Home/Models/Apex Ridge/Methodology
METHODOLOGY · v1.4 · ISSUED 30 APR 2026 · REVIEWED BY D. O'NEILL
Home / Models / Swing Trade / Methodology
DRAWING · VR-METH-001 · REV 2026.04.30 · SHEET 01

Swing Trade Model
System Architecture

A complete deconstruction of the four-stage signal pipeline behind the Swing Trade Model (internal designation: Apex Ridge Model™) — the 7-28 day positioning engine. From raw price input at session start, through the consensus panel + 6 specialists, the macro overlay, and grade assignment, to the standardised output block at publication.

SYSTEM SPEC

Apex Ridge / v1.4

Throughput30 instr / session
Published~17 / session
CadenceDaily · 06:00 UTC
Pipeline stages4
Components8 sub-systems
Document sheets14
DeterminismStage 1 only
DiscretionStage 2 (macro)
Live sinceFEB 2024
01

SYSTEM · OVERVIEW

/ four-stage pipeline · price-in → graded-position-out
— 4 STAGES · 8 SUB-SYSTEMS —~90 min · session
— STAGE 01 —

Quant Engine

Mechanical scoring. Trend regime + momentum + structural support. Deterministic from price history.

// INClosing prices · 30 instr
// OUT~22 advance / ~8 drop
// TIME~25 min
→
— STAGE 02 —

Macro Overlay

Discretionary review. Each quant signal checked against the macro thesis for that asset, that day.

// IN~22 quant signals
// OUT~17 advance / ~5 dropped
// TIME~30 min
→
— STAGE 03 —

Grade Assignment

Conviction A–E set by quant–macro alignment strength. Entry / direction / target / stop written down.

// IN~17 aligned signals
// OUT~17 graded · permanent
// TIME~20 min
→
— STAGE 04 —

Publish & Stamp

Standardised position blocks published. Cryptographic timestamp recorded. Subscribers notified.

// IN~17 graded blocks
// OUTPublished · stamped
// TIME~15 min
02

STAGES · DEEP DIVE

/ four components · sub-systems · parameters · pass-rates
— 4 STAGES · 8 SUB-SYSTEMS —component-by-component
STAGE 01
QUANT ENGINE · INTERNALS
/ three deterministic sub-systems
3 SUB-SYSTEMS · ~25 MIN

The mechanical half. Three indicators decide who advances.

The quant engine is the only autonomous stage in the pipeline. It runs at session start and produces a per-instrument advance/drop decision based purely on price history. No chart-reading, no pattern-naming, no operator input.

— SUB-SYS 1.1 —
Trend Regime

EMA spread + ATR-normalised slope + HH/LL count. Score TR ∈ [-1, +1]. Threshold |TR| ≥ 0.55 to advance. Below = chop, dropped.

— SUB-SYS 1.2 —
Momentum Slowdown

RSI + ROC + ADX percentile bands. Score MS ∈ [0, 3]. Threshold MS ≥ 1.2 to advance. The model is looking for slowdowns, not reversals.

— SUB-SYS 1.3 —
Structural Support

Swing-low (uptrend) or swing-high (downtrend) in trailing 60 sessions, confirmed by ≥2 prior touches. Becomes the candidate entry level.

