feat(kalshi): add ohlcv_hourly model

[los-xyz]

Summary

Adds kalshi.ohlcv_hourly — hourly OHLCV candles for Kalshi prediction markets. Depends on #9549 (market_details + market_trades).

What it does

• Sparse OHLCV aggregation from kalshi_market_trades with deterministic ordering (created_time, trade_id)
• Bounded hour spine per market via utils.hours, forward-fill via ASOF LEFT JOIN
• Resolution correction: close price set to 1.0/0.0 for settled markets past expiration
• is_forward_filled flag, VWAP null on no-trade hours
• Category extracted from product_metadata JSON
• Full history (2021+)
• Output schema aligned with polymarket_polygon.ohlcv_hourly for cross-venue consistency (BlackRock trial)

Kalshi vs Polymarket OHLCV differences

• Kalshi has no condition_id/token_outcome — uses ticker as market ID
• outcome is always 'Yes' (each Kalshi ticker is a single market; No price = 1 - Yes price)
• Ordered by created_time + trade_id (off-chain, no block_time/evt_index)
• Resolution via result field + expiration_time (vs Polymarket's market_outcome + market_end_time)

Test plan

• dbt compile passes
• CI builds and tests pass
• Unique on (hour, market_id, outcome)
• Spot-check OHLCV against raw trades for sample markets
• Verify forward-fill and resolution correction

🤖 Generated with Claude Code

[jeff-dude]

pushed some changes to get the *ohlcv_hourly table to match the polymarket one, assuming that's our goal for now to be in sync across the two main providers. a static table of 2025 data only to give sample, prior to full release later. also removed the data explorer display for now.

we are also blocked here until we finalize the other PR which builds the upstream kalshi tables. they also exist in this PR, but not with latest changes.

[cursor]

PR Summary

Medium Risk
Changes Polymarket’s OHLCV model materialization and incremental logic, which can affect rebuild behavior and historical candle outputs (especially around window boundaries and settlement correction).

Overview
Adds a new incremental model kalshi.ohlcv_hourly that builds hourly OHLCV candles from kalshi_market_trades, forward-fills no-trade hours via an hour spine + ASOF join, and overwrites all OHLC fields to terminal 1.0/0.0 after market expiration when the market is settled; includes market metadata fields, is_forward_filled, _updated_at, and uniqueness tests on (block_month, hour, market_id, outcome).

Refactors polymarket_polygon.ohlcv_hourly from a static table into an incremental merge partitioned by block_month, adds a prior-window sparse anchor to keep forward-fill correct across incremental boundaries, applies settlement correction to all OHLC fields (not just close), and updates schema docs/tests accordingly (including block_month and _updated_at).

^{Reviewed by Cursor Bugbot for commit 72b2402. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 3 potential issues.

Autofix Details

Bugbot Autofix prepared a fix for the issue found in the latest run.

• ✅ Fixed: Resolution correction makes close exceed high/low bounds
  • Applied resolution correction to open, high, and low in addition to close to maintain the OHLCV invariant for post-expiration settled markets.

Or push these changes by commenting:

@cursor push 4216e8e531

Preview (4216e8e531)

diff --git a/dbt_subprojects/daily_spellbook/models/_projects/kalshi/kalshi_ohlcv_hourly.sql b/dbt_subprojects/daily_spellbook/models/_projects/kalshi/kalshi_ohlcv_hourly.sql
--- a/dbt_subprojects/daily_spellbook/models/_projects/kalshi/kalshi_ohlcv_hourly.sql
+++ b/dbt_subprojects/daily_spellbook/models/_projects/kalshi/kalshi_ohlcv_hourly.sql
@@ -116,9 +116,6 @@
 		f.ticker,
 		f.market_name,
 		f.event_ticker,
-		f.open,
-		f.high,
-		f.low,
 		case
 			when m.expiration_time is not null
 				and f.hour > m.expiration_time
@@ -127,6 +124,42 @@
 				case
 					when m.result = 'yes' then 1.0
 					when m.result = 'no' then 0.0
+					else f.open
+				end
+			else f.open
+		end as open,
+		case
+			when m.expiration_time is not null
+				and f.hour > m.expiration_time
+				and m.result in ('yes', 'no')
+			then
+				case
+					when m.result = 'yes' then 1.0
+					when m.result = 'no' then 0.0
+					else f.high
+				end
+			else f.high
+		end as high,
+		case
+			when m.expiration_time is not null
+				and f.hour > m.expiration_time
+				and m.result in ('yes', 'no')
+			then
+				case
+					when m.result = 'yes' then 1.0
+					when m.result = 'no' then 0.0
+					else f.low
+				end
+			else f.low
+		end as low,
+		case
+			when m.expiration_time is not null
+				and f.hour > m.expiration_time
+				and m.result in ('yes', 'no')
+			then
+				case
+					when m.result = 'yes' then 1.0
+					when m.result = 'no' then 0.0
 					else f.close
 				end
 			else f.close

_{You can send follow-ups to the cloud agent here.}

[jeff-dude]

@cursor review

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

Autofix Details

Bugbot Autofix prepared a fix for the issue found in the latest run.

• ✅ Fixed: Resolution correction applied to all OHLCV fields inconsistently
  • Removed resolution correction from open, high, and low fields in Kalshi model to match Polymarket's behavior of only correcting the close price.

Or push these changes by commenting:

@cursor push ac418747a5

Preview (ac418747a5)

diff --git a/dbt_subprojects/daily_spellbook/models/_projects/kalshi/kalshi_ohlcv_hourly.sql b/dbt_subprojects/daily_spellbook/models/_projects/kalshi/kalshi_ohlcv_hourly.sql
--- a/dbt_subprojects/daily_spellbook/models/_projects/kalshi/kalshi_ohlcv_hourly.sql
+++ b/dbt_subprojects/daily_spellbook/models/_projects/kalshi/kalshi_ohlcv_hourly.sql
@@ -147,32 +147,14 @@
 		f.market_id,
 		f.outcome,
 		f.market_name,
+		f.open,
+		f.high,
+		f.low,
 		case
 			when m.expiration_time is not null
 				and f.hour > m.expiration_time
 				and m.result in ('yes', 'no')
 			then case when m.result = 'yes' then 1.0 else 0.0 end
-			else f.open
-		end as open,
-		case
-			when m.expiration_time is not null
-				and f.hour > m.expiration_time
-				and m.result in ('yes', 'no')
-			then case when m.result = 'yes' then 1.0 else 0.0 end
-			else f.high
-		end as high,
-		case
-			when m.expiration_time is not null
-				and f.hour > m.expiration_time
-				and m.result in ('yes', 'no')
-			then case when m.result = 'yes' then 1.0 else 0.0 end
-			else f.low
-		end as low,
-		case
-			when m.expiration_time is not null
-				and f.hour > m.expiration_time
-				and m.result in ('yes', 'no')
-			then case when m.result = 'yes' then 1.0 else 0.0 end
 			else f.close
 		end as close,
 		f.vwap,

_{You can send follow-ups to the cloud agent here.}

[jeff-dude]

@cursor review

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 72b2402. Configure here.}