, when Tidal Bay was entered this year, he had the highest prevailing OR (171) of the entries but, in framing the GN weights for all the entries, Smith adjusted Tidal Bay's mark down to 162 (- 9lbs). He also adjusted down the ORs of the next highest rated entries, Alberta's Run (-8lbs) Imperial Commander (-4lbs) What A Friend (-4lbs) and Weird Al (-8lbs) but left all the others to run off their prevailing marks (though he also tends to raise the marks of horses with particular aptitiude over the GN fences. Famously he did so by 8lbs in order to guarantee Amberleigh House a run in the race in 2003 [in which he was 3rd], for which he would not have made the cut off his previaling lower OR). This is "compression of the handicap", designed to give the highest-rated horses in the field a better chance of winning the race. Whether it has done so effectively is a matter of opinion. It certainly has increased entries from higher-rated horses, helped to ensure maximum fields, raised the average OR of runners and reduced (often eliminated) the number running from out-of-the-handicap). It also the effect of course of dragging more runners into the 11.00+ camp. Moreover, if the topweight at the time of the weights-framing subsequently defects, all of the weights rise and even more are dragged into the 11.00+ camp. The fact is that over the 12 years that the weights have been compressed only 1 horse favoured by the handicapper has won; Neptune Collonges (by a nose).
Sorry, I had to dash this morning. Of course Tidal Bay and Alberta's Run did not run in the end and, having both defected before declarations (when the actual weights to be carried are finalised), the weights for all runners went up 4lbs and Imperial Commander, having initially been allotted 11-06 became the new topweight with 11-10. His mark had been dropped 4lbs from his current OR when the handicapper framed the weights and so he remained theoretically 4lbs "well-in" against most other runners. But that counted for nothing (I would argue because of his missed prep and other poor stats) and he PU'd. If there were no compression and the handicapper had not dropped IC's mark (or alternatively if Tidal Bay had run), every other runner would have run with 4lbs less weight and instead of 15 horses running with 11-00+, only 9 would have carried 11-00+. That might have made a big difference to Teaforthree in particular as he would have had 10-13 (might have been neck and neck with Aurora's Encore at the line). It could also have made a big difference to Seabass's finishing position (he would have carried 11-02 without compression, only 4lbs more than in 2012 but with a more targetted prep this time) who was still in touch before running out of gas around the last fence (might have been placed as last year). No compression would also have meant that 7 (incl Aurora's Encore by 1lb) would have been running from out-of-the-handicap instead of only the 3 that actually did. Hence, my view that compression could actually favour lightly-weighted runners in certain circumstances by pushing those with otherwise strong GN profiles a crucial few lbs above (or further above) 11-00. It would be churlish and ridiculous to say that compression has been negative for the race but the effects are far from straightforward and vary from race to race. In general, it certainly hasn't made life easier for most horses with 11-00+ (arguably the reverse) and it doesn't necessarily help materially those whose marks have actually been adjusted down. Regardless of a relative weight advantage against other runners, a horse has to have the right profile to win the GN and it needs to have a particularly strong one to do it with 11-00+. The horse's ability to carry an absolute level of weight in this unique test over 4.5m is still critical (some can only do it with much less than 11.00, fewer can do it with 11-00+ and a very rare one could do it with topweight - if only he had been able to handle the occasion and fences, I firmly beleive that Synchronised had the exceptional stamina and quality to have done so. IMHO the run of 11-00+ winners from 2009~12 is simply a matter of numbers. The fact is that the appearance (largely an illusion) of a better chance for top-rated horses, that has been created by compression, has attracted many more entries from higher up the ratings, including more with the strength of profile to win a GN with 11-00+. All but 1 of the 5 11-00+ wins since 2005 have been (at best) an indirect rather than a direct result of compression.