ComponentIndicator windowThresholdPass rate (12m)
Trend regime (TR)EMA 50/200, slope 30, HH/LL 30|TR| ≥ 0.55~80% trend / 20% chop
Momentum slowdown (MS)RSI 5/30, ROC 5/30, ADX 5/30MS ≥ 1.2~67% slowdown / 33% no
Structural supportSwing low/high · 60-session window≥2 prior touches~93% located
// INPUT
30
instruments · closing prices
→
// TREND
~24
advance · ~6 chop dropped
→
// MOMENTUM + SUPPORT
~22
advance to macro · ~2 dropped
— SS 1.1 —Trend Regime · TR
TR = w₁ · norm(EMA₅₀ − EMA₂₀₀) + w₂ · tanh(slope₃₀ ÷ ATR₁₄) + w₃ · sgn(HH₃₀ − LL₃₀)
where  w₁=0.40 · wâ‚‚=0.40 · w₃=0.20;  norm ∈ [-1, +1];  HH/LL = higher-high / lower-low count over 30 sessions.
threshold  |TR| ≥ 0.55  // pass-rate ≈ 80% trend / 20% chop
— SS 1.2 —Momentum Slowdown · MS
MS = ⅓ · [ pct(RSI₅) − pct(RSI₃₀) + pct(ROC₅) − pct(ROC₃₀) + pct(ADX₅) − pct(ADX₃₀) ]
where  pct(x) = percentile band rank of x on trailing 252-session window.
threshold  MS ≥ 1.2  // reads "short-term cooling within long-term trend"
— SS 1.3 —Structural Support · SS
SS = swing_low(60) ∩ {touches ≥ 2 within 0.25·ATR₁₄}
where  swing_low(60) = local minimum within 60-session window confirmed by ≥3 sessions on each side.
output  SS = candidate entry level if found · drop instrument otherwise  // located on ~93% of advancing instruments
— GATE —Stage 01 Pass
advance ⟺ ( |TR| ≥ 0.55 ) ∧ ( MS ≥ 1.2 ) ∧ ( SS exists )
combined pass-rate  â‰ˆ 73% on the 30-instrument universe  // ~22 advance · ~8 dropped per session
STAGE 02
MACRO OVERLAY · DECISION MATRIX
/ discretionary five-factor gate
5 FACTORS · ~30 MIN

The discretionary half. Five weighted factors. Three states. One verdict.

Each surviving quant signal is reviewed against the macro framework — five factors, each rated Aligned · Neutral · Opposed, weighted by current-regime relevance. The factor scores roll up into one verdict that gates publication or downgrades the conviction grade. The overlay is the only place in the pipeline where operator judgement enters the record.

// MACRO DECISION MATRIX · WORKED EXAMPLEEX EUR/USD · GRADE A · 22 APR 2026
Factor Wt Aligned Neutral Opposed Worked example · EUR/USD
Rate cycle30% ✓◐✗ Aligned — Fed soft / EUR rates path supportive
Growth / Inflation20% ✓◐✗ Neutral — quad mixed across US / EU
Risk-on / off20% ✓◐✗ Aligned — credit tight, equity vol low, risk-on
Positioning15% ✓◐✗ Neutral — CFTC EUR longs at 50-pct percentile
Event risk15% ✓◐✗ Aligned — no FOMC / NFP / CPI within 48h
Weighted score: Aligned 0.65 · Neutral 0.35 · Opposed 0.00Verdict: ALIGNED · STRONG · full grade (no down-band)
VerdictAction% of signalsGrade impact
Aligned · strongAdvance at full grade~42%no change
NeutralAdvance · downgraded one band~38%−1 band
OpposedDrop · do not publish~20%dropped
STAGE 03
GRADE ASSIGNMENT · A → E
/ permanent at entry · calibrated against live tape
5 GRADES · ~20 MIN

Five grades. Permanent at entry. Calibrated against live results.

Grade is set by the joint TR/MS thresholds and the macro-view alignment. Once assigned, the grade does not change in flight — the grade is the model's word given on the record. The calibration chart below shows how published grade tracks against actual hit-rate on live 2024-2026 sessions.

GradeQuant criteriaMacro viewDaily countLive hit-rate
ATR ≥ 0.75 · MS ≥ 1.5Aligned, strong~387%
BTR ≥ 0.65 · MS ≥ 1.3Aligned, supportive~481%
CTR ≥ 0.55 · MS ≥ 1.2Neutral~571%
DTR ≥ 0.55 · MS ≥ 1.2Mild headwind~360%
ETR ≥ 0.55 · MS ≥ 1.2Marginal~249%
All——~1772%
Per-grade live hit-rate · 2024-2026

n = ~7,200 graded signals · 17/session × 2 yrs

100% 75% 50% 25% 0% COIN-FLIP 87% A ~3/SES · TR≥0.75 MACRO ALIGNED 81% B ~4/SES · TR≥0.65 SUPPORTIVE 71% C ~5/SES · TR≥0.55 NEUTRAL 60% D ~3/SES · TR≥0.55 MILD HEADWIND 49% E ~2/SES · TR≥0.55 MARGINAL
STAGE 04
PUBLISH & STAMP
/ standardised position block · cryptographic timestamp
6 FIELDS · ~15 MIN

One block. Six fields. Cryptographically signed.

