Minimum marktkapitalisatie "off by one" error

[roytomeij]

Voor de minimum marktkapitalisatie wordt gebruikt gemaakt van marktkapitalisatie > waarde terwijl dit marktkapitalisatie >= waarde zou moeten zijn. Nu worden fondsen met een marktkapitalisatie van 93% alleen getoond wanneer de ingevoerde waarde lager is dan 93.

[jakedane]

It does use >=. See at https://github.com/nicwortel/indexfondsenvergelijken.nl/blob/master/src/index.ts#L57 (not the =>, the >= after):

    portfolios = portfolios.filter((portfolio: Portfolio) => portfolio.getMarketCapPercentage().getPercentage() >= minimumMarketCap);

I think what you describe is caused by rounding. For example the 3 NT funds have a market cap of 92.8681203778817%. See at https://github.com/nicwortel/indexfondsenvergelijken.nl/blob/master/data/indices.json#L121. That is rounded up to 93% for display but it compares that 92.8681203778817% with the market cap filter, so you have to set it to 92% to match.

[nicwortel]

Good catch @jakedane, that seems to be the case indeed.

However I can understand the confusion as most people will not look into the data files, so I'm open for suggestions on how to make this more intuitive. For that reason I will leave this issue open.

[jakedane]

Given that the "Minimum marktkapitalisatie" field on the site allows only integer numbers, maybe change the filter code from:

portfolio.getMarketCapPercentage().getPercentage() >= minimumMarketCap

to:

portfolio.getMarketCapPercentage().getPercentage() >= minimumMarketCap - 0.5

Basically a "Minimum marktkapitalisatie" of 93 would match portfolios with a market cap that rounds to at least 93. For example the 3 NT funds at market cap 92.8681203778817, which the site shows rounded as 93, would now match if the filter is set to 93.

[roytomeij]

Sorry, only seeing this one now. Good catch @jakedane! Lazy me hadn't looked at the source code, so I'm glad others are better open-source citizens than me :)

I'd round up the funds' market cap to the nearest integer instead of subtracting 0.5 from the input integer. That seems like a more failsafe way to handle these edge cases.