Different results

Why would these two create different results? Should they not be the same?

Thanks
Tony


P123_composite_2.PNG


P123_composite.PNG

The actual formulas are not visible in the screenshots, only their labels. You might want to check them. You might also might want to check if Na’s are neutral or negative.

Assume they are the same, which they are. Should these not produce the same result?
I may misunderstand how composites change percentages.

Yes, these should produce the same result. I’m unable to reproduce your problem. Check in the text editor to make sure the weights are both exactly 2% and not, say, 2.03% or 1.97%.

-edit-

There are two reasons that they might give different results: One, they’re different; or two, there is a bit of a random component in the formula.

The first one is probably not the case, if you just copied and pasted. If you retyped it then you can email me links to the systems and I’ll take a look and verify. (In fact, I can do that regardless.) :slight_smile:

The random bit might happen because there is an explicitly random element to the formula (which is rare, but I’m going in blind to what your formula actually looks like) or we are forcing the computer to choose between two or more companies that are ranked equally.

Just looking at your screenshot, you’re looking at the current items, which do not exist for most financial companies. This means that I suspect strongly that you’re forcing a random choice here. But I’d be happy to take a look and verify it.

I will see if I can isolate this a bit before sending it over. But, who are you? You look like me, but you’re not me. That’s very weird.

Edit:

OK, you’re Paul.

If I isolate just these two elements then it appears to work fine. But if I use it in my very large RS, it produces tiny variances in the results.

I will email you the links.

incorrect post–deleted

Thanks Yuval. Anything I can do to make them consistent?

I’m still stumped by this. I think what I wrote above was incorrect. Now I’m no longer able to reproduce the performance difference, or the rank differences. Does your node formula include FRank or FOrder or Aggregate or any Eval statements? Are you using NAs neutral or NAs negative?

I am not using Frank or Forder, but I think I do have at least one eval

I sent you the links.