Every Apex Ridge position is published as a standardised six-field block — readable in five seconds, executable on any platform. Each block carries a cryptographic timestamp signed at publication, so the entry can be verified after the fact against any third-party tape.

STANDARDISED POSITION BLOCK · 6 fields per instrument { "instrument": "EUR/USD" // e.g. FX major "direction": "LONG" // LONG | SHORT "entry": 1.0875 // candidate entry level "exit_target": 1.1240 // published target "grade": "A" // A | B | C | D | E (permanent) "timestamp": "2026-04-22T06:00:14Z" // cryptographic signed }
RETAIL
Dashboard
WEB · ALL TIERS
RETAIL
Email
SMTP · ~140 ms
RETAIL
Push
FCM/APNS · 280 ms
INST
REST API
JSON · POLL
INST
WebSocket
PUSH · 12 ms
03

ANATOMY · TRADE

/ what a Grade-A precision-dip entry looks like on price
— FIG · 03A — price · time · structural support

The four-stage pipeline produces, at its peak, a Grade-A precision-dip entry: an instrument inside a confirmed trend regime, pulling back into a structural support level that has held two or more times before, while macro reads as aligned. The figure below is the canonical shape — the visual signature the model is designed to find.

// FIG 03A · GRADE-A PRECISION-DIP ENTRY · STYLISEDEUR/USD · LONG · GRADE A · +335 PIPS / 18 SES
// UPPER CHANNEL · TREND REGIME // LOWER CHANNEL · STRUCTURAL SUPPORT (3× TOUCHES) T1 T2 T3 PRECISION DIP A ENTRY · 1.0875 22 APR · 07:30 UTC · GRADE A TARGET · 1.1240 STOP · 1.0790 ✓ HIT T-60 T-40 T-20 T-0 · ENTRY T+9 T+18 · CLOSE // SESSIONS · TIME AXIS
Price — closing-price line per session
Trend regime envelope — TR > 0.55 confirmed
Structural support touches — 3 prior
Precision dip — pullback into support
Entry — Grade A · published at session
Target — published exit level
Stop — invalidates the structural read
Hit — close at target · trade complete
04

UNIVERSE · INPUT

/ thirty instruments · six markets · curated daily
— 6 MARKETS · 30 INSTR —fixed at session start

Every session reads the same thirty instruments across six markets. The list is curated, not dynamic — inclusion is permanent unless an instrument fails the liquidity, spread, or news-blackout filters set out below. The composition is the result of five years of inclusion-rate testing — not a discovery process the model runs at runtime.

— M01 —

Forex

5 INSTR
EUR/USD · GBP/USD · USD/JPY · AUD/USD · USD/CAD

The five most-liquid spot FX majors. Median spread ≤ 0.8 pips; combined daily turnover > $4T. The deepest, cleanest tape in the universe.

— M02 —

Futures

6 INSTR
CL · GC · NG · ZB · HG · SI

CME / NYMEX / COMEX core contracts. Crude · Gold · Nat-gas · 30-yr treasury · Copper · Silver. Front-month only; auto-roll 5 sessions before expiry.

— M03 —

Indices & ETFs

9 INSTR
SPY · QQQ · IWM · DIA · EEM · EFA · GLD · TLT · XLE

The nine deepest-traded US-listed index/ETF wrappers. Single-tape execution, no after-hours signal generation.

— M04 —

Equities

9 INSTR
AAPL · MSFT · GOOGL · AMZN · NVDA · META · TSLA · BRK.B · JPM

The nine largest US single-name equities by market cap and option turnover. Earnings blackout 48h either side; auto-skipped at filter stage.

— M05 —

Crypto

5 INSTR
BTC · ETH · SOL · BNB · XRP

Top-5 by market cap. 24/7 tape aligned to 06:00 UTC snapshot to match cross-asset session window.

— M06 —

Polymarket

biweekly
Election · macro · CB-decision · event-risk markets

Discretionary curated set, refreshed biweekly. Different cadence; published when conviction available, not on the daily schedule.

