MARKET SENTINEL
Benchmark-Relative Signal Tracker — Live Sample Data — Market data as of 2026-07-27
US Equities — Relative Watchlist
| Ticker | Sector | Price | Range | Rel. level | State | Signal |
|---|---|---|---|---|---|---|
| MSFT | AAPL | $381.58 | 0.9430 | Inflection | Near Inflection | |
| META | GOOGL | $606.10 | 1.1402 | Inflection | Near Inflection | |
| NVDA | AMD | $208.76 | 1.0843 | Inflection | Near Inflection | |
| AMZN | SPY | $233.66 | 0.7890 | Recovering | Recovering | |
| AAPL | MSFT | $321.66 | 1.3524 | Trending | Outperforming | |
| Most assets currently near the neutral zone — the engine abstains when there is no clear signal. This is correct behavior, not silence. See worked examples below for what clear signals look like. | ||||||
Showing 5 of 14 tickers — create a free account to see the full watchlist, switch horizons, and run deep dives
How to Read This
| Recovering | Below its benchmark-relative floor. Historical relationship favors catch-up versus the reference. |
| Inflection | Near its benchmark-relative anchor. Decision point. No clear directional edge. |
| Trending | Above its benchmark-relative floor. Near the floor, watch for support; when extended, watch for mean reversion. |
The same stock can be Trending on the 1-year view but Recovering on the 3-month view. This is not a contradiction — it means the long-term trend is intact but short-term is pulling back. Multi-horizon analysis shows you the full picture.
TIME MACHINE
Pick any past date. See what the system would have said — using only data available that day.
Then see what actually happened.
Then see what actually happened.
SPY vs XLK
Try any date — earnings, crashes, recoveries, quiet periods
Suggested: Apr 2025 Sep 2025 Aug 2024 crash
Suggested: Apr 2025 Sep 2025 Aug 2024 crash
Why relative, not absolute?
A Trending call during a crash has not failed if SPY drops 10% while the benchmark drops 20% — SPY outperformed its benchmark. Grading on absolute returns just measures market beta.
Note on this pair: SPY contains ~30% tech, so SPY vs XLK is a difficult benchmark — SPY is diluted across 11 sectors competing against a concentrated tech index. A “Recovering” call here asks whether broad market can close a gap against the sector driving it. That’s the right question to ask — and it’s harder to get right than most pairs.
Example calls — graded against what actually happened
INTC / XLK
★ 17m held
Recovering
Very high confidence
Apr 2024–Aug 2025 — relative level fell to 0.43 ($20). Signal held 17 consecutive months. Floor-break warning stayed clear.
✓ +422% vs XLK · $20 → $110 by May 2026 · graded at 12m
Replay-verified · 17 months sustained Recovering · 12-month signal window
ORLY / SPY
★ 20yr
Trending
Recovering
2006–2026 — 49 graded calls across GFC, COVID, and four market cycles. Benchmark-relative floor held throughout. Entry 2007: relative level = 0.47.
81.6% historical hit rate · N=49 · most statistically robust ticker
20-year corpus · GFC included · consistent floor
DVN / SPY
May 2022
Trending
Relative level = 3.08 — Devon Energy 3x above its benchmark-relative floor vs market. Energy supercycle signal. Floor-break warning stayed clear.
✗ −8.8% vs SPY over 12m — sector reverted
We publish the misses. 23.5% of analyzed calls failed; partial outcomes are shown separately.
We show the misses too. Full call log →
Analyst-readable outcomes — not one up/down number
Simple up/down scoring is the wrong measure for a relative timing model.
We grade against what each signal actually claims. Confirmed, partial, and failed outcomes are shown separately.
US Equity · N=3,132 analyzed · 2018–2026 · zero look-ahead
77%
N=3,132
ALL SIGNALS — confirmed or partial
17% confirmed, 59% partial, 23% failed. Partial outcomes mean the relationship was useful,
but timing or magnitude was imperfect.
74%
N=1,549
RECOVERING SIGNALS
Below the benchmark-relative floor. Outcome is graded at the signal window or the
model-implied recovery window for stronger setups.
79%
N=1,583
TRENDING SIGNALS
Above the benchmark-relative floor. Near-floor calls are support checks; extended calls
are mean-reversion checks.
91%
N=432
EXTENDED TRENDS
Deeply extended relationships are graded on movement back toward the relative floor.
This is the strongest Trending subset in the public audit.
78%
N=2,279
CROSS-ASSET PAIRS
Separate audit across crypto, ETFs, commodities, currencies, and futures.
Reference-adjusted price translation: 93% within bounded range or partial.
Graded against what each signal claims — not against simple up/down scoring.
Gate computed from data available at prediction date only (zero look-ahead).
Equity rows use the public stock corpus; cross-asset pairs use a separate tracked-pair audit.
Methodology →
Full call log →
Why these numbers are honest
✓ Benchmark floor frozen at signal time — never recomputed
✓ Seven pair-validity checks — contaminated pairs excluded
✓ Floor-break warning uses only past data — zero look-ahead
✓ Grading window fixed before testing — not swept to find best number
✓ Seven pair-validity checks — contaminated pairs excluded
✓ Floor-break warning uses only past data — zero look-ahead
✓ Grading window fixed before testing — not swept to find best number
✓ Graded against what we predict — not against simple up/down scoring
✓ No fitted knobs — timing rule fixed before testing
✓ Every call public — wins and losses, full log
✓ N=3,132 analyzed calls · confirmed, partial, failed shown separately
✓ No fitted knobs — timing rule fixed before testing
✓ Every call public — wins and losses, full log
✓ N=3,132 analyzed calls · confirmed, partial, failed shown separately
Create Free Account
No credit card required · Starter signup is open