# Daily MBS Routine — 5 min/day, single screen

**Created:** 2026-07-06 (Session 17, Block 011 carry-forward)
**Source:** User request — "if MBS signals are high value, add a time-efficient daily check to the existing yield curve / RR / BF routine" (generalized from Block 010 daily-vol-routine)
**Status:** Active practice
**Companion to:** `daily-vol-routine.md` (Block 010 parallel for vol surface). Do these back-to-back, vol first then MBS, to keep the rhythm.

**Purpose:** A daily routine that turns the Block 011 MBS material into lived experience. Five minutes, one screen, five numbers, one exercise, one decision. Same time, same place, same sequence every day. The MBS market is a *signal layer* for rates, household response, and Fed transmission — most macro PMs use it that way rather than trading the basis directly. This routine reads the signals, not the trade.

**Anchoring principle:** The MBS market is a read on positioning and expectations, not a direct tradable. We're learning to read the prepay-and-basis pulse, not predict mortgage rates.

**Key to abbreviations used in this document:**

| Abbrev | Full name | Notes |
|---|---|---|
| **PMMS** | Primary Mortgage Market Survey | Freddie Mac weekly mortgage rate survey, Thursdays |
| **UMBS** | Uniform Mortgage-Backed Security | Post-2019 single TBA agency MBS (Fannie/Freddie/Ginnie combined) |
| **TBA** | To-Be-Announced | Generic forward-settling agency MBS, the maximally prepayment-exposed benchmark |
| **PSA** | Public Securities Association | Prepayment speed benchmark — 100 PSA = 6% CPR baseline, ramps 0.2%/month to month 30 |
| **CPR** | Conditional Prepayment Rate | Annual prepayment rate (the flow measure PSA proxies for) |
| **WAC** | Weighted Average Coupon | Pool-level; drifts with selective prepays |
| **WAM** | Weighted Average Maturity | Pool-level; extends on no-prepay, contracts on refi |
| **DV01** | Dollar Value of a Basis Point | Per-$100-face price change for a 1bp yield move |
| **FHFA** | Federal Housing Finance Agency | Regulates Fannie/Freddie; sets conforming loan limits |
| **MBA** | Mortgage Bankers Association | Publishes weekly applications survey, Wednesdays |
| **Fed runoff** | Federal Reserve balance sheet reduction (UST + MBS run-off) | Affects TBA supply/demand |

---

## Tool stack (single source of truth)

**Primary:** Freddie Mac PMMS — `https://www.freddiemac.com/pmms` (free, weekly Thursdays)
- 30y FRM and 15y FRM — the primary-secondary spread denominator

**Primary:** FRED H.15 Treasury curve — `https://fred.stlouisfed.org/series/DGS10` and `DGS30` (free, daily)
- 10y UST and 30y UST for basis comparison

**Primary:** FRED MBA Weekly Applications Survey — `https://fred.stlouisfed.org/series/MORTGAGE30US` adjacent; or directly `https://www.mba.org/news-and-research` (free, weekly Wednesdays)
- Refi share of total applications, purchase index level

**Secondary (current-coupon tracking):** UMBS TBA pricing via dealer runs (Bantix on CME if institutional, otherwise broker chat / Bloomberg if available, otherwise implied from PMMS math)
- Current 30y TBA coupon + price — implicit, not a single URL
- Fallback: current coupon ≈ PMMS mortgage rate − ~85-95bp g-fee + servicing + TBA discount margin

**Tertiary (prep speeds):** eNAMIC / eMBS dealer reports for cohort-level PSA speeds (subscription); for the daily routine, infer PSA direction from MBA refi share (no subscription required)

**Write destination:** `01-journal/` daily-journal template. One line per day. Five numbers. Add to the same line as the vol routine (vol first, then MBS):

```
2026-07-06 | ZN ATM 5.5% RR -0.3 BF 0.4 | CVOL 4.7 | FedWatch cut-by-Sep 35% | 30y PMMS 6.43% / 30y UST 4.97% / gross sprd 146bp / refi share 35% / current coupon ~5.58%
```

If running a separate MBS-only line, use:
```
2026-07-06 | 30y PMMS 6.43% | 30y UST 4.97% | gross sprd +146bp | MBA refi share 35% | current coupon ~5.58%
```

**Why this minute matters:** The act of writing is the recall. Anki covers recall over weeks; this covers recall over days. They're complementary, not redundant.

