Re-introduce small assets filter

[lajarre]

On Deepdao, Bankless DAO has 47 asset tokens listed.

On Whip, it has a bit less, probably because of the whitelist filtering mechanism. Nevertheless, more than 15 of them have < 0.5% allocation, but we're still doing all the calculations and data fetching for them.

I would suggest to re-introducing a small assets filter, so as to prevent:

• Loading all these pricing series and building the related statistics (returns, volatility).
• Having a huge covariance matrix and related calculations for risk contributions.

As a result of such a filter, these assets wouldn't show up in the front-end. But if needed, we could keep these assets in a "shallow assets" collection, that would not be included in calculations but still shown in the summary. However, I am not sure this brings any value, I guess we'd prefer just showing the quick disclaimer explaining how assets are filtered.

So I'm in favor of just re-introducing the filter as it was before for an immediate performance gain.

@acemasterjb what do you think?

[acemasterjb]

The performance gains are tempting... however the small asset filter was removed to account for cases when a DAO has a large holding of (a) token(s) that results in other significant tokens to be filtered out.

E.g. OWL and COW filtering out FRAX and stETH among others on this Gnosis DAO

There might be other cases where a DAO might have ETH but it doesn't show and it doesn't factor into the calculations because it has a holdings that < 5%. I'm not saying that as a reason for not reintroducing the small asset filter, just wanted to bring up that possibility. Having such a small holding of ETH (as well as other "top" tokens) might very well be insignificant.

What do you think about only applying the small asset filter when calculating the statistics, including the risk contributions? That way the users would not think that Whip is "broken", but that those tokens have such a small holding that they're effectively not significant when calculating statistics.