I almost don't want to mention that a few of those places have a turf and an AW course. But not being a follower of racing, I don't know if the turf courses are just used for jumps.
Newcastle AW
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
- ruthlessimon
- Posts: 2094
- Joined: Wed Mar 23, 2016 3:54 pm
Yeah...?? I've "publicised" so much, I (& you) don't even know the edge I've leaked I was waiting for the "is it real money" question !!
.. & if I've publicised a lot here, I could make a fortune selling shite menial trade advice, & mentaility ebooks (I've kept the proper secrets well under wraps) Don't tempt me Fox!!!
- ruthlessimon
- Posts: 2094
- Joined: Wed Mar 23, 2016 3:54 pm
ShaunWhite wrote: ↑Mon Oct 29, 2018 7:56 pmI almost don't want to mention that a few of those places have a turf and an AW course. But not being a follower of racing, I don't know if the turf courses are just used for jumps.
Ty for these insights though guysfoxwood wrote: ↑Mon Oct 29, 2018 7:29 pmWind direction is also supposedly relevant at Newcastle - forecasts for tracks at http://www.myweather2.com/Horse-Racing/ ... harts.aspx
Savage; what else you hiding eh!!
- ruthlessimon
- Posts: 2094
- Joined: Wed Mar 23, 2016 3:54 pm
Really frustrates me that post - because I was looking a strategy today - & the turf courses are the weakest aspect - which means I'm missin' somethin'.
It's not the fact it's Turf - there's something else
You do not have the required permissions to view the files attached to this post.
Sample size too small to be meaningful - although 36 out of 36 is worth a second look.
Idle thought - maybe on AW you are acting on the signature of bots for a low volume race - that should be exploitable !
However, as soon as you act on that and put money out for them you will probably change what they do so the signature will change - catch 22
- ruthlessimon
- Posts: 2094
- Joined: Wed Mar 23, 2016 3:54 pm
It's a difficult one.
There's not much I can do about the sample size, as the dataset range is May1st - Nov11th (albeit the refinements of the strategy mean I avoid the majority of races - but that's what allows it to be profitable - weakening the edge to increase the sample, will probably errode the p&l)
It's a preemptive strategy (ins't utilizing what I would call "order flow" - i.e. the combo of price, volume, speed). If anything, perhaps it suggests the opposite - that it's better to be "acting on signatures" in the turf races. i.e. If it's gonna be volatile & unpredictable, better to preempt than chase... but given that same argument I can't see why it's degraded on the turf - preempting should've had a similar ratio imo
So, really, your sample size is actually just over 8k and you're cherry picking around 5.5%.
Major likelihood of backfitting is my guess - probably triggered by a selective PL bias that showed itself to come mainly from AW on further analysis.
You need another 6 months dataset to prove it against or take the original data split it into two random halves and rework what the filters should be on one set then prove it on the second set. You shouldn't use the filters you currently have on a subset of the same data since the filters already have that "knowledge". Same principles as used when training AI.
No indication of staking level or single / multiple trades per race either - hopefully PL is based on single trade per race at minimum stakes.
If you want commentary on results you need to provide more background on methodology when posting - pretty graphs and tables without the background info are just a waste of everyone's time imho - apart from the occasional artistic merit
Major likelihood of backfitting is my guess - probably triggered by a selective PL bias that showed itself to come mainly from AW on further analysis.
You need another 6 months dataset to prove it against or take the original data split it into two random halves and rework what the filters should be on one set then prove it on the second set. You shouldn't use the filters you currently have on a subset of the same data since the filters already have that "knowledge". Same principles as used when training AI.
No indication of staking level or single / multiple trades per race either - hopefully PL is based on single trade per race at minimum stakes.
If you want commentary on results you need to provide more background on methodology when posting - pretty graphs and tables without the background info are just a waste of everyone's time imho - apart from the occasional artistic merit
- ruthlessimon
- Posts: 2094
- Joined: Wed Mar 23, 2016 3:54 pm
Really good points there fox
Tbh, in hindsight, I'm not quite sure what I was hoping to achieve. You're absolutely right, for help to be given - I need to be more specific.. yet I kinda don't want to.. A real nasty contradiction that!
Tbh, in hindsight, I'm not quite sure what I was hoping to achieve. You're absolutely right, for help to be given - I need to be more specific.. yet I kinda don't want to.. A real nasty contradiction that!