---

## The 5-minute routine

### Minute 0:00 — Read the mortgage-Treasury spread (45 sec)
**Question:** What is the primary-secondary market telling us about duration/credit risk in mortgages?
**Action:** PMMS 30y FRM (weekly, released Thursdays) vs FRED H.15 30y UST yield (daily). Compute: `gross_spread = PMMS_30y - DGS30`.
**Reference levels (recall):**
- Normal range (2024-2026): +150-180bp gross
- Tight (post-COVID, 2021 refi wave trough): ~+130bp
- Wide (2008-2013 stress, 2020 March COVID): +250-400bp
- Current 2026-07: +146bp — **near the tight end of the post-COVID range, but inside the normal range**

**Three states:**
- Spread compressing (PMMS dropping faster than UST, or UST rising faster than PMMS) → household balance sheet improving
- Spread widening → mortgage credit risk pricing higher, or Treasuries rallying faster than mortgages
- Flat → no new info, market stable

**State aloud:** "PMMS 30y at X, 30y UST at Y, gross spread +Z bp. [Direction vs last reading]."

### Minute 0:45 — Read the MBS-Treasury basis (45 sec)
**Question:** How rich or cheap is the agency MBS market vs duration-equivalent Treasuries?
**Action:** Current 30y UMBS TBA coupon (implied from PMMS: `current_coupon ≈ PMMS - 85-95bp`) vs 30y UST yield. Compute: `secondary_spread = current_coupon - DGS30`.
**Reference levels (recall):**
- Normal range: +50-80bp gross secondary
- Tight (option-overpriced): <+50bp — current market
- Wide (option-cheap): >+100bp
- Net basis (after g-fee + servicing) typically -25 to -75bp in steady state, with positive carry on the long-MBS-short-UST basis trade

**Three states:**
- Basis tight (<+50bp) → MBS rich vs Treasuries, the prepayment option is overpriced
- Basis normal (+60-80bp) → fair value, no edge
- Basis wide (>+100bp) → MBS cheap, basis trade is the high-conviction expression

**State aloud:** "Current coupon ~X, 30y UST at Y, secondary spread +Z bp. [Tight / normal / wide vs +60-80bp reference]."

### Minute 1:30 — Read the refi pipeline (60 sec)
**Question:** How many borrowers are *actively* in the refi queue?
**Action:** MBA Weekly Applications Survey — refi share of total applications + refi index YoY.
**Reference levels (recall):**
- Refi share <30%: low refi activity, no wave
- Refi share 30-50%: mild refi incentive
- Refi share 50-70%: clear refi wave in progress
- Refi share >70%: peak refi wave (2003, late 2020, early 2021)

**Connection to Block 011:** The MBA index leads prepay speeds by 4-6 weeks (rate-lock-to-funding lag). Rising refi share = rising PSA in 1-2 months. This is the **forward read on the prepay pulse** — the only forward indicator in this routine.

**State aloud:** "MBA refi share X%, refi index YoY Y%. [Building / steady / fading]."

### Minute 2:30 — Cross-check the coupon cohort (45 sec)
**Question:** Are the prepay speeds consistent with the refi incentive the spread suggests?
**Action:** Mental check: given current PMMS and the coupon distribution of outstanding MBS (5.0% / 5.5% / 6.0% / 6.5% / 7.0% vintages), which cohorts are in-the-money to refi?
- 5.0% pool (2020-2021 vintage): in-the-money by ~143bp → maximum refi incentive, but most have already prepaid
- 5.5% pool: in-the-money by ~93bp → refi-incentivized, current production proxy
- 6.0% pool: in-the-money by ~43bp → borderline
- 6.5% pool: out-of-the-money by ~7bp → no refi incentive, this is the floor for "no refi" baseline PSA

**Cross-check the prepay read from Minute 1:30 against this cohort distribution.** If MBA refi share is rising and 5.0/5.5% cohorts still have material outstanding, the prepay wave is just starting. If MBA refi share is rising but the 5.0/5.5% cohorts are mostly paid down, the wave is already in the back-end (the marginal refi incentive is exhausted).

**State aloud:** "PMMS 30y at X. 5.5% cohort in-the-money by Y bp. PSA speed estimate: Z PSA. [Match / mismatch with MBA read]."

