Home › Technical › Technical Overview › PL
Technical Overview
PL
Platinum
BEAR
1.0/10 · Most bearish reading since May 2015
Validated technical signals are pointing bearish, at an extreme level versus this market’s own history.
Flipped from LEAN BEAR to BEAR on 2026-09-28 — 29 trading days in current read.
Latest bar: 2026-11-05
The read at three time scales
The same tape, read on three different clocks. Daily is the fast-moving tactical read, Weekly is the slower structural read, and Confluence only speaks when both the daily and weekly reads agree with each other.
Daily
Fast-moving tactical read
BEAR
1.0/10
9 of 33 signals active
Weekly
Slower structural read
BEAR
1.0/10
19 of 26 signals active
Confluence
Daily and weekly agree
NEUTRAL
5.5/10
3 of 9 signals active
What each kind of signal is saying
Only showing categories with signals active right now. When trend signals and momentum signals disagree, that tension is the story.
Trend
Fractal dimension
NEUTRAL · 1/2 active
Trend
LEAN BULL · 1/2 active
Breakout
Breakout & range
BEAR · 2/2 active
Cycle
Ehlers DSP cycles
LEAN BEAR · 1/2 active
Regime
Information theory
BEAR · 3/3 active
Regime detection
BEAR · 7/9 active
Statistical tests
BEAR · 3/7 active
Structure
Microstructure
BEAR · 1/7 active
Strongest signals right now
The ten most influential technical signals currently active. Each has been validated against years of history — the h60 return column is the average 60-day move that historically followed the signal, and hit is how often it was in the right direction.
| Dir | Scale | Type | Signal | Edge | Recent fires | h60 return | Hit rate | Sample |
|---|---|---|---|---|---|---|---|---|
| BEAR | Weekly | Statistical tests | Jbnorm | Inverse | 4 | +5.02% | 69% | 600 |
| BEAR | Weekly | Regime detection | Vol regime (2-state HMM) | Inverse | 4 | +3.54% | 55% | 1670 |
| BEAR | Weekly | Statistical tests | ADF stationarity | Inverse | 4 | +2.69% | 58% | 1185 |
| BEAR | Weekly | Regime detection | Gaussian-mixture regime | Inverse | 3 | +3.76% | 56% | 1240 |
| BULL | Weekly | Breakout & range | Donchian channel (55) | Direct | 2 | +2.50% | 53% | 1867 |
| BEAR | Weekly | Regime detection | HMM Fwd 2State | Inverse | 2 | +7.57% | 69% | 448 |
| BEAR | Weekly | Information theory | Mutual information | Inverse | 2 | +3.71% | 56% | 838 |
| BULL | Weekly | Information theory | Mutual information | Direct | 2 | +3.35% | 55% | 824 |
How the read has shifted (last 60 trading days)
Each bar is one trading day. Bars above the zero line are bullish; bars below are bearish. Height shows how strong the read was.
How the verdict is built
The verdict aggregates every classical technical signal we track on this market — momentum, trend, volatility, cycle, breadth, and so on — into a single read. Each signal has been validated out-of-sample on this market’s history (currently 68 signals pass the bar for PL). Four factors determine how much each signal contributes.
- Edge strength. How large and reliable is the historical edge? A signal with a 70% hit rate and a 6% average 60-day follow-through contributes more than one with a 54% hit rate and a 2% follow-through.
- Recency. A signal that fired this week counts more than one that fired three months ago. Fast daily signals fade out over about six weeks; slower weekly signals stay relevant for about six months.
- Confirmation. After a signal fires, we watch the market. If price moves the way the signal predicted (vol-adjusted), the signal’s contribution is amplified. If the market moves the other way, the contribution fades toward zero.
- Market-adaptive ranking. The final reading is ranked against this market’s own history over the past ~6 years. A bearish bias that would barely register on crude oil could be an extreme reading on the S&P — percentile ranking handles this automatically, so verdicts are comparable across markets.
Confidence tier
How many independent signals are currently driving the verdict: HIGH (15+ active) · MODERATE (5–14) · LOW (fewer than 5). A high-confidence read is backed by many independent signals pointing the same way; a low-confidence read could be a handful of signals amplified by a lucky move.
What this is not
- Not a trade recommendation. No transaction cost, slippage, or position-sizing model is applied here.
- Not a backtest of the overall verdict. The individual signals are validated out-of-sample, but the way they’re combined into one number is a consensus device, not itself a trading system.
- Not treating correlated signals as independent. Several momentum indicators pointing the same way count as separate votes here.
Refreshed nightly. Last update: 2026-11-07T23:00:02+00:00.