......and 1 final stat nugget regarding the received wisdom that lightly-weighted horses were favoured pre-compression. In the 11 pre-compression GNs from 1988~1999, only 2 (18% of) winners came from the bottom half of the race-card (Little Polveir and Bobbyjo and both were in the top 60% of the card). In the 12 "compression GNs" from 2002~13, 5 (42% of) winners have come from the lower half of the card, 4 of them in the bottom 40% of the card. I'm not suggesting at all that handicap-compression has been a bad thing for the GN but this stat reflects the fact that, despite being a process that was conceived (and is commonly thought) to favour the higher-rated runners, the actual effects of compression are more complex. TTFN
Final comment on the subject: The lack of success pre-compression of the bottom 40% of the card is hardly a surprise since, in every race from 1988~1999, the bottom 40% (at least) were all out-of-the-handicap. So, while generalisations are always dangerous, the stats set out in the posts above would seem clearly to support the conclusion that, GENERALLY SPEAKING, the group of horses that were most disadvantaged pre-compression and to which compression has operated to give a fairer crack of the whip are: NOT (as commonly supposed) the highest-rated runners in the GN NOT those carrying 11.00+ (compression has acted against the average runner with 11.00+) BUT those in the bottom 40% of the card that previously would have run from out-of-the-handicap but now, courtesy of compression, now usually run at their correct marks relative to most other runners in the race. Of course they still have to have the right profile to have a chance of taking the prize but, as I said earlier, this is an example of that good old law of unintended consequences. That's it from me.........roll on Aintree 2014.
Peanuts, in addition to being highly entertained by your tips before the race and loving the winnings they provided me with, I have to say that I find the commentary of your model and the analysis of its effectiveness after the race to be absolutely fascinating. Even after studying economics at uni, I am yet to see a better real world, profit-making application of statistical and analytical theory than your model!
Many, many thanks for sharing its secrets and makeup.
Cheers Allez, you're too kind. Got to say you have the stamina worthy of a GN winner to have persevered with my post-race musings this year. All the best pal.
Thanks Peanuts. Good read even for non horsey people like myself.
Cheers Chief, much appreciated (and sorry it was such a long-winded answer to a simple question). I'll stop picking at this issue now but it only just dawned on me that Neptune Collonges' nose is perhaps responsible for keeping the "compression illusion" alive, at least for the timebeing. If Sunnyhillboy (another one from the bottom 40% of the card) had held him off, those comparative winners' stats that show that it is the lighter-weighted runners on the card that tend to get most benefit from compression might even get the journos scratching their heads.
Comments
Silver By Nature and Rose Of The Moon both non-runners .
Certainly more like the ground that Aurora's Encore will relish and he could well run a big one this afternoon despite it being only 14 days since Aintree. Like the GN, counter-intuitively the stats suggest that being “fresh” is not an advantage in the Scot Nat. AE’s close 2nd last year was an identical 14 days after his last run and win (though well short of 4.5m) and Hello Bud (2009) and Joes Edge (2005) both won the SN having won their previous races just 9 and 7 days before respectively.
So, expect a big run from the National hero……….but no “Rummy”.
On good ground, the topweight did win the SN twice in the 1990s and was runner-up to Hello Bud by 0.5L in 2009 but in the last 9 renewals on (officially) Good ground (excl GF), horses carrying <10.12 account for all winners and 7/10 of those placed <4L of them.
All 4 of mine below 10.12.
The Off is at 3.50 (ch 4).
2 Big Occasion
3 Mister Marker
4 Tour Des Champs
5 (I think) Fill The Power
Well, I'll be deleting that spreadsheet.
Roll on Aintree 2014.
Ps your sacked
;0)
Sorry Chief,
a man's got to know his limitations.
That's my excuse and I'm sticking with it!
:-(
In case anyone is interested in how I intend to do it, this is the 2nd of 3 Parts to my post-GN stat-analysis. It completes the explanation of the tweaks to the model, mainly in respect of the poor performance of runners with 11.00+. The 3rd Part (to come shortly) will be a statistical analysis of the actual (contrasting with the apparent) effects over the years of the handicapper’s compression of the GN weights.
GN ANALYSIS - PART 2: Final Tweaks To The Model
My initial reaction to the 2013 race was that it was a “pre-compression”-type result, with only 1 of the first 11 home carrying 11.00+ (15 runners carried 11.00+ of which 4 finished: half the completion rate of those carrying <11.00). However, as explained in Part 3 to come, there are some erroneous assumptions about the pre-compression era and the compression of the handicap may not have had the effects that are commonly supposed. My conclusion now is that the 2013 GN result is not some aberration from the “new normal” but confirmation that it is just as hard as ever (in certain cases it may even be harder!) to carry 11.00+ to victory in the GN.