### Minute 3:15 — The Fed transmission check (45 sec)
**Question:** Is the Fed's MBS runoff affecting the basis?
**Action:** Federal Reserve H.4.1 release (Thursdays, free) — Federal Reserve Bank of New York System Open Market Account (SOMA) holdings of agency MBS, week-over-week change. The Fed's MBS holdings are ~$2.2-2.4T post-2022 QT. Net runoff ~$10-15bn/month currently (subject to caps).
**Reference levels (recall):**
- Fed runoff accelerated (MBS runoff >$20bn/month equivalent) → TBA supply pressure → basis widens
- Fed runoff paused → TBA supply tightens → basis tightens
- Fed buying (rare, only in 2020-style emergency) → basis compresses sharply

**State aloud:** "Fed MBS holdings X, WoW change -Y bn. [TBA supply pressure easing / steady / building]."

### Minute 4:00 — Write the number (60 sec) — NOT OPTIONAL
**Action:** One line in your daily log. Five numbers, plus the cross-checks. Same format every day:
```
2026-07-06 | 30y PMMS 6.43% | 30y UST 4.97% | gross sprd +146bp | MBA refi share 35% | current coupon ~5.58%
```
**Why this minute matters:** The act of writing is the recall. Anki covers recall over weeks; this covers recall over days. They're complementary, not redundant.

### Minute 4:30 — The exercise (60 sec, not optional)
**Question:** "If I had to express a view based on today's MBS signals, what would it be? And why would I NOT act on it?"

**Rotate through four daily exercises (don't always do the same):**

**Exercise A — The rates confirmation read.** "30y UST is at 4.97%, PMMS at 6.43%, gross spread +146bp. If I have a view that the long end is going to rally, does the MBS signal confirm or contradict? What does the prepay pulse say about how much room there is for further rally?"

**Exercise B — The household response read.** "MBA refi share is X%. Where in the cohort distribution is the marginal refi borrower? If refi share is rising but the 5.0/5.5% cohort is largely paid down, the wave is exhausting. If the 6.0% cohort is starting to refi, the wave is just starting."

**Exercise C — The Fed transmission read.** "Fed MBS runoff is at X bn/month. If they slow it (cap drop), TBA supply tightens and basis compresses. If they accelerate runoff, basis widens. What's the right trade if I have a view on Fed policy?"

**Exercise D — The basis regime read.** "Secondary spread is +Z bp vs the +60-80bp reference. If tight (<+50bp), the prepayment option is overpriced — short the basis via long specified pool / short TBA. If wide (>+100bp), basis is cheap — go long the basis. Where are we?"

**Decision rule:** If you can't articulate why you'd act AND why you wouldn't, you don't have a view. That's the answer. Most days, you don't have a view. The exercise is teaching you to recognize that.

### Minute 5:00 — Set the watch (15 sec)
**Action:** Look at the calendar. PMMS release (Thursday) / MBA refi (Wednesday) / FHFA conforming loan limit reset (November) / Fed H.4.1 (Thursday) / Treasury refunding announcement (quarterly) in next 5 trading days? If yes, note it. The MBS signal in the 2-3 days *before* a PMMS release is the signal to watch, not the signal on release day (which is already priced).

---

## What to skip (don't add to daily)

- **The full prepayment vector across coupons.** You only need PMMS + 30y UST + MBA refi share. Three numbers. The cohort decomposition is exercise material, not daily material.
- **Specified pool payup tables.** These are trade-level, not signal-level. Quarterly review is enough.
- **Fed's daily RRP / SOFR print.** That's the Block 012 (SOFR-OIS) routine, not this one. Do that separately.
- **Bloomberg MBS function, Intex, Wall Street Analytics.** Institutional tools; not for daily use.
- **eMBS / Empirica subscription data.** Cohort-level PSA speeds. Useful weekly, not daily.
- **TYVIX, CVOL, vol surface.** That's the Block 010 routine. Different lens, different minute.
- **Geopolitical risk premium on mortgages.** Real but small, can't be read daily.

