Thanks for your reply jfc.

You seem to be on the same wavelength as me to a certain extent.
I consider my first four model to be a very good one. The fact that i turnover millions of dollars a year on this bet type and have continually beaten the 20-25% rake over time shows how good my model is. And no i dont use Bayes Theorem of using win probabilities to predict the placings. I have sufficient evidence to disprove this. The superfecta before the abolished it was my biggest winner. The takeout there was 25% and my POT was huge thanks to a couple of sweeps of the pool. A few times a multiple jackpot would occur and in a big field it was quite easy to hold the only live ticket.

At the end of the day a 20% rake is definitely easy to overcome if you are taking all of the "overs" combinations. I have recently started betting in Hong Kong on the six up and Triple Trios and the overlays in some of the combinations there are huge. An example is no 4 being an unlucky Chinese number. If a no 4 wins in any of these bet types the dividends are significantly inflated. Massive pools and huge carry over jackpots mean the dividends pay "overs" quite frequently.

Just because the TAB have such a huge rake doesnt mean that these overlays dont exist. For example the World Cup has plenty of them. I dont know who sets the odds but they stuffed up big time with the quarter final odds. The original market for teams in groups G and H to reach the quarters was 160% when 2 teams MUST qualify from the 8 teams within. A huge arbitrage situation that didnt last long once i took to it. My biggest problem is not finding overlays but getting barred by fixed odds betting agencies. I have hundreds of accounts in different names and still manage to get several a week closed down.

Cheers
ATC

There is money in Trifecta betting if one has the software to apply the maths.

It doesn't matter to me what percentage they take out or leave in (within reason, a variation of 10% is diddly).

Why?

Because I'm not betting against them, I'm betting against other punters.

All I need is a dividend which, when multiplied by the strike rate is greater than 100.

Simple.

To find that occurrence is the challenge, but exotics are the easiest way to make a profit - full stop. The very fact that the approximates are unknown (to most) is why it is so easy.

Just a small amount of research turns odds on horses in 5/1 shots (I've posted re this some time ago).

__________________
RaceCensus - powerful system testing software.
Now with over 431,000 Metropolitan, Provincial and Country races!
http://www.propun.com.au/horse_raci...ng_systems.html
*RaceCensus now updated to 28/02/2026
Video overview of RaceCensus here:
http://www.youtube.com/watch?v=W821YP_b0Pg

It is possible to beat a rake as high as 20% in my view.
Most of my betting is on first fours which have a rake as high as 25% (NSW TAB) and i have managed to show a healthy POT on these bet types over the past few years. I do have an advantage however of having software to generate the hundreds of combinations per race and a robust mathematical model to predict the first four placings. My biggest problem is finding large enough pools. Small pools are bad for my method of betting because dividends are "unders" when the outsiders come in. The beauty now is that exotic bets can be placed with corporate bookmakers to avoid smothering the dividend. The problem however is that they are quick to ban you when you start collecting big payouts. I have been a fulltime pro for about 4 years now and have consistently beaten a 20% rake. You just have to be a little smarter than the average person, have a decent bankroll and a good computer program. And a PhD in mathematics helps too!

Cheers
ATC

I would have to disagree on that one Assume...

The numbers are there and anyone can come up with the right formula/method or whatever to pick a win.

Biggest advantage is software to do it.

Here are some relevant calculations to illustrate important but poorly understood concepts of Exotics.

I'm using these formulae:

a*b*c/(1-a)/(1-a-b) = Trifecta Prob

a*b*c*d/(1-a)/(1-a-b)/(1-a-b-c) = F4 Prob

First assume a ZERO-rake situation:

And that a, b, c and d are all paying as 10% chances.

But you know that they are 10% better than that - i.e. 11% chances

You can (as KV did) approximate your expected return by multiplication.

133.1% Trifecta
146.4% F4

But if you use the formulae above you get:

138.1% Trifecta
158.7% F4


Now apply the offical rakes of 21% and 25% to get a post rake:

109.0% Trifecta
119.0% F4


Leading to the intriguing situation where the F4 profit is over double that for the Trifecta even though the F4 rake is higher.

If you change the parameters you can get even more startling results.

For KV's case where he claims to be 18% better:

138.7% Trifecta
168.4% F4

Suggesting even better returns for the F4.

The figures get even weirder when you consider say 20% chances, but anyone interested can readily calculate such things on a spreadsheet.

ATC,

While I appreciate your quality post there are some observations I need to make.

I deliberately stipulated 21% Trifectas on Australian gallops for good reason.

Your situation with F4s is straying into different areas.

As I have just tried to demonstrate there are many situations where Trifectas are inferior propositions to F4s so success in F4s does not imply success in Trifectas.

The F4 rake hike to 25% is only recent so your 4 year record may not be able to be sustained.

Flexi-betting was only introduced in 2001. That offered a massive advantage to F4 players capable of rapidly automating that. That edge has now been eroded.

F4s are both more blind and less blind than Trifectas allowing appropriate exploitation.

And of course the big difference is that F4s have a pleasant habit of jackpotting - meaning big pools and effective rakes approaching ZERO. The sub-zero ones seems to have stopped now for some curious reason.

So all in all your F4 testimony does not change my scepticism about Trifectas.

I also note you frequent the Betfair Forum. That suggests you are not averse to minimum rake gambits.

Writing an effective F4 program is not difficult and certainly not only the province of PhDs.

Betfair Programming is however a different matter.

Essentially there is a massive first-mover advantage in such capers. It wouldn't surprise me if those pioneers who led the way in Overs, then Trifectas, Supers, F4s and noticed the diminishing returns as laggers finally arrived, are now making waves in the still brave new stochastic world of Betfair.

Consider a Trifecta punter successful at finding Overs.

Then the effect of halving the rake on the % of action is:

10% => 26%
5% => 20%

Asssuming Overs are distributed according to the spreadsheet NORMSDIST function.

That is someone previously with 5% action can now get 20%. 4 times the action suggests at least 4 times the profit.

That is a phenomenal improvement which could set someone up for life.

There are well documented articles worldwide about the extents big players go to reduce the rake. Yet when an actual opportunity arises which doesn't involve consorting with Triads or whatever, there is an inexplicable reluctance.

This simply is wrong. The few who know the approximates also have lots of savant money with which to tsunami the pools.

It has been like that for a painfully long time, and I'm amazed that this should be disputed.

These big players didn't get where they are by being philanthropists and intentionally leaving even a scrap for anyone else.

There's no need to take anyone's word either way. It's easy to satisfy yourself with a test similar to what I recently undertook on the NSW TAB.

11:43 $3,024
11:44 $3,788
11:45 $5,278 = AST + actual jump
11:45+ $6,797

In this piddly NZ jumps race the pool nearly doubled from 60 seconds before actual jump until close.

So nearly half the money in that pool was very well educated. That is an overwhelming proportion which must crush Overs - particularly with a 21% rake.

JFC - Ain't you assuming that ALL of the money going into the pools in the last 60 seconds or so is from 'educated' punters? Clearly, that would not be the case.