Problem with Percentiles calculations

[statspy-ml]

I am running a query to classify a price in to 4 quartiles. I am not exp'ed in Presto so wondering why this is happening:
you can see the price $86.36 classified as 1st and 2nd quartils but all the 4 quartis has the same value.

my query is below:

with cte as ( select
                     purchase_date, created_at, odin , seller_id, gmv
                            from table
                            where date(purchase_date) >= (cast(current_date as date) - interval '12' day)
                            group by purchase_date,created_at, odin, seller_id, gmv),

  quartiles as (select

                     odin,
                     approx_percentile(gmv, 0.25) AS "1_quartile",
                     approx_percentile(gmv, 0.5)  AS "2_quartile",
                     approx_percentile(gmv, 0.75) AS "3_quartile",
                     approx_percentile(gmv, 1)    AS "4_quartile"

  from cte
  group by odin

  )

select cte.purchase_date, cte.created_at, cte.seller_id, cte.gmv,
       ntile(4) over (partition by cte.odin order by gmv asc) quartile, quartiles.*
from cte
inner join quartiles on cte.odin = quartiles.odin

anyone to help me understand what i am doing wrong here?
Thanks