Is little Acorns genuine or a blag!?

Don't chase your losses, it doesn't work. You will eventually bust your bank.
Post Reply
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

... Excuse any bad maths I can't see my phone screen very well :)

£378 in 957 days, that's almost 40p a day, that should pay for the electric you've used to do it :D

But £378 from £23k turnover is about 1.6% so not stupidly small, just a shame it doesn't scale to a reasonable about because if you want to make £4 a day you'd stake 250s and move the price by approx the size of your edge.

Thx athos55, you've shown that a few hours due diligence can save months of wasted time and false dawns. Just a shame more don't do that rather than opening their wallet and crossing their fingers.
User avatar
Brovashift
Posts: 475
Joined: Tue May 18, 2021 12:35 am

Wow Athos55, that some quality Excel there, I've just been trying to do the Monte Carlo simulation and it's given me a headache lol. Thank you for taking the time to show me those numbers and saving a few more hairs on my head :lol:

I'll explain my thinking (and reply to you PM Shaun in a min 👍)....

Basically I just started with the avg lay the LA pdf states @1.76 (I used 1.75 just because Im a simplton lol), and when thinking about how the odds move (fluctuations in-play) I thought @2.0 being a crossover point and area of congestion it wouldnt take much contention for the odds to hit around 2.0. Possibly the data may say otherwise on e.g. short 5f races etc. but gotta start somewhere right?
So I thought an avg lay @1.75 backed @1.99 would be slightly over 20%, Wow £2 from a £10 L2B at very close odds, seems too good to be true, couldnt be could it? :D
I only looked back over past 3 days befair results data (manually... no sim), and quickly saw from the very few odds on favs per day, around 6 or 7, and not accounting for the selection precess, I could see that 20% would've inccured too many losses that would fall short of any wins. I know some people run automation and make pennies per trade but amount to decent returns over a large number of races day in day out. So where do I go from here, lets not be greedy and have a look at 10%. 10% on the very small sample size showed an almost 100% strike rate (again, on this very small sample size) when keeping the qualifying criteria >[email protected].

Pre Athos55's excellent work, I guess I would have had 2 choices to try and see if 10% was the right choice; one being to throw some more £1's at my new findings and be back on here in a week with a whole new bunch of questions about my 10% decision, or the second, and more sensible thing would be to take on your words Shaun and learn how to create a sim and run 1000 iterations in a second (once I worked out what data I needed). But to be honest, and you have to be honest with yourself in trading, I am not a numbers guy. Its like the guy says in the YouTube video shared in my other thread on modeling data "Making a Modeler", its not for everyone, some people are quantitatively or qualitatively slanted. This might sound mad but I don't trust the numbers :lol: their always historical. I know peoples profitability from data/auto is proof, but there's definitely a trust issue when it come to me and historical data. Its cheated on me before it'll cheat on me again! :lol:
I find Athos's work really interesting and impressive, but do I want to sit at my computer for hours creating something similar, or would I rather perfect my race reading skills and focus on trading in-play. I think it's the latter. :)
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

Historical data is only useful if you know how to use it. Otherwise it just leads to back fitting and trend following and it's worse than useless.

People think automation is some sort of simple cousin to manual trading, truth is it takes the same effort to learn. But with manual trading you get the fun of clicking the ladder on day 1, and with technical trading that first bet might be in 6 months time. Instant gratification results in a desk job, have patience and study hard and that could result in a passive income (and a better looking cv if you fail). Take your pick.
User avatar
wearthefoxhat
Posts: 3206
Joined: Sun Feb 18, 2018 9:55 am

Ok, so it still looks to be "profitable" in the last 3 years. (nice stats Athos5)

Can see what Shaun's driving at with learning data analytics/automation. My youngest lad just graduated with a 1:1 in his Computer Science Masters Degree. Quite sure he could steer me in the right direction, but he's too busy working on all some VR/Augmented reality stuff for an american company.

Still believe the L2B method has legs as the 2f/3f markets compress putting pressure on the favourite price. Reckon there's still a fair few that use the L.A in some form or other. Dallas put up a directional bot that could be incorporated in some way that when the price direction (Lay) is confirmed, a trade could be fired in and out a few times to take advantage of the moves. If this was an approach in some way, then the odds range is likely to go beyond 2.0, so still a low risk liability approach.
User avatar
Brovashift
Posts: 475
Joined: Tue May 18, 2021 12:35 am

wearthefoxhat wrote:
Mon Aug 15, 2022 4:27 pm
Ok, so it still looks to be "profitable" in the last 3 years. (nice stats Athos5)

Can see what Shaun's driving at with learning data analytics/automation. My youngest lad just graduated with a 1:1 in his Computer Science Masters Degree. Quite sure he could steer me in the right direction, but he's too busy working on all some VR/Augmented reality stuff for an american company.

