Australian racing

[Anbell]

I tested a new bot today, taking uncovered/naked bsp for a fixed payout of $20, given certain conditions. Active in every race in ANZ, had 500 bets in 90 markets.

It had POT after commission of 15%, which is higher than expected. I'll be very happy if it continues at half that return.

ETA: if anyone is interested I'll post tomorrow's results.

ETA2: I'm only offering this up as an exercise because a) there's been some recent chatter in the forum that you can't make money with prop bets and b) it'll show the variance in any model. If you have a losing few days, like GOAT's trading journey (for just one example) should you change your plan? This articular bot had a good day: should I continue or discontinue? What if I was down 15% on the first day/week? Do I have enough information to make decisions going forward?

I was going to turn this into a thread so that newbies could see how a bot comes from idea to fruition (and success or otherwise) but I'm too busy atm.

So in short, Saturday:
600 bets in 100 markets
Turnover: $1745
Profit after comm: $251
POT= 14.4%

Sunday:
260 bets in 46 markets
Turnover: $695
Profit after comm: $60
POT= 8.6%

Monday (with a couple of races to go in WA)
70 bets in 13 markets
Turnover: $188
Profit after comm: $60
POT= 32%

All of the above have been on the lay side, I had some technical issues with implementing the mirror image back side, but I solved that today.

[rik]

I tested a new bot today, taking uncovered/naked bsp for a fixed payout of $20, given certain conditions. Active in every race in ANZ, had 500 bets in 90 markets.

It had POT after commission of 15%, which is higher than expected. I'll be very happy if it continues at half that return.

ETA: if anyone is interested I'll post tomorrow's results.

ETA2: I'm only offering this up as an exercise because a) there's been some recent chatter in the forum that you can't make money with prop bets and b) it'll show the variance in any model. If you have a losing few days, like GOAT's trading journey (for just one example) should you change your plan? This articular bot had a good day: should I continue or discontinue? What if I was down 15% on the first day/week? Do I have enough information to make decisions going forward?

I was going to turn this into a thread so that newbies could see how a bot comes from idea to fruition (and success or otherwise) but I'm too busy atm.

So in short, Saturday:
600 bets in 100 markets
Turnover: $1745
Profit after comm: $251
POT= 14.4%

Sunday:
260 bets in 46 markets
Turnover: $695
Profit after comm: $60
POT= 8.6%

Monday (with a couple of races to go in WA)
70 bets in 13 markets
Turnover: $188
Profit after comm: $60
POT= 32%

All of the above have been on the lay side, I had some technical issues with implementing the mirror image back side, but I solved that today.

good results so far, why just 70 bets on Monday compared to 600 on Saturday? markets need to have good volume?

[Anbell]

I tested a new bot today, taking uncovered/naked bsp for a fixed payout of $20, given certain conditions. Active in every race in ANZ, had 500 bets in 90 markets.

It had POT after commission of 15%, which is higher than expected. I'll be very happy if it continues at half that return.

ETA: if anyone is interested I'll post tomorrow's results.

ETA2: I'm only offering this up as an exercise because a) there's been some recent chatter in the forum that you can't make money with prop bets and b) it'll show the variance in any model. If you have a losing few days, like GOAT's trading journey (for just one example) should you change your plan? This articular bot had a good day: should I continue or discontinue? What if I was down 15% on the first day/week? Do I have enough information to make decisions going forward?

I was going to turn this into a thread so that newbies could see how a bot comes from idea to fruition (and success or otherwise) but I'm too busy atm.

So in short, Saturday:
600 bets in 100 markets
Turnover: $1745
Profit after comm: $251
POT= 14.4%

Sunday:
260 bets in 46 markets
Turnover: $695
Profit after comm: $60
POT= 8.6%

Monday (with a couple of races to go in WA)
70 bets in 13 markets
Turnover: $188
Profit after comm: $60
POT= 32%

All of the above have been on the lay side, I had some technical issues with implementing the mirror image back side, but I solved that today.

good results so far, why just 70 bets on Monday compared to 600 on Saturday? markets need to have good volume?

Nah - just that there were only 3 meetings today vs 15 on Sat (aus horses)

[goat68]

I tested a new bot today, taking uncovered/naked bsp for a fixed payout of $20, given certain conditions. Active in every race in ANZ, had 500 bets in 90 markets.

It had POT after commission of 15%, which is higher than expected. I'll be very happy if it continues at half that return.

ETA: if anyone is interested I'll post tomorrow's results.

