Excel add-in: Bloomberg-grade overhaul of the bond pricing surface

.github/workflows/ci.yml

@@ -104,8 +104,9 @@ jobs:
       - name: Cache cargo registry
         uses: Swatinem/rust-cache@v2
 
+      # `excel` profile = panic=unwind; see [profile.excel] in Cargo.toml.
       - name: Build convex-ffi
-        run: cargo build --release -p convex-ffi
+        run: cargo build --profile excel -p convex-ffi
 
       - name: Setup .NET
         uses: actions/setup-dotnet@v4
@@ -115,8 +116,11 @@ jobs:
       - name: Build Excel Add-in
         run: dotnet build excel/Convex.Excel/Convex.Excel.csproj --configuration Release
 
+      - name: Test Excel Add-in (headless)
+        run: dotnet test excel/Convex.Excel.Tests/Convex.Excel.Tests.csproj --configuration Release
+
       - name: Copy native DLL
-        run: copy target\release\convex_ffi.dll excel\Convex.Excel\bin\Release\net472\
+        run: copy target\excel\convex_ffi.dll excel\Convex.Excel\bin\Release\net472\
 
       - name: Create Add-in Package
         shell: pwsh

.github/workflows/java-release.yml

@@ -43,13 +43,14 @@ jobs:
           sudo apt-get update && sudo apt-get install -y gcc-aarch64-linux-gnu
           echo "CARGO_TARGET_AARCH64_UNKNOWN_LINUX_GNU_LINKER=aarch64-linux-gnu-gcc" >> "$GITHUB_ENV"
 
+      # `excel` profile = panic=unwind; see [profile.excel] in Cargo.toml.
       - name: Build cdylib
-        run: cargo build -p convex-ffi --release --target ${{ matrix.target }}
+        run: cargo build -p convex-ffi --profile excel --target ${{ matrix.target }}
 
       - uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4
         with:
           name: ${{ matrix.classifier }}
-          path: target/${{ matrix.target }}/release/${{ matrix.lib }}
+          path: target/${{ matrix.target }}/excel/${{ matrix.lib }}
           if-no-files-found: error
 
   # Stage every cdylib into the jar resources and publish to Maven Central.

.github/workflows/java.yml

@@ -48,14 +48,15 @@ jobs:
       - name: Cache cargo
         uses: Swatinem/rust-cache@v2
 
+      # `excel` profile = panic=unwind; see [profile.excel] in Cargo.toml.
       - name: Build convex-ffi cdylib
-        run: cargo build -p convex-ffi --release
+        run: cargo build -p convex-ffi --profile excel
 
       - name: Stage native library
         run: |
           dest=bindings/java/src/main/resources/native/linux-x86_64
           mkdir -p "$dest"
-          cp target/release/libconvex_ffi.so "$dest/"
+          cp target/excel/libconvex_ffi.so "$dest/"
 
       - name: Set up JDK 22
         uses: actions/setup-java@v4

.github/workflows/release.yml

@@ -85,8 +85,9 @@ jobs:
       - name: Cache cargo registry
         uses: Swatinem/rust-cache@v2
 
+      # `excel` profile = panic=unwind; see [profile.excel] in Cargo.toml.
       - name: Build convex-ffi
-        run: cargo build --release -p convex-ffi
+        run: cargo build --profile excel -p convex-ffi
 
       - name: Setup .NET
         uses: actions/setup-dotnet@v4
@@ -97,7 +98,7 @@ jobs:
         run: dotnet build excel/Convex.Excel/Convex.Excel.csproj --configuration Release
 
       - name: Copy native DLL
-        run: copy target\release\convex_ffi.dll excel\Convex.Excel\bin\Release\net472\
+        run: copy target\excel\convex_ffi.dll excel\Convex.Excel\bin\Release\net472\
 
       - name: Create Add-in Package
         shell: pwsh

.gitignore

@@ -61,6 +61,8 @@ www/pkg/
 npm-debug.log*
 /excel/Convex.Excel/bin
 /excel/Convex.Excel/obj
+/excel/Convex.Excel.Tests/bin
+/excel/Convex.Excel.Tests/obj
 __pycache__/
 *.pyc
 /data

Cargo.toml

@@ -141,6 +141,14 @@ inherits = "release"
 lto = "thin"
 codegen-units = 4
 
+# Host-embedded cdylib profile (Excel add-in, JVM bindings): panics must
+# unwind so convex-ffi's catch_unwind guards can turn them into error
+# envelopes instead of aborting the host process (Excel/JVM). Build with:
+#   cargo build --profile excel -p convex-ffi
+[profile.excel]
+inherits = "release"
+panic = "unwind"
+
 # Distribution profile for binaries
 [profile.dist]
 inherits = "release"

