Login Page - Create Account

Support Board


Date/Time: Thu, 03 Jul 2025 19:31:25 +0000



Post From: bid-ask volume on VBP

[2025-06-17 16:35:23]
joshtrader - Posts: 519
Thanks for your reply John.

I understand your argument 100%. And if the original BidAskDiff Bars were exactly scaled correctly with respect to each other, without rounding, it would be 100% correct.

However, if the maximum BidAsk difference value is something like 10000, then the bar displayed on the screen is definitely not 10000 times wider/longer than the smallest which has a value of 1, as the screen just isn't wide enough. So, the bars on the screen are being scaled down to match the appropriate VbP width. When this happens, the BidAskDiff bars which have a value of 1 (or a similarly small number) are having their values massively overstated by giving them a width on the screen of 1. You have to do it this way, to show that there is a nonzero value there (you could round down to 0 but you're opting to show something there if it's > 0).

So, in scaling the bars, you are scaling the value which has been significantly rounded up, rather than scaling the original value.

Quick practical example:

A study of the height of 50000 people is conducted. The largest "172cm" category has 10000 people. Only one person is in the "235cm" category. On a bar chart which can only show 1000 units on the y-axis, a normalized 172cm category will show a value of 1000, scaled down 10x to fit the chart's axis constraints. The 235cm category will still have a value of 1, even though it should be 0.1. It is rounded up to show the existence of the person, but it overstates the number of people with that height by a factor of 10x.

If you could now have the ability to have greater granularity on the y-axis and a scale of 10000 instead of 1000, then the 172cm category would show its actual value of 10000. For the 235cm category, would you use the original data and show a single value of 1 for the 235cm category? Yes, as this is the proper ratio and most accurate. You wouldn't just multiply the rounded values from the original scaled-down graph by a factor of 10, and have a value of 10 for the 235cm category, as this would significantly overstate the value. It would make it appear that there were 10 235cm tall people for every 10000 172cm people, rather than the actual ratio of 1 to 10000.
Date Time Of Last Edit: 2025-06-17 16:41:55