Ratings maths

[thorns]

Say for example you had a set of ratings which showed a 79% strike rate from the top 5 rated horses, and the average win dividend was $5.47, would this show a level stakes profit? I'm not enitrely sure which way you would calcutlate the profit/loss of this.

Would it be $5.47/5=1.094

Then 1.094x79=86.426 to work out the P/L?

Any help on this would be appreciated.

And if that is the correct method that I have used above, any suggestions on how to try and improve this? Perhaps only backing the top five horses which are paying $5 or more?

[Dennis G]

Say for example you had a set of ratings which showed a 79% strike rate from the top 5 rated horses, and the average win dividend was $5.47, would this show a level stakes profit? I'm not enitrely sure which way you would calcutlate the profit/loss of this.

Would it be $5.47/5=1.094

Then 1.094x79=86.426 to work out the P/L?

Any help on this would be appreciated.

And if that is the correct method that I have used above, any suggestions on how to try and improve this? Perhaps only backing the top five horses which are paying $5 or more?Hi thorns,
I believe your maths are correct, but you seem to be assuming that the 86.426 is the return on 79, when it actually the return on 100.00 resulting in a loss.

AFA only backing horses paying $5.00 or more, I think you would find a significant reduction in your strike rate. Maybe someone with a database could help confirm this.

Rgds,
Den

[thorns]

thanks for that, I thought that the 86 was LOT, but didnt make that very clear, but the fact that your maths comes up the same is what I was really after.

[Silver_and_sand]

Another way to estimate how your system would perform would be to estimate the results after a certain # of races, say 200 for the sake of convenience.

# of selections = 200 races x 5 horses per race = 1,000 selections
# of winners = a winner found in 79% of 200 races = 158 winners
average price per winner = $5.47

So based on $1 bets, your 1,000 selections will return $5.47 x 158 = $864.26, which means after 200 races, you could expect a loss of roughly $1,000 - $864.26 = $135.74, or alternatively, a loss on turnover (LOT) of 13.574%.

Personally I would steer clear of trying to bet on so many selections in each race. Also, one of the most important lessons I've learnt so far is that you don't have to bet in every single race. When developing a system, you should constantly be trying to think of new filters that both increases your profit on turnover (POT), as well as your strike rate. Look for similarities in the majority of your winners, as well as the majority of your losers. Weeding out more losers from your system is just as important as finding the winners.

And then when you've refined your selection process to be as good as it can be, then focus on what kind of staking plan is most appropriate. Personally I prefer betting a certain percentage of the bank on each selection, which helps preserve the bank when a run of outs strikes, but also helps compounds profits when the winners come rolling in. Hope that helps. Good luck mate.

[thorns]

Thanks for the reply. I realise betting that many horses would be hard work to make a profit, I guess really I was more interested in finding the P/L, as those stats are form a particular ratings provider, and the way they were advertising made it sound impressive. But now that I have seen the actual "real life" figures, its not that great after all.

Cheers

[Grand Armee]

thorns - no ratings method ever devised in the history of the world can level stake 5 horses a race regardless of price over virtually every race and make a profit. It's an unrealistic expectation. Don't look for that as your answer because you won't find it, anywhere. 79% strike rate in the top 5 is a very good strike rate, and given that the LOT is only 11% level staking all 5 of them (and I do mean ONLY, because backing 5 horses a race flat stake regardless of price is a fast way to the poor house, therefore 11% is a good result in that scenario), then it is highly likely that you can turn that into a sustainable POT, with sound betting practises. As always, the key is to look for value in the selections, not to flat stake regardless of price.

Good luck.

[Benny]

I back 5 horses in a race according to my ratings, but usually only 3 or 4 races a day.

Up to the 22/1 the results are

55 races
275 selections
47 winners

out 275.00
in 306.10
+31.10

[michaelg]

Grand Armee,

I am currently testing a sort of mechanical method with real money in qualifying races where I bet half the field, less the fave, to Win using level stakes.

Admittedly there have only been 44 races (for 39 winners) but the POT on Win betting is currently 69.80%. I also box the selections (including the fave) in quinellas for a current POT of 40.80%, and trifectas for a current POT of 47.09%.

Today I missed Swan Hill R9 because I was not at my computer, which was a qualifying race. I would have snared the winner of $14.40, the quinella of $111 but I'm not sure of the $1,450 trifecta. This race will not be included in my results.

I am saying all this not out of conceit or anything like that (I will reveal the rules of the system if it continues to perform but will test the method for a much longer period of time) but just like Benny has stated it is not impossible to bet a large number of horses, but IMHO only if you have rules that select particular races.

Grand Armee, I fully agree when you say virtually every race - I believe you have to be selective in the races you chose.

[Grand Armee]

michael g and benny - that's why i said "virtually every race" and "regardless of price", because they are the stipulations that thorns is talking about.

Both Benny's method and michael's method target a few races only, and therein lies the difference.

Also, I don't want to burst any bubbles lads b/c your results are very promising but you obviously need a much bigger sample.