Still believe the L2B method has legs as the 2f/3f markets compress putting pressure on the favourite price. Reckon there's still a fair few that use the L.A in some form or other. Dallas put up a directional bot that could be incorporated in some way that when the price direction (Lay) is confirmed, a trade could be fired in and out a few times to take advantage of the moves. If this was an approach in some way, then the odds range is likely to go beyond 2.0, so still a low risk liability approach.
Do you mean the directional scalping bot? I thought of something like this as a way to monitor price direction, a bit like a trailing stop, but a trailing lay entry order lol. Thing is though how would you decide how much wiggle room to give it....
If you gave it a 10 tick trail and it traded down to @1.6 and bounced back to enter you @1.7, have you just got 10 ticks worse price.... or if you set it to 5 tick trail and price started e.g. @1.85, traded down to @1.75, retraces to enter you @1.80, to then trade down to @1.58... you'd be thinking 'you bugger'. lol
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

L2B is a tempting proposition. All but one of them are guarenteed to payout and that makes it more scalable than picking one B2L contender. It's just about not having a hair trigger on the B's :) Well that's the theory but because it's so 'easy' I think most of us would have tried it... And I don't see many posts from private yatchs :D
User avatar
Realrocknrolla
Posts: 1903
Joined: Fri Jun 05, 2020 7:15 pm

Wifi shit here! 😉
C198E95C-ABE4-406D-BD4B-63F4588509E5.jpeg
You do not have the required permissions to view the files attached to this post.
User avatar
wearthefoxhat
Posts: 3206
Joined: Sun Feb 18, 2018 9:55 am

Realrocknrolla wrote:
Mon Aug 15, 2022 7:12 pm
Wifi shit here! 😉

C198E95C-ABE4-406D-BD4B-63F4588509E5.jpeg
You look good in white!...
User avatar
wearthefoxhat
Posts: 3206
Joined: Sun Feb 18, 2018 9:55 am

Brovashift wrote:
Mon Aug 15, 2022 6:38 pm

Do you mean the directional scalping bot? I thought of something like this as a way to monitor price direction, a bit like a trailing stop, but a trailing lay entry order lol. Thing is though how would you decide how much wiggle room to give it....
If you gave it a 10 tick trail and it traded down to @1.6 and bounced back to enter you @1.7, have you just got 10 ticks worse price.... or if you set it to 5 tick trail and price started e.g. @1.85, traded down to @1.75, retraces to enter you @1.80, to then trade down to @1.58... you'd be thinking 'you bugger'. lol

I can see what you mean about the trailing stop loss. There's always a trade that frustrate you that's for sure, it's just part of doing business.


The steamer bot could be converted into a Lay Bot.

viewtopic.php?f=53&t=11425

----------------

Same for The Scalp the mover bot.

viewtopic.php?f=53&t=11785

--------------

Here's the directional scalping bot.

viewtopic.php?f=53&t=12610

--------------
britguy
Posts: 22
Joined: Fri Dec 08, 2017 4:44 pm

Just to throw some confusion into this thread, I played with some numbers on my data to see how it compared to Athos's conclusion.
"Over this period returns for a £25 Lay bet would have circa £378 and numerically you would have had 420 bets that won and 501 that lost over this period."
Here is what I got

Code: Select all

 
   --------------------------------------------------------------
   |    Race Strategy Results for 01-Jan-2020 to 14-Aug-2022    |
   --------------------------------------------------------------

Bank Total             = £ 0.00
Total races count      =   24756
Lay candidates count   =     841
Actual lays count      =     841  (3.4%)
Winning bets count     =     376
Losing bets count      =     465
Skip on last out count =       0
So for my data only 3.4% of all races met the 3 price thresholds, and 841 lays were placed. The statistics gods were trolling me :shock: That gave me exactly a breakeven profit of 0.00. I assumed it was a bug, but more checking showed it was just a bizarre coincidence.

My results are quite a bit off compared to Athos, he had 921 qualifying races, compared to my 841. No idea where the discrepancy is, if Athos is reading this, how many races did your data cover ? how does it compare to my 24,756?

I ran a further test where it takes into account the last time out performance of the qualifying horse, skipping the lay if it was a beaten favourite.

Code: Select all

   --------------------------------------------------------------
   |    Race Strategy Results for 01-Jan-2020 to 14-Aug-2022    |
   --------------------------------------------------------------

Bank Total             = £587.50
Total races count      =   24756
Lay candidates count   =     841
Actual lays count      =     700  (2.8%)
Winning bets count     =     326
Losing bets count      =     374
Skip on last out count =     141

This skips 141 of the lay bets, giving a noticeable improvement of £587 profit.

Dont think I will be booking my Seychelles holiday anyway with this strategy !
Atho55
Posts: 637
Joined: Tue Oct 06, 2015 1:37 pm
Location: Home of Triumph Motorcycles

Hi Britguy, no idea how many records as I altered the data ranges after posting as seemed a dead end and looks like you came to the same conclusion. Might be the way I/we treat dead heats that`s the difference.

