Research on roughies

[hermes]

Numerous comments on this forum have prompted me to look at a totally different category of runners lately. For instance, Bhagwan says never bet on TAB numbers beyond 10 for statistical reasons. But Placegetter, I think it was, says he deliberately does the opposite to other punters. Then someone sung the virtues of TAB 12. So...

This time I've isolated runners TAB 11, 12, 13, and 14 where they are within four kilos of the top zip weight (The Zip Star) given in the Sportsman ratings. (The great majority of winners overall are within 4 of zip.) What I am after are cases where the zip ratings suggest the handicapper has underestimated a horse. Looking for long shots.

Its a fascinating category of runners. A surprising number of favourites but they rarely get up. Some race meets, no qualifiers. Other meetings, stacks. Some meetings, no results at all - wipe out. Other meetings, big returns. It is very uneven as you'd expect when chasing roughies.

I looked at random race meetings, metro and country, all races, any races back to early May - but obviously only for races where the Zip rating is available. I found 345 qualifiers. Only 34 winners. A disappointing one in ten. Return on places = $316. Return on wins = $483.70. (Queensland TAB prices.) POT of 40%. Average return = $14.20. The lowest return was $3.40 to win.

BUT, the figures are blown out by a $92 winner and a $70 winner. Without these, we'd be behind. This is worrying with such a low strike rate. So are they freaks or can you rely on some mega-winners in 350+ bets? Qualifiers regularly start at 60/1+. Sooner or later one will get up. The question is how often? In my sample, twice.

If you remove the $92 winner and only bank on one mega-winner in 350+ bets the sums still work, still ahead. Or let's call it a $30 winner instead of $92. That's still $421 return for $345. And in the sample there are $24 winners. I think the within-4-of-zip filter pins enough of them down. The method is sustained, though, by the fact that in bigger fields TABS 11, 12, 13, 14 are often not too bad quality-wise and still pay better than $6 or $7.

These are the results for the raw category with only the zip filter applied. You can then try to tweak it further. For instance, you save more than you lose if you eliminate the TAB 14s. TABS 11, 12, 13 only equals $288 outlay for $419 return. Even taking the mewgawinners out of the equation TAB 14 doesn't pay.

The drawback are the long runs of outs and the temptation to bet for place. There are some great place prices. Best was $23 to place, Pi Meson, R4 Belmont, 13/7/02.Then $15 to place. Lots of $10 to place. But not enough of them. Bet win only.

What if there is more than one qualifier in a race? I have no luck at all trying to pick which will win and which won't, so back 'em all. If any of them win you'll pay for the others. To make a viable system out of this though you need a further filter at this point.

You also need deep pockets and you need to get used to the idea that you'll collect nothing for 20 or 30 or more bets in a row. For example, my sample went:

Strike - 21 outs
Strike - 7 outs
Strike - 12 outs
Strike - 15 outs
Strike - 12 outs
Strike - 42 outs
Strike - 32 outs
Strike - 16 outs
Strike - 7 outs
Strike
Strike - 3 outs
Strike - 2 outs
Strike
Strike - 1 out
Strike - 7 outs
Strike
Strike - 30 outs...

And so on. As you can see, some meetings lots, many meetings zero. You'd be getting mighty thirsty after 42 and 32 outs in a row and there'd be whole weeks you did nothing but lose. Not recommended for those suffering from depression. But when you hit a fertile patch it pays well.

Running a bank of $100, betting $1 to win over the 345 bets you end up with $238.70. The lowest ebb was $72 at bet 43. But therein is the problem. Remove the megawinners and you go bust at bet 147.

(I spent several sleepless nights trying to get the figures for this system to add up. Eventually I realised I once again hadn't been eliminating scratchings, only checking winners. There's often lots of scratchings in this category. It enhanced my figures no end to realise that a horse didn't win because it hadn't run!)