Before the race I had suggested that the withdrawals of Tidal Bay (dropped 9lbs from his current mark) and Alberta’s Run (-8lbs) and consequent 4lb weight-rise for the field had half-neutered the compression and, from the model’s perspective, could materially affect the result. To recap, this was because:
1. All 3 remaining runners favoured by the handicapper (Imperial Commander [-4lbs], What A Friend [-4lbs] and Weird Al [-8lbs]) had had interrupted campaigns and/or missed preps or simply poor seasons (no amount of favourable relative weight treatment would be enough to compensate); and
2. The 4lb rise in weights changed the task for several former GN heroes from tough-assignment to mission-impossible (especially Seabass, Sunnyhillboy, Ballabriggs and Big Fella Thanks).
These horses together accounted for 7 of the top 9 on the card and all carried 11.04+. All of them under-performed even the model’s moderate (at best) rating of their prospects (some significantly).
It led me to trial a tweak to the model, which increased (by a stepped scale according to weight to be carried, starting at 11.00+ with a step-up at 11.06+) the standard penalties applied across a range of criteria, including age, runs in 3m chases, form over 3.5m+, days since last run, quality of prep, season’s form, “spring ground” form and, if applicable, weight increase on prior GN. Not only did this tweak reconcile the poor runs this year of most horses with 11.00+ but, in back-testing the tweak against all GNs since 1988, I was amazed at how robust this change to the algorithm is, even when applied to the recent “compression GNs” when horses with 11.00+ won and/or filled 3 of the first 5 places (all GNs 2008~12) and even when there was no weight-rise after the handicapper had framed them (2010~12).
So this tweak is now confirmed and its logic is simple: despite compression, it is still so demanding to carry 11.00+ in the GN that there must be virtually NO chink in the stats-armour of such runners if they are to have a chance of being in the shake-up, and none if they are to win.
The tweak also corrected the model’s underestimation of the the effect of the weight-rise on those that were brought into the 11.00~11.03 range. That was (IMO) the critical factor in T43 not quite getting home (the tweak reconciles his virtual dead-heat with Cappa Bleu) and Join Together (an 8 year-old with poor spring-ground form) running OK but well below expectation.
Finally, together with other small adjustments, the “11.00+ tweak” reconciles the starkly different runs of two horses with very similar profiles but different weight burdens: Oscar Time (4th 20L with 10.11) and Ballabriggs (PU with 11.04). Both 12 year-olds, respectively the close runner-up and winner of the 2011 GN but with no win since and a 10L near-miss over 3m the best of their season’s runs, the difference in absolute weight burden would seem to have been highly significant; in addition to Oscar Time clearly relishing a return to Aintree and decent ground (he seems to be one of those horses, like Mon Mome, that had excellent form on heavy in his early days but, since injury, simply doesn’t want to run on the testing ground he faced all season).
So that is the model tweaked for another year. It’s now in shape to run the slide-rule over 80 or so entries and their weights next February and, hopefully, see us right.
Part 3 to come (stat-analysis of the effects handicap compression).
In his reply (below) Phil Smith makes the point that he was right to raise AE 11lbs since this equated to the 9L margin over Cappa Bleu (due to go up 2lbs for his good run at Ascot). If he had agreed with Harvey, CB would only have had a 3lb pull for a 9L defeat. But that argument only makes sense if you equate 1lb to 1 length, which is palpable nonsense over 4.5m.
It is commonly accepted that over 5f 1lb = 0.33L and only over 2m does 1lb = 1L. Over greater distance, logically every 1lb of weight should have a larger effect upon performance (probably exponentially so). Willie Mullins has estimated it at 1lb = 2~3L for the GN. If one takes Mullins 2-3L guestimate, that suggests that Harvey Smith is spot on and AE should have had a 5lb rise.
