diff --git a/src/sql/orderbook/barn_batch_rewards.sql b/src/sql/orderbook/barn_batch_rewards.sql
index b93cdae2..53664f3d 100644
--- a/src/sql/orderbook/barn_batch_rewards.sql
+++ b/src/sql/orderbook/barn_batch_rewards.sql
@@ -361,6 +361,11 @@ reward_data AS (
         end as fee,
         -- scores
         winning_score,
+        case
+            when block_number is not null
+            and block_number <= block_deadline then winning_score
+            else 0
+        end as observed_score,
         reference_score,
         -- auction_participation
         participating_solvers,
@@ -390,12 +395,12 @@ reward_per_auction as (
         surplus,
         protocol_fee, -- the protocol fee
         fee - network_fee_correction as network_fee, -- the network fee
-        surplus + protocol_fee - reference_score as uncapped_payment,
+        observed_score - reference_score as uncapped_payment,
         -- Capped Reward = CLAMP_[-E, E + exec_cost](uncapped_reward_eth)
         LEAST(
             GREATEST(
                 - {{EPSILON_LOWER}},
-                surplus + protocol_fee - reference_score
+                observed_score - reference_score
             ),
             {{EPSILON_UPPER}}
         ) as capped_payment,
diff --git a/src/sql/orderbook/prod_batch_rewards.sql b/src/sql/orderbook/prod_batch_rewards.sql
index b93cdae2..53664f3d 100644
--- a/src/sql/orderbook/prod_batch_rewards.sql
+++ b/src/sql/orderbook/prod_batch_rewards.sql
@@ -361,6 +361,11 @@ reward_data AS (
         end as fee,
         -- scores
         winning_score,
+        case
+            when block_number is not null
+            and block_number <= block_deadline then winning_score
+            else 0
+        end as observed_score,
         reference_score,
         -- auction_participation
         participating_solvers,
@@ -390,12 +395,12 @@ reward_per_auction as (
         surplus,
         protocol_fee, -- the protocol fee
         fee - network_fee_correction as network_fee, -- the network fee
-        surplus + protocol_fee - reference_score as uncapped_payment,
+        observed_score - reference_score as uncapped_payment,
         -- Capped Reward = CLAMP_[-E, E + exec_cost](uncapped_reward_eth)
         LEAST(
             GREATEST(
                 - {{EPSILON_LOWER}},
-                surplus + protocol_fee - reference_score
+                observed_score - reference_score
             ),
             {{EPSILON_UPPER}}
         ) as capped_payment,
diff --git a/tests/integration/test_fetch_orderbook.py b/tests/integration/test_fetch_orderbook.py
index 3acf8ec7..6475dd50 100644
--- a/tests/integration/test_fetch_orderbook.py
+++ b/tests/integration/test_fetch_orderbook.py
@@ -252,22 +252,22 @@ def test_get_batch_rewards(self):
                     "8912931423922369",
                 ],
                 "uncapped_payment_eth": [
-                    "405658669881356",
+                    "405658366132340",
                     "292618683544035",
                     "-13360904627424245",
                     "-12938581202699558",
                     "2823075137791897",
                     "9345865552057182",
-                    "1607086130097916",
+                    "1607086130097915",
                 ],
                 "capped_payment": [
-                    "405658669881356",
+                    "405658366132340",
                     "292618683544035",
                     "-10000000000000000",
                     "-10000000000000000",
                     "2823075137791897",
                     "9345865552057182",
-                    "1607086130097916",
+                    "1607086130097915",
                 ],
                 "winning_score": [
                     "11327955072945657",