I am trying to write a sql query to calculate RSI(Relative strength index) in Postgresql & implement in grafana.

I have just 3 columns in table -

time_stamp bigint, 
metric_name text,
value.max int

the RSI is about value.max for plotting in grafana as a timeseries query

RSI Reference: https://school.stockcharts.com/doku.php?id=technical_indicators:relative_strength_index_rsi

the query for RSI I used is as below:

SELECT
  $__unixEpochGroup(timestamp/1000, '1m') AS "time",
  value.max - lag(value.max) OVER (ORDER BY $__unixEpochGroup(timestamp/1000, '1m')) AS "diff",
  case when diff > 0 then diff else 0 end as "Gain",
  case when diff < 0 then diff else 0 end as "Loss",
  avg(Gain) OVER (ORDER BY $__unixEpochGroup(timestamp/1000, '1m') ROWS 20 PRECEDING) as "avg_gain",
  avg(Loss) OVER (ORDER BY $__unixEpochGroup(timestamp/1000, '1m') ROWS 20 PRECEDING) AS "avg_loss"
FROM spectrum_schema.my_table
WHERE
  $__unixEpochFilter(time) and
  metric_name = 'CPUUtilization' 
ORDER BY 1

this gives ERROR**

window function calls may not be nested

Can anyone help with building RSI sql query?

the normal query for timeseries data plotting is working as below:

SELECT
  $__unixEpochGroup(timestamp/1000, '1m') AS "time",
  value.max AS "value.max"
FROM spectrum_schema.my_table
WHERE
  $__unixEpochFilter(time) and
  metric_name = 'NetworkThroughput' and 
ORDER BY 1

Can this above working query be tuned as well?