I tried filtering by last start = meaningless. And place percentage = meaningless. The only other filter I've tried is matching qualifiers to the ratings on the Queensland TAB. Result = meaningless. Except the biggest priced winners were all rated below 90 - genuine pedigree roughies. Narrowing the range to within 3 or 2 of the zip doesn't work. So, to net long shots:

Rule 1. TAB numbers 11, 12, 13 (and 14)
Rule 2. Within 4 kilos of the Zip Star horse.
Rule 3. Bet win only.
Rule 4. Where there are several qualifiers in a single race, back them all. (Or find a reliable way to distinguish between them.)
Rule 5. Take up another interest while you wait.

Be good to find ways and means of making it more stable. I'd like to get the strike rate to around 15% at least. Ultimately I fear this method is heading for a situation of break even and you sink or swim on one-in-150 race megawinners.


Hermes

[hermes]

PS:- I anticipate (albeit from a small sample) a continuing pattern of feast or famine. You don't get one winner at a race meeting, you get three. Or none. I can't see any reason in this though. One day Sandown is great, another day catastrophic. A good day at Scone. A bad day at Randwick. A brilliant day at Geelong, wipe out at Wang. What makes a roughies day?

[Equine Investor]

Hermes a great idea, but have you ever considered say backing Numbers 1-2-3-4 only if they are at 10/1 or better? Throws up some amazing winners and placegetters! Usually the horses are Classy types which are just out of form. These horses regularly surprise.

Just an idea - Still think your idea has merit too.

:wink:

[becareful]

Interesting stuff once again Hermes. There is nothing wrong with a strike rate of 10% as such but the inconsistancy would worry me. As I've said before my strike rate is just over 10% but I rarely go for more than 20 bets without a winner and usually pick up at least one per day. You would need balls of steel to be placing bets on 80-1 shots when you have had 30 outs in a row!

Relying on the really big winners to give a profit is also a concern - I think you need a REALLY BIG sample to be confident that these are going to keep coming in. One thing you may like to try is eliminating all starters over say $20 or $25 - it will cut out your two big winners but should also get rid of a lot of other long-shot bets and hopefully improve the strike rate.

Also I don't particularly like systems based on TAB numbers but that's just my preference.

[becareful]

On 2002-08-16 16:26, hermes wrote:
What makes a roughies day?


Wet track at Sandown always seems good for the roughies!

Actually - have you tried analysing your results by track condition and/or rails position - just a "gut feel" but wet tracks and rails well out seems to give a lot of roughies. Don't have any solid data to back this up though.
_________________
"So certain are you. Always with you it cannot be done. Hear you nothing that I say?" - Yoda

<font size=-1>[ This Message was edited by: becareful on 2002-08-16 16:52 ]</font>

[hermes]

Got it! Reducing the zip range doesn't work but going the opposite direction does.

Rule 1. TABs 11, 12, 13, 14.
Rule 2. Only those EXACTLY four kilos from the zip star.

In my sample, 146 bets. Win = $345.50. Place = $173.30.

ALL the good winners are in this sub-category. And if we remove the megawinners the results are still healthy:

144 bets. Win = $182.60. Place = $145.80.

But longest run of outs still equals 21.But if places pays or breaks even this could be broken up with place bets where longest run of outs is 12.

By these figures, a megawinner now and then would be the icing on the cake, not the salvation of the system.

Quite a few in this category tommorow.

Rosehill - R7 #12 Conzeal.
Moonee V. - R3 #13 Yallah Limit, R7 #11 Star Doc, R8 #12 Monde Special.
Eagle Farm - R5 #Flighting and #13 Janneke
Vic Park - R5 #12 Spar, R7 #12 Street Parlour.

Sandown on Sunday - R6 #13 Keltic Verse, R8 #12 Yakama.

Thanks for all contributions and suggestions. Much appreciated.

Hermes

[hermes]

And let's add Belmont Park. R3 #12 Sergesky. R5 #14 Proven Destiny.

Bringing my database up to date I'm currently on a run of 8 outs. But Sat 10th August was a beauty! Flemington R3 #13 Royal Calcutta - $40.30. Then Randwick R7 #14 In the Night = $8.10. Looking good.

Hermes

[hermes]

I can't believe just how tight the punt is!

