nimo23

But @TheCookieLab If you do not agree with this PR, please create a new PR with the following points: - remove the `base` from `ReturnCriterion` and all other Criteria where...

>Another potential problem with this change is negative returns. In a scenario where one wants to calculate aggregate returns of multiple investments, say, +10%, -2%, and +5% the current representation...

>This approach is also using the endTime (ZonedDateTime) as key instead of the index. This would avoid issues with limited bar series where the indices of the underlying array may...

>It seems that my IDE has swapped a lot of imports in files that are not related to changes of this PR. I will try to revert this soon -.-...

What about the old caching implementations. If you include the old caching logic into another `IndicatorValueCache`-implementation within caching/`NativeIndicatorValueCache` caching/`NativeRecursiveIndicatorValueCache` _(with changes as needed, e.g. `endTime` instead of `index`)_ We need...

>Let’s not normalize raising concerns on blank lines in between imports, comment style, etc in code reviews, that’s frankly anal retentive and causes more harm than good. There’s a reason...

>I aggree, especially in PRs like this where the focus is on the general design and solving different challenges to reach the objectives. @team172011 I don't agree. It does **not**...

>..one goal of this PR is to eliminate alle the difficulties we have with the **indexed based caching** and **the sync between** the circular array in CachedIndicators and in the...

>I just used the endTime of each bar as primary key, it is mandatory and it makes no sense to have two bars for the same endTime -> primary key....

@team172011 very good, you understood exactly what I meant. And the solution you provided is very promising 👍