stevehopwoodforex.com
https://www.stevehopwoodforex.com/phpBB3/
Print view

Help needed from any mathematicians/statisticians please!
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=6019
Page 1 of 1
Author:  Jamesmsv [ Wed Mar 10, 2021 5:25 pm ]
Post subject:  Help needed from any mathematicians/statisticians please!

I am doing some analysis with a data feed and I have a problem that I'm struggling to solve using Google, probably because my mathematical vocabulary is not sufficient to find what I'm looking for. It's probably quite simple to solve once I explain:

I have one value (A) that normally sits in the 0.3000000 to 0.4500000 range.
The second value (B) is on a vastly different scale : anything from -78,000,000 to +200,000,000
My goal is to compare A to B on a daily basis as part of a trading plan. Prior day values do not feature in the trading plan (so a relative change %age is not what I'm looking for).

I need to find a way of calculating a relationship between A and B (I'm guessing a ratio) but due to the massive swings of 'B' I can't get any useful numbers that sit within a sensible range. I have tried just moving the decimal places but the massive swings of B seem to throw this simple method off too. It may be that I need to use different decimal shifts each day but I don't know if that action 'compromises' my data.

Is there a mathematical function I can perform to normalise both sets of data so that, on any given day, I can compare A to B and compute a figure that is consistent (say from -1.00 to 1.00)?

Any help much appreciated, thanks.
Author:  tomele [ Wed Mar 10, 2021 6:08 pm ]
Post subject:  Help needed from any mathematicians/statisticians please!

You can transform each value X in the range of Xmin..Xmax to a Y value in the range Ymin..Ymax by an affine transformation like this:

Y=(X-Xmin)*(Ymax-Ymin)/(Xmax-Xmin)+Ymin

Transform A and B to the range 0..1 and the multiply them. The result is again in a range of 0..1.
Author:  Jamesmsv [ Wed Mar 10, 2021 6:22 pm ]
Post subject:  Help needed from any mathematicians/statisticians please!

That sounds like exactly what I was after, thank you very much :clap:
All times are UTC Page 1 of 1