feat: Use benchmarks API for historical data #3138
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
5 Skipped Deployments
|
fhqvst
force-pushed
the
fhqvst/ui-276-use-benchmarks-api-for-historical-data
branch
from
a008adb to
a697c35
Compare
cprussin
left a comment
cprussin
left a comment
There was a problem hiding this comment.
The change is overall good! But I think there's a few more things before we ship it:
- I think the
/historical-pricesendpoint is now dead code, can you check and clean it up if so? - @ali-behjati can you tell us how hard it would be to add CI to the benchmarks endpoint? Is it feasible? I really hate losing CI from the historical charts, plus it does seem reasonable that we might want to have CI in the benchmarks apis.
| // Transform OHLC data to our format using close prices | ||
| const data = benchmarkData.t.map((timestamp, i) => ({ | ||
| time: timestamp as UTCTimestamp, | ||
| price: benchmarkData.c[i], // Use close price |
There was a problem hiding this comment.
I don't love this -- I think we will need to put more thought in here. If we aren't showing OHLC candles, I think the chart will be misleading. Plus, we should have full resolution (individual points) when zoomed all the way in, so OHLC doesn't really make sense at that resolution. Let's grab some time to chat through this
|
@cprussin it shouldn't be so hard to add it but takes a bit of time. we need to change the materialized views to add it and need to expose a new endpoint that returns it in the benchmarks. |
fhqvst
force-pushed
the
fhqvst/ui-276-use-benchmarks-api-for-historical-data
branch
from
a697c35 to
d52a2b4
Compare
Summary
With this PR we now consume historical price data from the Benchmarks API instead of using our own hand-rolled Clickhouse query.
One downside is that we don't have confidence intervals in the Benchmarks API...
Rationale
Our clients' users have complained that they're seeing inconsistent data in insights hub vs. the data on our clients' pages. With this change we now make sure that the data our clients are consuming is the same one we're using on insights hub, meaning there shouldn't be any inconsistencies.
How has this been tested?