Postgres index not used when using BETWEEN inside join - postgresql

Question

I have the following table in my database:
business_db_dev=# \d schedules2
Table "public.schedules2"
Column | Type | Collation | Nullable | Default
-------------+--------------------------------+-----------+----------+----------------------------------------
id | bigint | | not null | nextval('schedules2_id_seq'::regclass)
monday | boolean | | not null |
tuesday | boolean | | not null |
wednesday | boolean | | not null |
thursday | boolean | | not null |
friday | boolean | | not null |
saturday | boolean | | not null |
sunday | boolean | | not null |
start1 | time(0) without time zone | | |
end1 | time(0) without time zone | | |
start2 | time(0) without time zone | | |
end2 | time(0) without time zone | | |
user_id | bigint | | not null |
inserted_at | timestamp(0) without time zone | | not null |
updated_at | timestamp(0) without time zone | | not null |
Indexes:
"schedules2_pkey" PRIMARY KEY, btree (id)
"schedules2_start1_end1_DESC_NULLS_LAST_index" btree (start1, end1 DESC NULLS LAST)
"schedules2_start2_end2_DESC_NULLS_LAST_index" btree (start2, end2 DESC NULLS LAST)
"schedules2_user_id_index" UNIQUE, btree (user_id)
Foreign-key constraints:
"schedules2_user_id_fkey" FOREIGN KEY (user_id) REFERENCES users(id) ON DELETE CASCADE
I also have other tables that I use to do a join with that one (users and strategies) which I will not post here for brevity, but if it is needed you can just ask and I will update the question with their structures too.
Giving this table, I'm trying to do the following query
select u.token
from strategies as st
inner join users as u on (st.user_id = u.id)
inner join schedules2 as sc on (st.user_id = sc.user_id)
where st.exchange = 'binance'
and st.market_pair = 'btc_usdt'
and st.timeframe = 'five_minutes'
and st.name = 'stoch_oscillator'
and st.inputs = '{5,3,3,80,20}'
and (sc.start1 is null or ('13:00:01'::time between sc.start1 and sc.end1) or ('13:00:01'::time between sc.start2 and sc.end2));
Running this query with explain analyze I got this result:
Nested Loop (cost=1.27..215.56 rows=16 width=6) (actual time=0.076..6.050 rows=942 loops=1)
Join Filter: (st.user_id = u.id)
-> Nested Loop (cost=0.98..197.89 rows=17 width=16) (actual time=0.070..3.650 rows=942 loops=1)
-> Index Only Scan using unique_strategy_and_user_id on strategies st (cost=0.69..7.29 rows=80 width=8) (actual time=0.056..1.083 rows=1000 loops=1)
Index Cond: ((exchange = 'binance'::text) AND (market_pair = 'btc_usdt'::text) AND (timeframe = 'five_minutes'::text) AND (name = 'stoch_oscillator'::text) AND (inputs = '{5,3,3,80,20}'::character varying[]))
Heap Fetches: 0
-> Index Scan using schedules2_user_id_index on schedules2 sc (cost=0.29..2.38 rows=1 width=8) (actual time=0.002..0.002 rows=1 loops=1000)
Index Cond: (user_id = st.user_id)
Filter: ((start1 IS NULL) OR (('13:00:01'::time without time zone >= start1) AND ('13:00:01'::time without time zone <= end1)) OR (('13:00:01'::time without time zone >= start2) AND ('13:00:01'::time without time zone <= end2)))
Rows Removed by Filter: 0
-> Index Scan using users_pkey on users u (cost=0.29..1.03 rows=1 width=14) (actual time=0.002..0.002 rows=1 loops=942)
Index Cond: (id = sc.user_id)
Planning Time: 0.834 ms
Execution Time: 6.130 ms
The important part is this one:
Index Scan using schedules2_user_id_index on schedules2 sc (cost=0.29..2.38 rows=1 width=8) (actual time=0.002..0.002 rows=1 loops=1000)
Index Cond: (user_id = st.user_id)
Filter: ((start1 IS NULL) OR (('13:00:01'::time without time zone >= start1) AND ('13:00:01'::time without time zone <= end1)) OR (('13:00:01'::time without time zone >= start2) AND ('13:00:01'::time without time zone <= end2)))
As you can see, Postgres is using Filter to check the values start1, end1, start2 and end2, but I was expecting that Postgres would use the two indexes I created for this exact condition:
"schedules2_start1_end1_DESC_NULLS_LAST_index" btree (start1, end1 DESC NULLS LAST)
"schedules2_start2_end2_DESC_NULLS_LAST_index" btree (start2, end2 DESC NULLS LAST)
Removing the join with schedules2 table and its condition basically halves the query time.
So, my question is, why is Postgres using Filter instead of my indexes, and how can I change the query or the indexes itself to optimize this query?
Edit: Note that the values used in the query (like '13:00:01'::time) are just examples, in my system this can be anything.

Answer 1

Index Cond: (user_id = st.user_id)
Filter: ....
Rows Removed by Filter: 0
The index it is already using is already perfect. It found no extra rows which had to be removed by the filter (or at least, so few that they rounded to zero). How could that be improved upon by using more indexes?