My own analysis supports Mullins’ 2~3L estimate but only at the 10.00~10.07 weight range (which is the relevant range here because AE carried 10.03). In my view, the equation also depends significantly upon (and grows exponentially with) the absolute weight that a horse is carrying over a marathon trip such as 4.5m. To take an extreme example, if you loaded 15 stone on any horse and asked it to run at 15~15.5 secs per furlong they probably would not be able to go as far as 3m let alone 4.5m. Therefore, at the margin, every 1lb added at and above 11.00 will be more of a burden that at 10.00). Based upon my stats, in the GN, roughly, 1lb of extra weight is at least 4-5L at 11.00.
I shall be returning shortly to the matter of Phil Smith in my analysis of the effects of compression of the GN weights but I do have doubts as to whether this bloke really is up to the job…….apart from which his arrogance (later in his blog, he tells Sue and Harvey Smith they should run AE in the Bet365 at Sandown, not the Scottish National) is totally unbecoming of a BHA Head of Handicapping…………IMHO!
QUOTE
National winners and their runs in handicaps
BHA head of handicapping's blog - by Phil Smith
LAST week I explained how I came to the decision of raising Auroras Encore from 137 to 148 after his win in the John Smith's Grand National. It seemed a fairly logical decision to me, but it seems it has upset Harvey Smith who thought he should have been raised by only 5lb not 11lb.
As I explained Cappa Bleu was due to go up by 2lb before the race so that if I had indeed only raised Auroras Encore by the 5lb suggested by Harvey then Cappa Bleu would only have had a 3lb pull for a nine-length defeat. I think most impartial observers would feel that in that case Auroras Encore would have had a better than equal chance if the first two in the Grand National were to meet next time…..
UNQUOTE
COMPRESSION OF THE GRAND NATIONAL WEIGHTS:…………….separating fact from fiction.
When betting in the GN, particularly if one has a stats-based approach, it’s important to try to determine what effect the handicapper’s compression of the GN handicap (at his discretion) tends to have on the prospects for certain types of runner, if not the whole field.
Compression started in 2002 and such is its apparent success that it is now the received wisdom that to carry 11.00+ is no longer the impossible burden that it once was. No GN winner carried 11.00+ from 1989 to 2004 inclusive but all 4 winners from 2009 to 2012 carried 11.00+. So, it’s obvious that compression has succeeded in its mission of making the GN easier for horses carrying 11.00+ and/or assisting the best-rated horses in the field to go well……..…………………...isn’t it?
Well, the 4 wins are in the record book but, if you think that it is easier to carry 11.00+ now and/or that the horses favoured by the handicapper are reaping the benefits, you might be surprised by the following stats, which compare 11 pre-compression GNs (from the last 11.00+ winner in 1988 to Bobbyjo’s win from a full stone out-of-the-handicap in 1999) and the 12 “compression GNs” (from 2002 to 2013):
• Proportionate to their representation, in the pre-compression period horses with 11.00+ were more than x2 as successful at filling one of the 1st~5th places in the GN as those with 11.00+ in “compression GNs”.
• Pre-compression, horses with 11.00+ had a 58% completion rate (41% of horses with <11.00 completed). In the “compression GNs”, horses with 11.00+ have a 32% completion rate (38% for horses with <11.00); i.e. while completion rates for <11.00 are broadly unchanged, there has been a significant fall in the rate of completion by those with 11.00+ in the compression GNs.
Yes, but that’s surely because compression has encouraged more entries from higher up the handicap and, by definition, dragged more horses into the 11.00+ camp, many of them with unsuitable credentials for the GN. That is all true but consider these stats:
• Horses from the top 5 on the race-card (i.e. those horses likely to be most favoured by the compression) filled 31% of the 1st~5th finishing places in the 11 pre-compression GNs. In the 12 “compression GNs” they have filled only 15% of the 1st~5th places.
• Spreading the net wider, horses from the top 10 on the race-card filled 42% of the 1st~5th finishing places in the pre-compression GNs, compared to 20% in the “compression GNs”
The figures and ratios change only slightly if one adjusts for the specific (often lower) field-size of the pre-compression GNs (average 36.5 vs 39.9): the only change being that the top 10 pre-compression rate falls to 40% from 42%.
