Rates Intelligence · Methodology · Updated April 24, 2026
How the Rate Lock Index Works
The Rate Lock Index answers one question every morning: if you had to quote a mortgage rate lock today, should you lock it now or wait? Seven market signals, one number (0-100), one decision: lock or float.
The Scale
Above 50 leans LOCK. Below 50 leans FLOAT. The further from 50, the stronger the bias. Stars show how confident the system is — five stars only when the seven factors agree.
0 - 35
Lean FLOAT
Conditions favor waiting — rates likely to drift lower in the relevant window.
35 - 65
No edge
Signal is mixed. Let the closing date drive the decision, not the index.
65 - 100
Lean LOCK
Conditions favor locking — rates likely to drift higher or hold.
Where the Data Comes From
Two public, free sources. Every number on the live dashboard is traceable to one of them. No Bloomberg terminal, no paid feeds.
| Source | What we pull |
|---|---|
| FRED St. Louis Fed | UST yields (2y, 5y, 10y, 30y), Freddie Mac PMMS 30-year and 15-year mortgage rates, the investment-grade corporate OAS used as an MBS-spread proxy, and the Fed funds rate. |
| Alpaca Market data API | Daily price bars for MBB (iShares MBS ETF) and IEF (iShares 7-10 year Treasury ETF). |
FRED
St. Louis Fed
UST yields (2y, 5y, 10y, 30y), Freddie Mac PMMS 30-year and 15-year mortgage rates, the investment-grade corporate OAS used as an MBS-spread proxy, and the Fed funds rate.
Alpaca
Market data API
Daily price bars for MBB (iShares MBS ETF) and IEF (iShares 7-10 year Treasury ETF).
The Seven Factors
Each factor outputs a number from −1 (strong float) to +1 (strong lock). The contributions are weighted, summed, and rescaled to 0-100. Weights add to 100% and were chosen for diversification, not fit.
| # | Factor | Weight | What it watches | How it reads the tape |
|---|---|---|---|---|
| 1 | Market-implied path | 25% | Where is the 10-year Treasury heading vs what the forward curve says it should? | 10y yield z-score vs its forward-curve-implied path over the last 60 days. Stretched above implied path → mean-reverts down → float. Below implied path → revert up → lock. |
| 2 | Bond vol regime | 25% | How jittery is the bond market right now? | ICE BofAML MOVE index (bond-market VIX) + realized vol on the 10y. High vol regime → unpredictable tape, asymmetric downside on a delay → lock. Calm regime → no signal. |
| 3 | Primary-secondary spread | 21% | How much margin are lenders baking into today's rate? | Gap between 30-yr Freddie PMMS and 10-yr Treasury (historical norm ~180 bps). Wide and tightening → margin compression pulling rates down → float. Already tight (<150 bps) → no headroom, lock. |
| 4 | CC MBS OAS momentum | 10% | Is the MBS market richening or cheapening week-over-week? | Current-coupon MBS option-adjusted spread, 5-day slope. Tightening OAS → MBS bid, rates follow down → float. Widening → credit stress, rates back up → lock. |
| 5 | Event-risk density, 14d | 8% | What's on the Fed + inflation + jobs calendar over the next 14 days? | Weighted count of FOMC (1.0), CPI (0.85), NFP (0.70), PCE (0.55), Treasury auctions (0.30-0.35). Heavy calendar → lock through it. Light calendar → no signal. |
| 6 | Dealer/flow proxy | 8% | Are dealers long or short MBS relative to their hedges? | MBB returns regressed against IEF over a 20-day window. MBB persistently underperforming the hedge → dealers long-axis (supply overhang) → eventual compression → float. |
| 7 | RSI (10y) | 3% | Where is the 10-year on its own chart? | 14-day RSI on 10-year yield. Overbought (>70) → expect pullback → float. Oversold (<30) → expect bounce → lock. Mid-range (30-70) → no signal. |
Market-implied path
Where is the 10-year Treasury heading vs what the forward curve says it should?
10y yield z-score vs its forward-curve-implied path over the last 60 days. Stretched above implied path → mean-reverts down → float. Below implied path → revert up → lock.