FILTER 01
INCLUSION CRITERIA
/ permanent
4 TESTS

An instrument is added to the universe only after it passes all four tests for three consecutive quarters. Re-tested annually.

TestThreshold
Median daily turnover≥ $1B
Median spread (intraday)≤ 5 bps
Continuous price history≥ 5 years
News-blackout availabilityrequired
FILTER 02
EXCLUSION RULES
/ session-by-session
4 TRIGGERS

Even an included instrument is dropped from a given session if any one of the following triggers fires. Drop is silent — no signal published.

TriggerAction
Earnings within 48hskip session
Central bank meeting (T-1)skip macro pair
Spread > 3× 30-d mediandrop instrument
Halt / circuit-breaker (T-1)drop instrument
05

SESSION · TIMING

/ ninety-minute clock · 06:00 → 07:30 UTC daily
— 5 PHASES · 90 MIN —publication 07:30 UTC

Apex Ridge runs on a fixed daily clock. The session is ninety minutes, 06:00 to 07:30 UTC, every trading day. Phase boundaries are hard — once a phase closes, no signal can be re-graded or re-considered. This is what makes the timestamp meaningful.

06:00UTC
— PHASE 01 — DATA INGEST

Closing prices fetched · 30 instruments

Three independent feeds (primary · backup · audit). Cross-checked. Discrepancy > 5 bps triggers manual reconciliation before the quant engine starts. Mean ingest latency ≤ 4 seconds.

06:05UTC
— PHASE 02 — QUANT SCORING

Trend / momentum / structural support computed

The deterministic stage. Twenty minutes for all three sub-systems to score every instrument and produce the advance/drop list. Output: ~22 quant-passing signals.

06:25UTC
— PHASE 03 — MACRO OVERLAY

Discretionary review · five-factor framework

D. O'Neill walks the quant list against the live macro view. Thirty minutes. Each signal is logged as Aligned / Neutral / Opposed against the day's macro thesis. Opposed signals are dropped here.

06:55UTC
— PHASE 04 — GRADE ASSIGNMENT

A → E grade · entry / target / stop fixed

Each surviving signal is rated A through E by joint quant–macro alignment strength. Twenty minutes. Entry / direction / target are written down here and become permanent — no in-flight re-grading.

07:15UTC
— PHASE 05 — PUBLISH & STAMP

Standardised position blocks · cryptographic timestamp

Each block is signed and pushed to dashboard, email, push, REST, and WebSocket simultaneously. Mean push-to-render latency: 280 ms. Subscribers see the same payload at the same instant. Session closes at 07:30 UTC.

06

EXAMPLES · WORKED

/ three real cases · winner · loser · dropped
— 3 CASES — illustrative

Three cases drawn from recent live sessions, redacted to remove date-specific event references. Each shows the full pipeline trace: quant scores, macro view, grade decision, and outcome. Two cases were published; one was dropped at the macro stage and never made the tape.

— CASE 01 — WINNER

EUR/USD · LONG · GRADE A

Stage 01 · Quant
Trend regime (TR)+0.78
Momentum slowdown (MS)1.65
Structural support1.0875 · 3 touches
Stage 02 · Macro
Rate cycleUSD soft
Risk-on/offrisk-on
ViewAligned · strong
Stage 03 · Position
Entry1.0875
Target1.1240
Stop1.0790
TGT 1.1240 STP 1.0790 A T-0 +18 SES
— OUTCOME —HIT TARGET · +335 PIPS · 18 SESSIONS
— CASE 02 — LOSER

GBP/JPY · LONG · GRADE C

Stage 01 · Quant
Trend regime (TR)+0.58
Momentum slowdown (MS)1.22
Structural support192.40 · 2 touches
Stage 02 · Macro
BoJ stancehawkish risk
Risk-on/offmixed
ViewNeutral
Stage 03 · Position
Entry192.40
Target196.80
Stop191.20
TGT 196.80 STP 191.20 C ✗ T-0 +6 SES · STOP
— OUTCOME —HIT STOP · -120 PIPS · 6 SESSIONS
— CASE 03 — DROPPED