ETA2: I'm only offering this up as an exercise because a) there's been some recent chatter in the forum that you can't make money with prop bets and b) it'll show the variance in any model. If you have a losing few days, like GOAT's trading journey (for just one example) should you change your plan? This articular bot had a good day: should I continue or discontinue? What if I was down 15% on the first day/week? Do I have enough information to make decisions going forward?

I was going to turn this into a thread so that newbies could see how a bot comes from idea to fruition (and success or otherwise) but I'm too busy atm.

So in short, Saturday:
600 bets in 100 markets
Turnover: $1745
Profit after comm: $251
POT= 14.4%

Sunday:
260 bets in 46 markets
Turnover: $695
Profit after comm: $60
POT= 8.6%

Monday (with a couple of races to go in WA)
70 bets in 13 markets
Turnover: $188
Profit after comm: $60
POT= 32%

All of the above have been on the lay side, I had some technical issues with implementing the mirror image back side, but I solved that today.

i'd be interested in seeing what sort of variance you get

[Anbell]

i'd be interested in seeing what sort of variance you get

Me too!

I'm expecting quite a lot. And that's ok.

The story I'm trying to tell here, in part, is about how I (hopefully) turned a reliable high volume bot from 0.5% return to a 5% return bot. I can reduce stakes to cope with that, or just ride the variance. Or maybe it'll be another one for the scrap heap.

[dm1900]

i'd be interested in seeing what sort of variance you get

Me too!

I'm expecting quite a lot. And that's ok.

The story I'm trying to tell here, in part, is about how I (hopefully) turned a reliable high volume bot from 0.5% return to a 5% return bot. I can reduce stakes to cope with that, or just ride the variance. Or maybe it'll be another one for the scrap heap.

This is an interesting project. What's the strike rate Vs implied strike rate?

[Anbell]

i'd be interested in seeing what sort of variance you get

Me too!

I'm expecting quite a lot. And that's ok.

The story I'm trying to tell here, in part, is about how I (hopefully) turned a reliable high volume bot from 0.5% return to a 5% return bot. I can reduce stakes to cope with that, or just ride the variance. Or maybe it'll be another one for the scrap heap.

This is an interesting project. What's the strike rate Vs implied strike rate?

whats the formula for calculating those?

[goat68]

i'd be interested in seeing what sort of variance you get

Me too!

I'm expecting quite a lot. And that's ok.

The story I'm trying to tell here, in part, is about how I (hopefully) turned a reliable high volume bot from 0.5% return to a 5% return bot. I can reduce stakes to cope with that, or just ride the variance. Or maybe it'll be another one for the scrap heap.

i'm watching with interest...

[dm1900]

Me too!

I'm expecting quite a lot. And that's ok.

The story I'm trying to tell here, in part, is about how I (hopefully) turned a reliable high volume bot from 0.5% return to a 5% return bot. I can reduce stakes to cope with that, or just ride the variance. Or maybe it'll be another one for the scrap heap.

This is an interesting project. What's the strike rate Vs implied strike rate?

whats the formula for calculating those?

Strike rate: how many times were you right?
(Sum of stakes of winning bets)/(sum of all stakes)

Implied strike rate: how many times does market think you should have been right?
For each bet, define X as: (1/odds)*stake
Then calculate implied strike rate as (sum of X for all bets)/(sum of all stakes)


E.g. imagine these are my bets:

£2 @ 4.0, outcome = WIN
£2 @ 3.0, outcome = LOSS
£4 @ 2.0, outcome = WIN

Strike rate: (2+4)/(2+2+4) = 6/8 = 0.75

Implied strike rate = ((1/4 * 2)+(1/3 * 2)+(1/2 * 4 ))/(2+2+4)=0.395

---------------------
Alternative method that doesn't weight by stake size - also useful to calculate.

Actual SR = (number of winning bets)/(total number of bets)

Implied SR = for each bet, compute 1/odds as X. Then market SR is average of X of all bets.

[ShaunWhite]

given certain conditions

Same conditions I use

should I continue or discontinue?

I prefer a more non-binary approach, you look at the stats and decide how much you want to spend on buying the next day's raffle ticket. Stake according to confidence rather than just yes or no. Something that's previously performed really well might even be worth running at a small loss to see if it comes back. You need a reasonable R&D budget you expect to burn doing automation otherwise every wobble will throw you off or you'll miss good things that just happened to start in a downward trend. There's no hard and fast rule on this one as far as I can tell from asking around a lot, it really is just experience and personal pain threshold.