Bond vol regime
How jittery is the bond market right now?
ICE BofAML MOVE index (bond-market VIX) + realized vol on the 10y. High vol regime → unpredictable tape, asymmetric downside on a delay → lock. Calm regime → no signal.
Primary-secondary spread
How much margin are lenders baking into today's rate?
Gap between 30-yr Freddie PMMS and 10-yr Treasury (historical norm ~180 bps). Wide and tightening → margin compression pulling rates down → float. Already tight (<150 bps) → no headroom, lock.
CC MBS OAS momentum
Is the MBS market richening or cheapening week-over-week?
Current-coupon MBS option-adjusted spread, 5-day slope. Tightening OAS → MBS bid, rates follow down → float. Widening → credit stress, rates back up → lock.
Event-risk density, 14d
What's on the Fed + inflation + jobs calendar over the next 14 days?
Weighted count of FOMC (1.0), CPI (0.85), NFP (0.70), PCE (0.55), Treasury auctions (0.30-0.35). Heavy calendar → lock through it. Light calendar → no signal.
Dealer/flow proxy
Are dealers long or short MBS relative to their hedges?
MBB returns regressed against IEF over a 20-day window. MBB persistently underperforming the hedge → dealers long-axis (supply overhang) → eventual compression → float.
RSI (10y)
Where is the 10-year on its own chart?
14-day RSI on 10-year yield. Overbought (>70) → expect pullback → float. Oversold (<30) → expect bounce → lock. Mid-range (30-70) → no signal.
How the Final Number Is Built
The final score is a weighted sum of the seven factor signals, rescaled from the [−1, +1] sum into the 0-100 band.
weighted_sum = 0.25 × path + 0.20 × event + 0.15 × spread
+ 0.15 × vol + 0.10 × oas + 0.08 × flow
+ 0.07 × tech
RLI = 50 + 50 × weighted_sum (clamped to [0, 100])Worked example — 54 / 100 NEUTRAL ★
A reading from a recent morning, broken down factor by factor. Each factor's contribution = signal × weight × 50.
| Factor | Signal | Weight | Contribution |
|---|---|---|---|
| 10y stretched high (path) | −0.50 | 25% | −6.2 pts (lean float) |
| Calendar light (event) | +0.42 | 20% | +4.2 pts |
| Spread tight (primary-secondary) | +0.51 | 15% | +3.8 pts (lean lock) |
| MBB outperforming IEF (flow) | +0.68 | 8% | +2.7 pts (lean lock) |
| Other factors (smaller) | — | ~ −0.2 pts | |
| Net | — | 50 + 4.3 ≈ 54 | |
10y stretched high (path)
Signal
−0.50
Weight
25%
Contrib.
−6.2 pts (lean float)
Calendar light (event)
Signal
+0.42
Weight
20%
Contrib.
+4.2 pts
Spread tight (primary-secondary)
Signal
+0.51
Weight
15%
Contrib.
+3.8 pts (lean lock)
MBB outperforming IEF (flow)
Signal
+0.68
Weight
8%
Contrib.
+2.7 pts (lean lock)
Other factors (smaller)
Signal
—
Weight
—
Contrib.
~ −0.2 pts
Result: 54 falls in the NEUT band (45-55) with one star — system has no meaningful edge today.
How the Star Rating Works
Stars require two conditions to both be true: distance from 50, and factor agreement. A 5-star reading means seven independent signals all point the same way. That's rare.
| Stars | Meaning |
|---|---|
| ★★★★★ | Score ≥ 85 or ≤ 15, and ≥70% of factor weight aligned |
| ★★★★ | Score ≥ 75 or ≤ 25, and ≥60% aligned |
| ★★★ | Score ≥ 65 or ≤ 35, and ≥50% aligned |
| ★★ | Score moderately away from 50 |
| ★ | Score near 50 — no edge |
The stars answer a single question: is the whole system agreeing, or is one loud factor dragging the score? If the RLI reads 75 (LOCK) but only path-deviation is pulling up and the rest of the table is flat, that's 25% agreement — three stars, not five.