So, not only are runners with 11.00+ now faring worse collectively with weight-compression but runners that are likely to be most favoured by the compression (defined crudely either as the top 5 or top 10 on the race-card) are only half as successful at making the frame as were their counterparts pre-compression.
A similar ratio of outperformance applies for wins aswell:
• Pre-compression, the top 5 on the race-card contributed 2 wins in 11 GNs. This compares with 1 win in 12 “compression GNs” (and that solitary win was by a Nose last year)
• The top 10 on the race-card contributed 7 wins in 11 pre-compression GNs (6 when adjusted for low field-size), compared to 3 wins in 12 “compression GNs”.
• Moreover, only 1 of those 3 compression wins (Neptune Collonges’ Nose) was by a horse whose weight was dropped by the handicapper from its prevailing mark (2lbs in his case), though it is just as relevant to note that his GN prep (after the GN weights were framed) had been so impressive that his real mark was due to rise by 5lbs; so he was in fact 7lbs well-in in the GN. Both of the other “top 10” wins were by horses (Don’t Push It in 2010 and Ballabriggs in 2011) not specifically advantaged by the handicapper, though there were runners in both races whose weights had been dropped by 4lbs+ from their current marks (none of them completed).
As they say, “chew on that”.
I shall be doing exactly that and doing some more data-crunching over the months to come. I have some working theories as to these counter-intuitive stats (some you can guess at if you’ve read my earlier post-GN analysis) but I’d like to verify these statistically. By the time we start to look at the GN 2014 entries and weights, I aim to post a more detailed explanation of the sometimes weird and wonderful side-effects of compression of the GN weights……….IMHO, the BHA’s equivalent to the Bank of England’s Quantitative Easing programme…..…….to both of which “the law unintended consequences” appears to be apposite.
Have a great summer.
COYR
The GN is open to horse of all ratings (though it now requires a minimum of OR 120) but because top-weights have such a poor record in the race (Rummy winning it with topweight in 1977 was the last to do so and the first since the 1930s) by 2000 the BHA was concerned that horses with high ratings tended not to enter. As a result you had smaller than maximum fields and a lot (up to 81% !) of the field running from out-of-the-handicap (i.e. carrying the minimum 10-00 but this representing more weight (sometimes as much as 2 stone more) than their OR). In 1999 Bobbyjo won the GN running from a full stone out-of-the-handicap (he had a rating of just 128).
Handicapper Phil Smith said: "Looking back over the history of the race, we realised that the highly weighted horses had a moderate record, so we thought something needed to be done to try to not overburden the better horses."
So the BHA gave the handicapper discretion to "compress the handicap". The GN is the only race in which he has such discretion.
So, when Tidal Bay was entered this year, he had the highest prevailing OR (171) of the entries but, in framing the GN weights for all the entries, Smith adjusted Tidal Bay's mark down to 162 (- 9lbs). He also adjusted down the ORs of the next highest rated entries, Alberta's Run (-8lbs) Imperial Commander (-4lbs) What A Friend (-4lbs) and Weird Al (-8lbs) but left all the others to run off their prevailing marks (though he also tends to raise the marks of horses with particular aptitiude over the GN fences. Famously he did so by 8lbs in order to guarantee Amberleigh House a run in the race in 2003 [in which he was 3rd], for which he would not have made the cut off his previaling lower OR).
This is "compression of the handicap", designed to give the highest-rated horses in the field a better chance of winning the race.
Whether it has done so effectively is a matter of opinion. It certainly has increased entries from higher-rated horses, helped to ensure maximum fields, raised the average OR of runners and reduced (often eliminated) the number running from out-of-the-handicap). It also the effect of course of dragging more runners into the 11.00+ camp. Moreover, if the topweight at the time of the weights-framing subsequently defects, all of the weights rise and even more are dragged into the 11.00+ camp.
The fact is that over the 12 years that the weights have been compressed only 1 horse favoured by the handicapper has won; Neptune Collonges (by a nose).
Of course Tidal Bay and Alberta's Run did not run in the end and, having both defected before declarations (when the actual weights to be carried are finalised), the weights for all runners went up 4lbs and Imperial Commander, having initially been allotted 11-06 became the new topweight with 11-10. His mark had been dropped 4lbs from his current OR when the handicapper framed the weights and so he remained theoretically 4lbs "well-in" against most other runners. But that counted for nothing (I would argue because of his missed prep and other poor stats) and he PU'd.