[Anbell]

This is an interesting project. What's the strike rate Vs implied strike rate?

whats the formula for calculating those?

Strike rate: how many times were you right?
(Sum of stakes of winning bets)/(sum of all stakes)

Implied strike rate: how many times does market think you should have been right?
For each bet, define X as: (1/odds)*stake
Then calculate implied strike rate as (sum of X for all bets)/(sum of all stakes)


E.g. imagine these are my bets:

£2 @ 4.0, outcome = WIN
£2 @ 3.0, outcome = LOSS
£4 @ 2.0, outcome = WIN

Strike rate: (2+4)/(2+2+4) = 6/8 = 0.75

Implied strike rate = ((1/4 * 2)+(1/3 * 2)+(1/2 * 4 ))/(2+2+4)=0.395

---------------------
Alternative method that doesn't weight by stake size - also useful to calculate.

Actual SR = (number of winning bets)/(total number of bets)

Implied SR = for each bet, compute 1/odds as X. Then market SR is average of X of all bets.

Thanks.
Total Stake=$2700
Stake of winners=2090

SR=78%
Implied SR=25%

Definitely over-over-performing over the first 1000 bets.

[Anbell]

given certain conditions

Same conditions I use

should I continue or discontinue?

I prefer a more non-binary approach, you look at the stats and decide how much you want to spend on buying the next day's raffle ticket. Stake according to confidence rather than just yes or no. Something that's previously performed really well might even be worth running at a small loss to see if it comes back. You need a reasonable R&D budget you expect to burn doing automation otherwise every wobble will throw you off or you'll miss good things that just happened to start in a downward trend. There's no hard and fast rule on this one as far as I can tell from asking around a lot, it really is just experience and personal pain threshold.

Cheers ShaunWhite. They were mostly rhetorical questions but the non-binary framing is a useful one. Unfortunately BA doesnt generally allow us to run many experiments simultaneously which is suboptimal

[dm1900]

whats the formula for calculating those?

Strike rate: how many times were you right?
(Sum of stakes of winning bets)/(sum of all stakes)

Implied strike rate: how many times does market think you should have been right?
For each bet, define X as: (1/odds)*stake
Then calculate implied strike rate as (sum of X for all bets)/(sum of all stakes)


E.g. imagine these are my bets:

£2 @ 4.0, outcome = WIN
£2 @ 3.0, outcome = LOSS
£4 @ 2.0, outcome = WIN

Strike rate: (2+4)/(2+2+4) = 6/8 = 0.75

Implied strike rate = ((1/4 * 2)+(1/3 * 2)+(1/2 * 4 ))/(2+2+4)=0.395

---------------------
Alternative method that doesn't weight by stake size - also useful to calculate.

Actual SR = (number of winning bets)/(total number of bets)

Implied SR = for each bet, compute 1/odds as X. Then market SR is average of X of all bets.

Thanks.
Total Stake=$2700
Stake of winners=2090

SR=78%
Implied SR=25%

Definitely over-over-performing over the first 1000 bets.

Interesting. So you're essentially betting at odds of ~4 which would imply you win 25% of the time, but you're actually winning 78% of the time

[Anbell]

Strike rate: how many times were you right?
(Sum of stakes of winning bets)/(sum of all stakes)

Implied strike rate: how many times does market think you should have been right?
For each bet, define X as: (1/odds)*stake
Then calculate implied strike rate as (sum of X for all bets)/(sum of all stakes)


E.g. imagine these are my bets:

£2 @ 4.0, outcome = WIN
£2 @ 3.0, outcome = LOSS
£4 @ 2.0, outcome = WIN

Strike rate: (2+4)/(2+2+4) = 6/8 = 0.75

Implied strike rate = ((1/4 * 2)+(1/3 * 2)+(1/2 * 4 ))/(2+2+4)=0.395

---------------------
Alternative method that doesn't weight by stake size - also useful to calculate.

Actual SR = (number of winning bets)/(total number of bets)

Implied SR = for each bet, compute 1/odds as X. Then market SR is average of X of all bets.

Thanks.
Total Stake=$2700
Stake of winners=2090

SR=78%
Implied SR=25%

Definitely over-over-performing over the first 1000 bets.

Interesting. So you're essentially betting at odds of ~4 which would imply you win 25% of the time, but you're actually winning 78% of the time

According to my interpretation of your description! ROI is only 15% (pre-comm) though, so I may have misunderstood something along the way.

[Derek27]

Didn't realise they're so many group races at Randwick.