To change the up-down bar width, select one of the related line chart series and format it. On the Options tab is a setting for gap width, which works the same as in a bar chart.

You have to fake the labels. Excel has no built in way to assign labels from an arbitrary range to a set of points. You can use Rob Bovey’s Chart labeler add-in (http://appspro.com/) to add the column D labels to any series in the chart. If you want fine control over the label positioning, you may want to add another series to the chart. Use an XY type, reassigned to the same axes as the line chart series, or a line chart type, where its plot order is not first or last (which affect the endpoints of the up-down bars). Hide the lines and markers for this series, and use the Chart Labeler to add your labels to this series.

This was a great explanation. I had one problem and one question.

Problem = I can’t get the up-down bars to change width. I use Excel 2003 SP3 at work. Can you explain how to make the change?

Question = Is there a way to get the changes (column D) to show in the report as data labels. I tried the ways I know and all I can get are the starting and ending numbers as labels.

Love the site. Thanks again.

Thanks, yes, this is an issue. If you format the up-down bars with a fill color and no border, they can be resolved down to a smaller size. If you use Excel 2003 and earlier, you can manually change the gap width of the bars to help with visibility (Excel 2007 requires some VBA).

I picked up the deviation values straight away for data label column for some deviation chart, ignoring your formulas. When I found that the data labels were not getting positioned correctly in the graph, I realised that even there, you had ensured a discipline whereby labels are displayed only above the X axis & on top of the stack.

I used all possible permutations & combinations to understand the reasoning behind your colour scheme & formulas & ended up with this:

Blank + & – columns ensure only the row deviation given by red or green is visible
Blank + is to make the patch above X axis from 0 upto cum after effect value invisible.
This implies that Blank + would have values only if both before & after effects are positive
Blank + would take the least value of before & after effects
Blank – is to make the patch below X axis from 0 down to cum after effect value invisible.
Blank – would have values only if both before & after effects are negative
Blank – would take the least ABS value of before & after effects

Red – & + to show negative deviations
Sum of ABS Value of Red + & Red + = Row deviation
Red – will have values only if the after effect is also negative
Red – will pick up either after effect or deviation, whichever is smaller number in ABS values
Red + will have values only if before effect is +
Red + will pick up before effect or deviation, whichever is smaller number in absolute values

Green – & Green + to show positive deviations
Sum of ABS Value of Green + & Green – = Row deviation
Green – will have values only if before effect is negative
Green – will pick up either before effect or deviation, which ever is smaller ABS
Green + will have values only if after effect is +
Green + will pick up after effect or deviation which ever is smaller ABS

Whatever, thankss for pointing out the typo in the example. The formulas were correct, but the typo in the table led to a misinterpretation of the protocol.

Gaaa! I see the error (I didn’t notice when I tested the same numbers yesterday). I’ve corrected these values. The formulas were always correct

Your reply to

“If deviation is +, Previous row cum is -, current row cum is +, Green + is current cum and green – is previous cum”
is
“No, because the green negative (which is a difference) cannot exceed the gain (B3). ”

Sir, is it not true that, if the previous row cum is picked, it will never exceed the gain as in any situation, the gain has to be more than the previous row –ve cum, to make the current row cum +ve?

I tried couple of options with + & – If the value in C9 is read as -4.87 instead of -9.87, then everything works fine whether it is your ‘=IF(AND(C20),MAX(-B3,C2),0)’ formula or as bland as mine wherein I pick earlier cum value

My son goes around telling everyone who bothers to listen, that I’ll end up in the next couple of days, plotting a waterfall chart with start point as his previous year performance in school & end point as current year’s

Thanks a Million for all the guidance,
Savithri

The previous after effect is -ve, current deviation ( price effect 6.35) is +, making the after effect also + 1.48.

Hence green – is picking up Max of -6.35 and – 4.87 which is -4.87 ( previous row after effect as I understand) .

This along with the value in green + is painting the entire stretch from a negative cum to positive cum ( which is the current favourable deviation) as green in the display

Savithri

“If deviation is +, Previous row cum is +, current row cum is + , Green + is current cum and green – is zero”

Current cum = previous cum + gain. Green + is gain (the difference between previous cum and current cum).

“If deviation is +, Previous row cum is -, current row cum is +, Green + is current cum and green – is previous cum”

No, because the green negative (which is a difference) cannot exceed the gain (B3).

“If deviation is +, Previous row cum is -, current row cum is – , Green + is zero and green – is previous cum”

No, Green – is current cum minus previous cum.