If there were no compression and the handicapper had not dropped IC's mark (or alternatively if Tidal Bay had run), every other runner would have run with 4lbs less weight and instead of 15 horses running with 11-00+, only 9 would have carried 11-00+. That might have made a big difference to Teaforthree in particular as he would have had 10-13 (might have been neck and neck with Aurora's Encore at the line). It could also have made a big difference to Seabass's finishing position (he would have carried 11-02 without compression, only 4lbs more than in 2012 but with a more targetted prep this time) who was still in touch before running out of gas around the last fence (might have been placed as last year).
No compression would also have meant that 7 (incl Aurora's Encore by 1lb) would have been running from out-of-the-handicap instead of only the 3 that actually did.
Hence, my view that compression could actually favour lightly-weighted runners in certain circumstances by pushing those with otherwise strong GN profiles a crucial few lbs above (or further above) 11-00.
It would be churlish and ridiculous to say that compression has been negative for the race but the effects are far from straightforward and vary from race to race. In general, it certainly hasn't made life easier for most horses with 11-00+ (arguably the reverse) and it doesn't necessarily help materially those whose marks have actually been adjusted down. Regardless of a relative weight advantage against other runners, a horse has to have the right profile to win the GN and it needs to have a particularly strong one to do it with 11-00+. The horse's ability to carry an absolute level of weight in this unique test over 4.5m is still critical (some can only do it with much less than 11.00, fewer can do it with 11-00+ and a very rare one could do it with topweight - if only he had been able to handle the occasion and fences, I firmly beleive that Synchronised had the exceptional stamina and quality to have done so.
IMHO the run of 11-00+ winners from 2009~12 is simply a matter of numbers. The fact is that the appearance (largely an illusion) of a better chance for top-rated horses, that has been created by compression, has attracted many more entries from higher up the ratings, including more with the strength of profile to win a GN with 11-00+. All but 1 of the 5 11-00+ wins since 2005 have been (at best) an indirect rather than a direct result of compression.
In the 11 pre-compression GNs from 1988~1999, only 2 (18% of) winners came from the bottom half of the race-card (Little Polveir and Bobbyjo and both were in the top 60% of the card).
In the 12 "compression GNs" from 2002~13, 5 (42% of) winners have come from the lower half of the card, 4 of them in the bottom 40% of the card.
I'm not suggesting at all that handicap-compression has been a bad thing for the GN but this stat reflects the fact that, despite being a process that was conceived (and is commonly thought) to favour the higher-rated runners, the actual effects of compression are more complex.
TTFN
The lack of success pre-compression of the bottom 40% of the card is hardly a surprise since, in every race from 1988~1999, the bottom 40% (at least) were all out-of-the-handicap.
So, while generalisations are always dangerous, the stats set out in the posts above would seem clearly to support the conclusion that, GENERALLY SPEAKING, the group of horses that were most disadvantaged pre-compression and to which compression has operated to give a fairer crack of the whip are:
NOT (as commonly supposed) the highest-rated runners in the GN
NOT those carrying 11.00+ (compression has acted against the average runner with 11.00+)
BUT those in the bottom 40% of the card that previously would have run from out-of-the-handicap but now, courtesy of compression, now usually run at their correct marks relative to most other runners in the race. Of course they still have to have the right profile to have a chance of taking the prize but, as I said earlier, this is an example of that good old law of unintended consequences.
That's it from me.........roll on Aintree 2014.
Many, many thanks for sharing its secrets and makeup.
Got to say you have the stamina worthy of a GN winner to have persevered with my post-race musings this year.
All the best pal.
I'll stop picking at this issue now but it only just dawned on me that Neptune Collonges' nose is perhaps responsible for keeping the "compression illusion" alive, at least for the timebeing. If Sunnyhillboy (another one from the bottom 40% of the card) had held him off, those comparative winners' stats that show that it is the lighter-weighted runners on the card that tend to get most benefit from compression might even get the journos scratching their heads.