[Grand Armee]

A good example of the sample size thing: I am currently "live" testing (ie not back testing but recording each day) a method which, after 133 races, is winning 43.8% on turnover. Promising, yes. But I know from past experience that I need far, far more than that.

Preferably, you want a sample of 10,000. That's a bit unrealistic though. IMO 1000 is bare minimum, but I've seen methods change their profile after more selections than that. 3000 is a number where, from my experience, it does seem to settle down. So, I'll be waiting on 3000 selections before I get too excited about this.

[michaelg]

Hi, Grand Armee,

you're not bursting my bubble because I'm aware of the dynamics and unpredictabilities of the game and that any method/ratings/staking plans,etc has to be tested extensively before any could be deemed worthwhile. So I'm not holding my breath but would be surprised if any of my three types of betting eventually went into the red because they are so far in front at the moment. Its early days but time will tell and might prove me wrong. Even though I haven't bet on Duets or First Fours, both are showing (on paper) excellent results, so its quite encouraging.

I started betting my system on Mon 21 Jan so I'm betting on approximately 2 races per day. To bet on 1,000 races would take about 1 and a half years. At the moment I'm considering one year so that it covers all four seasons to see if there are any hot or cold periods.

[thorns]

hi michaelg,

interesting method you seem to have come across there. The races that you decide to bet, I take it you are looknig for races where the favourite is obviously quite vulnerable. To decide this, are you going by the odds of the favourite? Or are you using another method? I understand however if at this stage you are not willing to share this info.

I too have played round the idea of backing as many horses as possible after the fav to try and show a profit. I was looking for races where the fav was paying around the $5 mark, so would in turn have a poor strike rate, and then depending on the next priced horse would back as many runners as possible. Ie if the 2nd fav was showing $6, back the next 6 horses, using the 2nd fav a saver type bet.

I had mixed success with the method, but then on the occasion where the fav, or a outsider which i was not on, got up, the losses were quite heavy. I quite liked the idea though because of the high strike rate, (i'm not the sort of person that could go 20-30 races without a winning bet) hence why i was looking at ratings in my first post. I guess with teh high strike rate, you could use a progressive staking pan, as the run of outs would be pretty minimal, its just finding teh right mix of agression and control to make it really work.

[michaelg]

Hi, Thorns.

We are more or less on the same page when it comes to selecting the bettable races.

I originally had quinellas in mind, because after testing many previous methods where I box multiple selections I realised that in every one of them that quinellas in the long term outperformed trifectas. Maybe its due to the lesser commission by the TAB on quinellas or maybe the impact of flexi betting on trifectas, or even both? But whatever the reason(s) it soon became apparant that quinellas was the way to go.

When I began my current selection method (without real betting) I only recorded the quinella results as I was looking for qualifying races where I expected high quinella divvies. I had little interest in Win or trifecta betting, but soon realised that there was a very good strike rate for both (it is striking one in every two trifectas) with surprisingly good divvies. So when I began betting with real money I included both. Also, previous systems I have worked on showed it was better to omit the fave in Win when betting on the TAB, which I am doing.

What I'm really getting at is even though our selection process is different, have you considered boxing your selections in quinellas because I will be surprised if my quinellas eventually do not outperform my Win and trifecta results. If you have records of your betting selections you might check them to see how quinellas have fared?

[thorns]

I will definitly be looking into the quinella method then as well, at this time I havent particulary been recording my results, more just watching the races when I get home form work and watching how many times, the method I outlined above seemed to succeed. But with your info as well, I might have to start looking at recording teh results to get true idea of how this method is achieving in real life.

Cheers
Paul

[michaelg]

Thorns, getting back to quinellas - there was one qualifying race today. The fave won so I lost on the Win bet, I did not have the third horse so the trifecta bet lost, but I had the quinella for a most acceptable 38% POT.

One race certainly does not support my preference for quinellas, but it might be indicative that they could be superior, even when the fave is involved in the divvy.

[thorns]

With your quinella method, I notice form your last post that you are including the fav, how many horses are you generally boxing up?

[michaelg]

I box half the field in win, quinellas and trifectas, but I omit the fave only in Win betting. If the fave does happen to win the race I am not too concerned because even though the Win bet has lost, there is a current 72% chance of snaring the quinella and about 50% with the trifecta. Of course these percentages may reduce, time will tell...

If there's an uneven number of runners I rate upwards. For example, if the qualifying race has 13 runners I box seven. Half the field (and slightly higher) is a huge number to box but I'm hoping the method identifies those races where the divvy should be worthwhile and that the selection process provides one or two horses that are "overs" in compensation for the large outlay. My best winner so far has been $42 which was an "over" on the TAB.

Thorns, like you I also prefer high strike rates, and this method provides it. All it needs is also to provide a profit.

[michaelg]

Yesterday at Ipswich race 7 the system snared the winner, quinella and trifecta, paying $23.10, $99 and $3,370.

But now looking at the Unitab divvies which paid $25.00, $209 and $6,150 I feel almost cheated. I have never seen exotic divvies double those of another state, and I have often compared them on the IAS website. And no, my minimum bets did not affect the pools.

I can accept that flexi-betting could affect the trifecta dividends, particularly the higher ones, but this does not apply to quinellas.