---

## The decision tree (the meta-routine)

After every reading, ask these four questions in order:

```
1. Is the mortgage-Treasury spread widening or compressing?
   - Widening: mortgage credit risk pricing higher, or Treasuries rallying faster
   - Compressing: household balance sheet improving, or Fed transmission working
   - Flat: no new info

2. Is the MBS-Treasury basis rich or cheap?
   - Rich (<+50bp): prepayment option overpriced, basis trade is short
   - Normal (+60-80bp): no edge
   - Cheap (>+100bp): basis trade is long, carry + convexity both work

3. Is the refi pipeline building or fading?
   - Building: rising MBA refi share, more borrowers active in the queue
   - Fading: refi share dropping, marginal refi borrower exhausted
   - Flat: steady state, no new wave

4. Is the Fed runoff affecting the basis?
   - Accelerating: TBA supply pressure, basis widens
   - Pausing: TBA supply tightens, basis compresses
   - Buying: emergency, basis compresses sharply
```

If (2) and (3) and (4) all point in the same direction, the MBS signal is high-conviction. If they conflict, the read is mixed — be cautious about acting. The MBS market is a *signal layer*; the vol surface (Block 010 routine) and the rates curve itself are the primary expression. The MBS signal is the *confirmation or contradiction layer*.

---

## Weekly extension (+5 min, run on Fridays)

After the daily routine, on Fridays add:

1. **Cohort PSA speed check (60 sec):** If you have eMBS / Empirica access, pull the 5.5% cohort PSA speed. Note vs 100 PSA baseline (6% CPR) and 200 PSA (12% CPR). If no subscription, infer from MBA refi share trend. Note: "5.5% cohort running ~Z PSA" or "inferred ~Y PSA from refi share momentum."

2. **Specified pool payup check (60 sec):** If you have dealer runs, pull payup for the 5.0% and 5.5% cohort specified pools vs TBA. If no access, skip. This is institutional-level data; for non-traders, the cohort PSA speed covers 80% of the read.

3. **Fed H.4.1 cumulative check (60 sec):** Federal Reserve Bank of New York SOMA agency MBS holdings — month-over-month change. Cumulative runoff trajectory. Note: "Fed MBS holdings down X bn this month, Y bn YTD."

4. **Trade log link (60 sec):** If you took a position during the week that the MBS signal influenced, write the entry price, the MBS signal reading at entry, and the thesis. If you didn't, write "no trade — MBS signal was: [one sentence]." The exercise is *not* acting on weak signals.

5. **MBS signal move summary (60 sec):** Compare this Friday's MBS numbers to last Friday's. Note moves: PMMS up/down X bp, MBA refi share up/down Y%, basis moved Z bp. This is the "what did the MBS signal do this week" record. Over time, you build an instinct for which moves matter (regime shifts) and which are noise (weekly data noise around the survey).

---

## What this routine does for retention

- **5 minutes/day × ~250 trading days = ~21 hours/year** of deliberate practice on the MBS signal
- **Spaced across 4 distinct concepts:** primary-secondary spread, basis, refi pipeline, Fed transmission
- **Written down daily:** look back at what you thought last week
- **Anki covers recall over weeks** (definitions, formulas, anti-patterns from Block 011)
- **Daily routine covers application over days** (reading live data, classifying regime, articulating a view)
- **You need both.** Recall without application is academic. Application without recall is unreliable.

---

## Practical notes

- **Set a daily reminder** for the same time every day. Variance is the enemy of routines. Suggest: do the vol routine (Block 010) first, then the MBS routine (Block 011) immediately after — total 10 minutes, both lenses covered.
- **Don't skip the write-the-number minute** even if "you'll remember." You won't.
- **Don't skip the exercise minute** even if "you don't have a view." Recognizing no-view is the practice.
- **The 5-minute routine works in any market regime** — the numbers change, the questions don't.
- **If you miss a day, don't double up the next day.** Just resume. The point is the cumulative pattern, not the daily perfect execution.
- **MBA refi share is weekly, not daily.** On non-Wednesday days, the number is the most recent reading. Don't fabricate intraday changes.

---

## Cross-references

- Block 011 synthesis: `~/Obsidian-Macro/00-methodology/study-program/blocks/011-mbs-treasury-basis-prepayment-risk.md`
- MBS resource stack: `~/Obsidian-Macro/03-research/deep/mbs-prepayment-resources.md`
- Vol routine (Block 010): `~/Obsidian-Macro/03-research/deep/daily-vol-routine.md`
- Risk framework v0.3 (sizing reference): `~/Obsidian-Macro/00-methodology/risk-framework.md`
- Anki deck: `Macro Study::Phase 1 — Components::Rates` (6 Block 011 cards tagged `block-011 mbs negative-convexity psa-mechanics basis-trade` + reverse)
- Pre-trade validation: process.md §4.5 reward:risk check (the MBS signal is a *confirmation layer*, not a *trade trigger* by itself)