Asset / Coupon | Scenario A (Soft Landing 45 %) | Scenario B (Correction 30 %) | Scenario C (Shock 25 %) | RCL Bias |
---|---|---|---|---|
US Equities (S&P) | 🟢 Buy on dips | 🟠 Hold / Hedge | 🔴 Sell → Re-enter lower | ↓ Beta |
Mega-Cap AI Tech | 🟢 Buy selectively | 🟠 Reduce to core holdings | 🔴 Sell vol-sensitive names | ↓ Exposure |
Energy / Defense | 🟢 Buy / Overweight | 🟢 Buy | 🟢 Buy | ↑ Allocation |
US Treasuries (10y) | 🟠 Neutral | 🟢 Buy (Receive Duration) | 🟢 Buy Strongly | ↑ Duration |
Credit (HY) | 🟢 Carry | 🟠 Reduce | 🔴 Avoid | ↓ HY risk |
MBS 6.0–7.5 % | 🟢 Buy carry | 🟢 Buy convexity | 🟢 Buy liquidity | ↑ Weight |
MBS 1.5–3.0 % | 🔴 Sell | 🔴 Sell | 🔴 Sell | ↓ Exposure |
RCL Bias Symbol | Meaning | Portfolio Implication |
---|---|---|
↑ (Positive bias) | RCL loop converges on accumulation signal | Gradually increase exposure / buy |
↓ (Negative bias) | RCL detects risk elevation or fragility | Decrease exposure / sell / hedge |
→ (Neutral bias) | RCL equilibrium; no dominant driver | Hold / wait for new trigger |
↔ (Volatile bias) | Competing nodes oscillate; uncertain regime | Use barbell or hedged positions |
For example: • "↓ Beta" under US Equities = the RCL lattice sees rising systemic uncertainty → reduce market beta (trim index exposure). • "↑ Duration" under US Treasuries = recursion favors duration as a counter-cyclical hedge. • "↑ Weight" under MBS 6.0–7.5 % = RCL suggests overweighting those coupons because lower yields improve convexity returns.
1. Macro Yield Curve Overview The U.S. Treasury yield curve has shifted downward and remains inverted. This suggests a pre-recessionary phase with expected Federal Reserve rate cuts. Lower yields indicate easing financial conditions, weaker growth outlook, and a transition from tightening to an accommodative cycle. 2. MBS Pricing Implications A downward yield curve impacts MBS pricing through refinancing incentives, carry dynamics, and OAS behavior across coupons: - Low-coupon MBS (1.5–3.0%) face accelerated prepayments as mortgage rates decline, reducing duration and richening valuations — reinforcing SELL signals. - Mid-coupon MBS (3.5–5.0%) show mixed behavior; carry remains positive, but refi risk rises. - High-coupon MBS (6.0–7.5%) benefit from limited prepayment exposure and stable carry, supporting BUY signals. 3. Updated Buy/Sell Heatmap
Coupon (%) | Model Signal | Macro Adjustment | Forward View |
---|---|---|---|
1.5–3.0 | Sell | Refi surge, duration collapse | Strong Sell |
3.5–4.5 | Neutral | Curve flattening compression | Soft Sell / Hold |
5.0–5.5 | Neutral | Carry positive, limited risk | Hold / Mild Buy |
6.0–7.5 | Buy | Duration anchor, convexity gain | Strong Buy |
4. Strategy Implications Portfolio positioning should shift up in coupon, emphasizing carry-rich and duration-stable sectors. Key tactical views include: • Overweight 6.0–7.5% FNMA/UMBS TBAs • Underweight 1.5–3.0% low-coupon pools • Rebalance 4.0–5.0% as curve steepens • Consider receiving swaps to hedge duration in lower coupons 5. Summary The downward-shifting yield curve implies imminent monetary easing. MBS relative value shifts toward higher coupons with resilient carry and convexity. Low coupons are vulnerable to renewed refinancing waves, warranting underweight exposure.
1. Overview This report analyzes the September 2025 Agency MBS prepayment data derived from FNMA, FHLMC, and GNMA flash reports. Using the DeepSeek MoE R1 RCL DSA MLA AI_v3 framework, a 3,000-path Monte Carlo simulation was run to model S-Curve prepayment dynamics, effective duration, convexity, and relative value across the coupon stack. 2. Methodology The model applies an incentive-driven S-Curve CPR function where prepayments accelerate as the primary mortgage rate falls below the weighted-average coupon (WAC). The 10-year Treasury rate was modeled as a mean-reverting process (OU), with a 1.0% monthly volatility and 0.18 mean reversion coefficient. Each coupon’s prepayment behavior was simulated over 3,000 random paths, computing discounted cashflows to obtain fair value under a 50 bps base OAS assumption. 3. Model Results
Coupon | BalMM | WAC_% | ModelPrice@50bp | EffDur | EffCvx | OAS_to_Par_bp | Signal |
---|---|---|---|---|---|---|---|
1.5 | 69759.0 | 2.5 | 99.757 | 3.577 | 17.835 | 43.2 | Sell |
2.0 | 682345.0 | 2.912 | 100.271 | 3.602 | 18.157 | 57.5 | Sell |
2.5 | 533006.0 | 3.318 | 100.839 | 3.631 | 18.515 | 72.9 | Sell |
3.0 | 310667.0 | 3.724 | 101.279 | 3.645 | 18.718 | 84.6 | Sell |
3.5 | 227468.0 | 4.155 | 101.895 | 3.675 | 19.118 | 100.6 | Neutral |
4.0 | 187802.0 | 4.65 | 102.526 | 3.701 | 19.488 | 116.6 | Neutral |
4.5 | 138641.0 | 5.11 | 103.168 | 3.728 | 19.879 | 132.4 | Neutral |
5.0 | 168725.0 | 5.641 | 103.784 | 3.743 | 20.11 | 147.4 | Neutral |
5.5 | 238714.0 | 6.125 | 104.49 | 3.771 | 20.497 | 164.0 | Neutral |
6.0 | 230620.0 | 6.645 | 105.123 | 3.786 | 20.776 | 178.7 | Buy |
6.5 | 100267.0 | 7.133 | 105.861 | 3.813 | 21.17 | 195.2 | Buy |
7.0 | 24200.0 | 7.684 | 106.682 | 3.841 | 21.591 | 213.0 | Buy |
7.5 | 2287.0 | 8.377 | 107.516 | 3.851 | 21.844 | 231.5 | Buy |
Table 1. Modeled FN30 Buy/Sell signals based on OAS-to-Par Monte Carlo outputs.
4. Interpretation & Insights Coupons with higher OAS-to-Par values are modeled as cheaper relative to the cross-sectional mean, hence marked as 'Buy'. Conversely, those with lower OAS-to-Par appear rich and are flagged as 'Sell'. The effective duration and convexity figures reflect path-dependent risk sensitivities. This model can be enhanced by integrating the live TBA screen prices and actual forward curve data for more precise relative value assessment.