This is one time that you wish this forum would not censor expletives.

[lomaca]

Yesterday at Ipswich race 7 the system snared the winner, quinella and trifecta, paying $23.10, $99 and $3,370.

But now looking at the Unitab divvies which paid $25.00, $209 and $6,150 I feel almost cheated. I have never seen exotic divvies double those of another state, and I have often compared them on the IAS website.

Hi michaelg, I feel your pain.
I thought NSW TAB and Supertab were the same?
It paid $29.6 win, $281.4 Quin amd $6736 Tri on supertab.
I got the winner (but I skew my rating for long shots), but don't play tri, or quin, alas.
better luck next time!

[thorns]

Yesterday at Ipswich race 7 the system snared the winner, quinella and trifecta, paying $23.10, $99 and $3,370.

But now looking at the Unitab divvies which paid $25.00, $209 and $6,150 I feel almost cheated. I have never seen exotic divvies double those of another state, and I have often compared them on the IAS website. And no, my minimum bets did not affect the pools.

I can accept that flexi-betting could affect the trifecta dividends, particularly the higher ones, but this does not apply to quinellas.

This is one time that you wish this forum would not censor expletives.Ouch!!! I bet that does hurt. Quite strange though that it was UniTAB with the highest prices, usually when I'm testing a system, I use there prices because they are generally the lowest. So if someting make a profit using there prices, it should be even better at other TAB's or bookies.

[michaelg]

Yes, Thorns. More often than not it is Unitab with the lowest divvies.

But as Lomaca says, even Supertab bettered Unitab.

Sooner or later NSW and Vic TABs will most likely be combined which should create some sort of stability, but the punters who bet on Best TAB will be adversely affected. Maybe also with IAS's Super Price.

[michaelg]

The system today snared the trifecta (also the quinella and first four) at Camperdown race 2.

I now don't feel too bad because the NSW TAB trifecta divvy was $1,482 yet $1,0510 on Unitab and a piddling $597 on Supertab, and surprisingly only $1,078 with IAS Superprice.

Swings and roundabouts...

[michaelg]

The system snared the exotics at Hawkesbury race 8. The trifecta here in NSW paid $807, with Unitab $1,484, and with STAB it was $998. Because our TAB is currently the only one, I think, to offer flexi-betting it could be assumed that this is adversely affecting the larger divvies.

Out of curiousity I have been monitoring some of the larger divvies and the above is not unusual.

[Crackone]

You would think it would affect the divi, you can go wider for less % same cost.
Do the smaller ones pay more?

[michaelg]

The smaller trifectas are more or less consistent with the other states. One of the smallest the system has struck was today's Mornington race 10. In NSW it paid $212, on Unitab it was $168 and STAB it was $185, so there's not too great a difference.

The trifecta pools are very small in relation to the number of combinations which easily leads to the the huge discrepencies between the TABs because a $1 bet on three, or even two, outsiders can drastically distort the divvy. In regard to the combinations factor, today's offending race (Hawkesbury race 8) had 14 starters which meant that there was a total of 2,184 trifecta combinations yet the pool only held $34,178 - approximately $17 for every combination. In contrast, there were 91 quinella combinations, the pool held $19,132 - approximately $210 for every combination. The First Four pool is even more inequitable - a few cents for every combination. In the first couple of weeks, when the method snared two particular First Fours where the divvy exceeded the total money held in the pool I then decided the inequity was too unrealistic and unpredictable, so I no longer record F.Fours in the method's results.

Because the faves fill the places for the smaller trifecta divvies, they are not prone to the distortions that the outsiders can create. So, there should not be too much difference in the 3 TABs for the lower divvies. After saying that, I'll probably have another rant if I snare a $100 divvy yet it pays double in the other 2 TABs.

[Crackone]

You mite be better with a bookmaker like IAS they offer (see below )


Bet IASbet SuperPrice and be guaranteed that the payout will exceed at least 2 Australian TABs . In fact SuperPrice often exceeds all Australian TABs for Win, Place, Quinella, Exacta, Trifecta and Multi-betting.

Cheers

[michaelg]

Yes, I'd be better off wIth IAS Super Price.

However the problem is that they only accept exotic bets up to 3 minutes prior to the start time. Because the prices are one of the method rules to identify a qualifying race there have on a few occasions where I did so with about a minute prior to the start, then I have to also identify the selections. Sometimes it's touch-and-go to get the bets on at the TAB. I've even missed some very good divvies because for one reason or another I was unable to get the bets on before the race began. Its a terrible feeling when this happens, even worse than snaring the exotics where NSW TAB has by far the worst divvies.

If IAS would let us bet right up to the start time I'd certainly do so, but until then I will have to stick with the TAB. Its also a pity that they don't offer First Fours because the method has struck some huge divvies.

I've just realised that I have been now betting for one month with real money on the method. There have been 73 races and results are:

Quinellas:
Strike rate of 80.82%
POT of 38.28%.

Trifectas:
Strike rate of 56.16%
POT of 57.81%.

Maybe its just a lucky streak its going through and might now start to go backwards.