Why This Is Credible
Every factor has a published reason, a published weight, and a current value visible on the live dashboard. Nothing is hidden. Accuracy is tracked daily by comparing each past RLI reading to the realized 30-year mortgage rate move (Freddie Mac MORTGAGE30US series) at 14, 30, and 60 day horizons.
14-day · n=32
72%
direction hit rate
30-day · n=20
90%
direction hit rate
Brier score
0.130
0 = perfect, 0.25 = coin flip
Calibration check
2.7★ when right vs 1.0★ when wrong ✓
The real test of whether the system is calibrated or just lucky: high-conviction calls should be right more often than low-conviction ones. They are.
Refresh cadence: RLI is computed weekdays at 7:00 AM PT after FRED settles. Accuracy is rescored at 7:30 AM PT. The dashboard auto-updates — no intraday override, one honest snapshot per trading day.
As of April 24, 2026. Live numbers on the dashboard.
Common questions
What is the Rate Lock Index?
The Rate Lock Index answers one question every weekday morning: do market conditions favor locking a mortgage rate today, or floating? It publishes a headline LOCK, FLOAT, or WATCH call backed by calibrated model probabilities, alongside the underlying 0-100 score computed from a 7-factor bond-market model. It is research and education, not personalized advice or a recommendation to lock or float a specific loan.
Where does the Rate Lock Index data come from?
Two free public sources. FRED (Federal Reserve Bank of St. Louis) provides Treasury yields (2-year, 5-year, 10-year, 30-year), the Freddie Mac PMMS 30-year and 15-year mortgage rates, and the investment-grade corporate OAS used as an MBS spread proxy. Alpaca market data provides daily price bars for MBB (the iShares MBS ETF) and IEF (the iShares 7-10 year Treasury ETF). No paid data feeds are used.
What are the seven factors in the Rate Lock Index?
Path deviation (25% weight) — how stretched the 10-year Treasury is from its 60-day moving average. Event risk (20%) — density of upcoming Fed and inflation events. Primary-secondary spread (15%) — the gap between mortgage rates and the 10-year Treasury. Volatility regime (15%) — MBB realized volatility. OAS momentum (10%) — investment-grade credit spread direction. Dealer flow (8%) — MBB performance vs IEF. Technical (7%) — 14-day RSI on the 10-year yield.
How does the model decide between LOCK, FLOAT, and WATCH?
Two model generations work together. A baseline model estimates the probability that mortgage rates rise over the next 30 days, and a regime-conditional model adjusts for the current Fed phase (hiking, holding, or cutting). When the probabilities show a clear edge, the published call is LOCK or FLOAT with a stated confidence level; when there is no meaningful edge, the call is WATCH. The live dashboard documents the decision logic and shows both probabilities every day.
How accurate is the Rate Lock Index?
Accuracy is tracked daily by comparing each past reading to the realized 30-year mortgage rate move (Freddie Mac MORTGAGE30US series) at 14, 30, and 60 day horizons. Out-of-sample, roughly two out of three high-conviction calls have been correct. Direction hit rate, Brier score, and a calibration check — whether high-conviction calls are right more often than low-conviction ones — are all published on the live dashboard.
Is the Rate Lock Index investment advice?
No. The Rate Lock Index is research and education only. It is not personalized advice, not a recommendation to lock or float a specific loan, and not an offer of credit. Past performance does not predict future rates. For a binding rate quote, contact your loan officer. NMLS #2636410.
Research only · not investment advice · not a rate lock commitment
The Rate Lock Index is research and education. It is not personalized advice, not a recommendation to lock or float a specific loan, and not an offer of credit. Past performance does not predict future rates. Markets can move sharply on data we do not yet have. For a binding rate quote and personalized analysis, contact your loan officer. NMLS #2636410. West Capital Lending, Inc. NMLS #1566096. Equal Housing Opportunity.
See today's reading
The live dashboard publishes the score, conviction stars, all seven factor bars, the 60-day sparkline, and rolling accuracy tiles. Refreshes weekday mornings at 7:00 AM PT after FRED settles.
Open the live Rate Lock Index