Equine Investor suggested backing TABs 1,2,3 where they start over $10. Seemned like a sensible strategy to me so I tried it. Result: I found 314 cases, return = $314.40. Similarly, I hit upon the idea of chasing the runner one point under the Zip Star (with a few other filters). I found 160 cases, return = $159.80. Then I tried zip stars where they are not the favourite. Found 210 cases, return = $211.20.

So I spent the whole day at it yesterday (between races) and came up with three methods that all break even.

The punt is as reliable as gravity. If you've got a method that pays well - consistently - well done, you're defying gravity.

Sandown today:

R6 #13 Keltic Verse, R8 #12 Yakama.

Hermes

[becareful]

Hermes,

I would say you are doing really well to find three different methods that all break even - given that most selection systems lose money. I would say they are all worth further investigation to see if you can find some additional filters to give you that extra 10% or so you need to make them profitable. Even something simple like looking at what Divi+ would have paid instead of TAB may be enough (or what the highest TAB price is instead of single TAB price if you are dealing with midweek results).

[Bhagwan]

Well done Hermes for you fortatude , I notice one of your selections got up on Sunday , which practicly won back the losses from Sat.

I would`nt give up ,if I were you.
You are knocking on the door of a breakthrough.
You`ve done very well indeed just to break even .
The Punt is a funny thing , the recurring percentages will not be denighed.

Frustrating stuff, but keep plugging away.
I`m only too happy to help you with maybe some fresh ideas , to research.

[hermes]

Becareful,

I looked over wet tracks in my samples. A higher proportion of long shots but not my long shots unfortunately. But I need a bigger sample, as you say.

Bhagwan, any fresh ideas you can throw my way would be greatly appreciated.

Here's a rather weird idea, but of all the methods I've trialed lately it shows the best results. It comes from the idea that statistics refer to a mythical world of averages where there are, say, fourteen and a half horses in every race, and every race is over some mythical distance that only exists in average-land. I set off looking for the MOST mythical horse...

*Only look at races with 10-20 starters (field minus scratchings).
*If ten starters, watch the market for the runner nearest to $5. Its the selection. If eleven starters look for the runner nearest $5.20. And so on by the following scale:

12 = $5.50
13 = $5.90
14 = $6.25
15 = $6.60
16 = $7.15
17 = $7.60
18 = $8.30
19 = $9
20 = $10

This will place you in a nice band of runners with a 15% strike rate and average return of $6.50.

My sample was 342 races, return on wins of $352.50. (Places deliver a LOT).

With a bit of filtering this could make a substantial system I think. It is sometimes difficult to judge which of two or more runners will be nearest the target at jump but in most races there is a clear selection reliable from the 3 min. mark.

The price targets are taken from the inverse raw chances of a runner in a field that size. In a field of ten if all things were equal - they never are in the real world! - all runners have a 10% chance. In a field of 20, a five percent chance. Invert those. The runner nearest that is the horse most like the great mythical horse, Mr Average. In the mythical world this horse will run a dead heat with every other horse every race. In the real world this horse will win 15% of races and pay POT on unfiltered selections.

Like I said, a slightly weird idea, and probably based on totally silly reasoning, but it shows promise. Any suggestions most welcome.

Hermes

[hermes]

The method I posted a few back continues to pay. That is, TABS 11, 12, 13, 14 that are rated exactly 4 kilos from the zip star horse. Back each way. Over a bigger sample now, still a healthy POT even without the megawinners included in the mix. Over 528 races 11% POT on wins, 4% on places.Set and forget. The only worry is a slump in the figures in the samples from Spring 2001 races.

Nevertheless,

Three placegetters in 5 bets at Seymour yesterday, Touch of Heart paid $3.20 for the place, Longmire paid $6.20, Southern Command paid $2.30.

Yakama won in the last at Sandown to rescue Saturday. Today, another winner at Weribee R2 #14 Noble Start, $29.30 the win, $5.20 the place. A typical win for this method. And you get enough of them.

This one has legs.

Hermes