I'm actually going round the houses. I've developed an indicator and I want to try and measure how well it's indicating. Seems reasonable to want to quantify it.
The only thing I can think of with my limited knowledge is to compare it to another indicator of 'known origin' ie something like the McGinley.
Better still if I can find something to pass my indicator through and it returns a simple answer with a table of how accurate it was X ms ahead
I'm in way over my head sb, but I think I can find my way with the occasional nudge in the right direction.
Maths homework
-
- Posts: 3140
- Joined: Sun Jan 31, 2010 8:06 pm
Lol , I wouldn't have a clue, Shaun, my daughter was in and got a 1st in maths last year so I asked her.
This link seems to explain it better
http://fxcodebase.com/wiki/index.php/McGinley_Dynamic
This link seems to explain it better
http://fxcodebase.com/wiki/index.php/McGinley_Dynamic
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
It might be time to put an advert on the uni noticeboards, I need a statistician who'll work for minimum wage or a few rupees.
-
- Posts: 1074
- Joined: Fri Nov 20, 2015 9:38 am
You could just take a random sample of the results of your indicator vs what its measuring, look at the differences and obtain the average difference & variance on the results... Bigger the sample size the better the results... (just make sure the sample is random - easier said than done)ShaunWhite wrote: ↑Fri Jun 22, 2018 7:02 pmI'm actually going round the houses. I've developed an indicator and I want to try and measure how well it's indicating. Seems reasonable to want to quantify it.
The only thing I can think of with my limited knowledge is to compare it to another indicator of 'known origin' ie something like the McGinley.
Better still if I can find something to pass my indicator through and it returns a simple answer with a table of how accurate it was X ms ahead
I'm in way over my head sb, but I think I can find my way with the occasional nudge in the right direction.
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
It's a tricky one sionascaig, It suggests direction and strength of move rather than an absolute value and I also don't know how far ahead it's working, if it is at all. I'm snookered at the 'those two wiggly lines look really similar' stage. I might post some examples after the weekend, I need a break and the best ideas sometimes sneak up on you while your attention's elsewhere.
-
- Posts: 1074
- Joined: Fri Nov 20, 2015 9:38 am
You've got me interested now !
I'd try measuring the two elements you are interested in separately to see how well the indicator matches each (and this will allow you to refine formula based on results)...
Might be able to have a look when things quieten down (or fire it past someone with a brain the size of a planet)....
I'd try measuring the two elements you are interested in separately to see how well the indicator matches each (and this will allow you to refine formula based on results)...
Might be able to have a look when things quieten down (or fire it past someone with a brain the size of a planet)....
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
Even if the lines match theoretically it's no use unless it can be turned into a trade trigger. I think I need one giant loop with various trigger thresholds and methods, inside another loop looking at all the time offsets, for each race, for each everything else that might be factor
Time to pop to my local Quantum Computer World, or start learning about neural nets.
Time to pop to my local Quantum Computer World, or start learning about neural nets.
Would the Forecast function in Excel be any use to project a few seconds forward then compare the results alongside each other. I`ve tried it with Back odds using 11 seconds of actual odds then projected the next 3 forward which seemed to work.
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
Thanks for that. It's a function I hadn't considered. I'll stick it on my todo list for next week.