Pity you did not post first as it would have saved me doing it. Anyway, did you look at the alternate set of figures I posted by any chance. R1, 2.26. R2, 6.70. R3, 10.10 to see if they performed better?
britguy
Posts: 22
Joined: Fri Dec 08, 2017 4:44 pm

Hiya

Yes played around with some more numbers after I posted

For my original I have got a test for the price of "less than", I swapped this now to "less than or equal" for the 3 clauses, this gave results closer to yours

Code: Select all

Bank Total             = £440.75
Total races count      =   24756
Lay candidates count   =     945
Actual lays count      =     945  (3.8%)
Winning bets count     =     434
Losing bets count      =     511
Skip on last out count =       0
Then i added the clause back in to skip if beaten fave last time out. Gives a tidy improvement

Code: Select all

Bank Total             = £1060.75
Total races count      =   24756
Lay candidates count   =     945
Actual lays count      =     794  (3.2%)
Winning bets count     =     380
Losing bets count      =     414
Skip on last out count =     151
Then I tried your 3 threshold values with the last time out ignored again. Gives good improvement for my testing too.

Code: Select all

Bank Total             = £2120.00
Total races count      =   24756
Lay candidates count   =    2091
Actual lays count      =    2091  (8.4%)
Winning bets count     =    1054
Losing bets count      =    1037
Skip on last out count =       0
Adding the last time out clause back in with your thresholds. Small improvement again

Code: Select all

Bank Total             = £2484.25
Total races count      =   24756
Lay candidates count   =    2091
Actual lays count      =    1756  (7.1%)
Winning bets count     =     902
Losing bets count      =     854
Skip on last out count =     335
As a final test I ran the last criteria on the 2 years prior to these tests
Again it gives a profit of £1,826 on 1,545 lay bets.

So not the worst strategy ever, maybe not Seychelles holiday but a weekend in Bognor!

Not for me though, wouldnt fancy the huge liability on 1,545 x 25 quid lay bets for < £2,000 profit
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

britguy wrote:
Mon Aug 15, 2022 11:14 pm

Not for me though, wouldnt fancy the huge liability on 1,545 x 25 quid lay bets for < £2,000 profit
But it's not 1545 * 25 at any given time thankfully and 2 grand on 38k seems like a fair 5% on paper.

But things like this are far too selective, profit is stake * edge * opportunity and I can't see any of those 3 factors ever getting very big with this one. 1,500 bets in 2½yrs is crazy, I know I'm not the only one who has trades/bets on 2,3 or 400 dogs a day, and a few more than that on the aus dogs. You don't need 5% when you're doing 2,000 selections a week instead of 1,500 in two and a half years, and your bet stats accumulate fast too. Stake is limitted, edge is always marginal, so the way to make money is by increased opportunity ie scale, not filtering the markets down to 2 bets a day.

Btw I might be teaching granny to suck eggs here but when you're refining results (rather than testing the core premise) I suggest you do it on half your data, then see if that also worked on the other half. If you don't then refining over the entire set introduces backfitting and a greater likelihood it will actually worsen it in the future due to reversion rather than improving it. Big data sets make you feel they're representative but backfitting is a trap even on years of data. In fact finding an edge in big data isn't hard, but finding one in a randomised sample that also exists in the remainder is. An edge has to be evident/proven in your 'unseen' data too because that's what tomorrows racing is, using the whole lot just tells you about the past. How many people say they found and edge but when they did it live it started to fail? That's why.
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

...i'm on my high horse now :roll:

Even with data I also like to find some real world possible explaination for an edge, otherwise you end up with bollocks like lay all dogs where the collar colour matches the trap colour, on Wednesdays. But something like, horses from a respected yard being overbacked late on, does at least sound like a plausable phenomenon. Markets are people not just numbers so it's not a bad idea to think of a 'punteresque' behavior and look for it in the numbers, rather than finding something in the numbers and then not having an explaination for it. It's also a good solution to having a heap of data and not knowing where to start.


Back to Little Acorns.....lots of chat about if it worked so far but not much about why it should work in the future? Can anyone explain what punters/pros and the entire exchange ecosystem are doing that creates this simple predictable mispricing situation, because I can't think of one. "It just does" won't convince anyone "it just will" still be a thing tomorrow.
User avatar
wearthefoxhat
Posts: 3206
Joined: Sun Feb 18, 2018 9:55 am

ShaunWhite wrote:
Tue Aug 16, 2022 2:37 am

Back to Little Acorns.....lots of chat about if it worked so far but not much about why it should work in the future? Can anyone explain what punters/pros and the entire exchange ecosystem are doing that creates this simple predictable mispricing situation, because I can't think of one. "It just does" won't convince anyone "it just will" still be a thing tomorrow.

IMO, the attraction of the whole Little Acorns existence is/was a low liability lay method within the odds-on price range. The punters of the day liked the idea that there was also a recovery aspect to it that could "cope" with a losing run of 9 that "has never happened before."

Back in the day (2003) betfair trading wasn't much of a thing, but a few caught on and used if for that reason based on its' price movements.

I vaguely remember that backing the 2f/3f at Betfair odds in a qualifying race was a thing, but proved a false dawn.

Will it work in the future? That's where the ongoing pre-off testing with automation, will give an indication. There no reason why L2B trades can't work in some form if a bot is crunched together using some of Dallas's shared automation bots.

Not trying to convince anyone, just a POV.
Post Reply

Return to “Loss recovery systems and methods”