crates/convex-analytics/src/dto.rs

@@ -522,6 +522,133 @@ pub struct CurveQueryResponse {
     pub value: f64,
 }
 
+// ---- YAS (one-call yield & spread analysis) --------------------------------
+
+/// Bloomberg-YAS-style analysis in one request: every yield convention,
+/// G/Z/benchmark/ASW spreads, risk metrics, and the settlement invoice.
+/// Backed by `crate::yas::YASCalculator`.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct YasRequest {
+    pub bond: Handle,
+    pub settlement: Date,
+    pub mark: MarkInput,
+    /// Spot/discount curve (Z-spread; also G-spread unless `govt_curve` set).
+    pub curve: Handle,
+    /// Government curve for G-spread and the benchmark spread.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub govt_curve: Option<Handle>,
+    /// Swap curve for I-spread.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub swap_curve: Option<Handle>,
+    #[serde(default)]
+    pub quote_frequency: Frequency,
+}
+
+/// Yields in percent, spreads in basis points, prices/amounts per 100 face.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct YasResponse {
+    pub clean_price: f64,
+    pub dirty_price: f64,
+    pub accrued: f64,
+    pub accrued_days: i32,
+    pub ytm_pct: f64,
+    pub current_yield_pct: f64,
+    pub simple_yield_pct: f64,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub money_market_yield_pct: Option<f64>,
+    pub g_spread_bps: f64,
+    pub z_spread_bps: f64,
+    pub benchmark_spread_bps: f64,
+    pub benchmark_tenor: String,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub asw_spread_bps: Option<f64>,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub oas_bps: Option<f64>,
+    pub modified_duration: f64,
+    pub macaulay_duration: f64,
+    pub convexity: f64,
+    pub dv01_per_100: f64,
+    pub principal_amount: f64,
+    pub accrued_amount: f64,
+    pub settlement_amount: f64,
+}
+
+// ---- Scenario analysis ------------------------------------------------------
+
+/// One bump component. Tagged JSON: `{ "kind": "parallel" | "steepener" |
+/// "flattener" | "key_rate" | "credit_spread", ... }`.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(tag = "kind", rename_all = "snake_case")]
+pub enum ScenarioBumpSpec {
+    Parallel {
+        shift_bps: f64,
+    },
+    Steepener {
+        short_shift_bps: f64,
+        long_shift_bps: f64,
+        #[serde(default = "default_pivot_tenor")]
+        pivot_tenor: f64,
+    },
+    Flattener {
+        short_shift_bps: f64,
+        long_shift_bps: f64,
+        #[serde(default = "default_pivot_tenor")]
+        pivot_tenor: f64,
+    },
+    KeyRate {
+        tenor: f64,
+        shift_bps: f64,
+    },
+    CreditSpread {
+        shift_bps: f64,
+    },
+}
+
+fn default_pivot_tenor() -> f64 {
+    5.0
+}
+
+/// One named scenario = a list of bumps applied together.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ScenarioSpec {
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub name: Option<String>,
+    pub bumps: Vec<ScenarioBumpSpec>,
+}
+
+/// Run N curve scenarios against one bond in a single call. Methodology:
+/// the mark implies a Z-spread over the base curve; each scenario bumps the
+/// curve and reprices holding that Z-spread fixed.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ScenarioRequest {
+    pub bond: Handle,
+    pub curve: Handle,
+    pub settlement: Date,
+    pub mark: MarkInput,
+    pub scenarios: Vec<ScenarioSpec>,
+    #[serde(default)]
+    pub quote_frequency: Frequency,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ScenarioRow {
+    pub name: String,
+    pub clean_price: f64,
+    pub dirty_price: f64,
+    /// Clean price change vs the base mark.
+    pub delta_clean: f64,
+    /// YTM at the scenario price, decimal.
+    pub ytm_decimal: f64,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ScenarioResponse {
+    pub base_clean: f64,
+    pub base_ytm_decimal: f64,
+    pub z_spread_bps: f64,
+    pub rows: Vec<ScenarioRow>,
+}
+
 // ---- Risk profile / hedge advisor -----------------------------------------
 //
 // `convex_risk_profile` returns a `RiskProfile`; being a serializable value it

crates/convex-analytics/src/spreads/zspread.rs

@@ -197,10 +197,14 @@ impl<'a> ZSpreadCalculator<'a> {
             }
         })?;
 
-        // Convert to basis points
-        let z_spread_bps = (result.root * 10_000.0).round();
+        // 1e-4 bp rounding: suppresses solver float noise without the
+        // DV01-scale round-trip price gap that whole-bp rounding caused.
+        let z_spread_bps = (result.root * 10_000.0 * 10_000.0).round() / 10_000.0;
+        // from_f64 (not from_f64_retain): the division is not exactly
+        // representable in binary, and retain would carry that noise into
+        // the Decimal (51.2346000000000001…).
         Ok(Spread::new(
-            Decimal::from_f64_retain(z_spread_bps).unwrap_or_default(),
+            Decimal::from_f64(z_spread_bps).unwrap_or_default(),
             SpreadType::ZSpread,
         ))
     }
@@ -318,10 +322,14 @@ impl<'a> ZSpreadCalculator<'a> {
             }
         })?;
 
-        // Convert to basis points
-        let z_spread_bps = (result.root * 10_000.0).round();
+        // 1e-4 bp rounding: suppresses solver float noise without the
+        // DV01-scale round-trip price gap that whole-bp rounding caused.
+        let z_spread_bps = (result.root * 10_000.0 * 10_000.0).round() / 10_000.0;
+        // from_f64 (not from_f64_retain): the division is not exactly
+        // representable in binary, and retain would carry that noise into
+        // the Decimal (51.2346000000000001…).
         Ok(Spread::new(
-            Decimal::from_f64_retain(z_spread_bps).unwrap_or_default(),
+            Decimal::from_f64(z_spread_bps).unwrap_or_default(),
             SpreadType::ZSpread,
         ))
     }