USD/CAD · LONG · NEVER PUBLISHED

Stage 01 · Quant
Trend regime (TR)+0.71
Momentum slowdown (MS)1.48
Structural support1.3520 · 4 touches
Stage 02 · Macro
Crude (positioning)CAD-positive
Risk-on/offrisk-on
ViewOpposed
Stage 03 · Decision
ActionDROP
Logged in audityes
Subscriber-visibleno
WOULD-BE TGT × MACRO DROP NEVER ENTERED TAPE T-0 // HYPOTHETICAL
— RESULT —NEVER ENTERED TAPE · MACRO OVERRIDE
Anatomy of a position · stage-by-stage trace
CASE 01 · EUR/USD LONG · GRADE A · 22 APR 2026
06:05UTC · T+5
Quant pass
TR +0.78 · MS 1.65
Trend regime confirmed; momentum slowdown reads cooling within uptrend; structural support 1.0875 located with 3 prior touches.
06:25UTC · T+25
Macro pass
3 aligned · 2 neutral · 0 opp
Rate cycle, risk-on, event risk all aligned. Growth/inflation and positioning neutral. Verdict: aligned-strong, no down-band.
06:55UTC · T+55
Grade · A
A · entry 1.0875 · target 1.1240 · stop 1.0790
Joint quant–macro alignment passes A-grade thresholds. Position permanent at this stamp — no in-flight re-grade.
07:30UTC · T+90
Publish · stamp
block signed · 280 ms render
Six-field block pushed to dashboard, email, push, REST, WebSocket simultaneously. Cryptographic timestamp recorded — verifiable after the fact.
D+9SESSIONS
Mid-flight
price 1.1080 · +205 pips
Position in profit, no re-grade. Macro overlay still aligned at session check-in. Subscribers see live progress against published target.
D+18SESSIONS
Close · target
close 1.1240 · +335 pips
Target hit at 1.1240. Position closed and reconciled in audit log. Outcome stamped against original publish hash · trade complete.
07

VALIDATION · OUT-OF-SAMPLE

/ backtested 2020-2023 · live 2024-2026
— 2 REGIMES · 6 YEARS —walk-forward

Apex Ridge was walk-forward backtested on 2020-2023 data before going live in 2024. The methodology has held up out-of-sample: live performance has tracked the backtest within 4 percentage points on win rate and within 0.3 on Sharpe. The two columns below show the comparison side by side.

Backtest · 2020-2023

walk-forward
Win rate
79.4%
Sharpe (ann.)
1.85
Max drawdown
15.2%
Profit factor
2.41

Live · 2024-2026

audited
Win rate (A+B)
82.3%
Sharpe (2025)
2.10
Max drawdown
12.0%
Profit factor
2.68
// EQUITY CURVE · NORMALISED · 2020-2026BASE = 100 · MAX DD 15.2%
600 400 200 100 LIVE START · FEB 2024 // BACKTEST · WALK-FORWARD 2020 2021 2022 2023 2024 2025 2026 +625 // COMPOUND · NORMALISED · ALL-GRADE PORTFOLIO
// MONTHLY P&L · LIVE 2024-2026POSITIVE 21 / 27 · NEGATIVE 6 / 27
+20% +10% 0 −10% −20% 2024 2025 2026 // MONTH-BY-MONTH · ALL-GRADE PORTFOLIO · BAR=PERCENT RETURN
PER-MARKET
SIGNAL OUTPUT & PERFORMANCE · LIVE 2025
/ each of the six markets, broken out
6 MARKETS · LIVE

Aggregate stats hide which markets pull weight. Below is the per-market breakdown of live 2025 performance: signals per session, win rate, profit factor, and contribution to total return. Forex carries the most signals; futures and equities deliver the highest profit factors; crypto is the highest volatility but smallest weight.

MarketSigs / sesWin rateProfit factorSharpeContrib. to return
Forex~579%2.712.18+38%
Futures~374%2.941.92+22%
Indices & ETFs~375%2.581.85+15%
Equities~372%2.811.78+12%
Crypto~266%2.341.42+8%
Polymarketbiwk68%2.121.21+5%
All markets~1775%2.682.10100%
VS BENCHMARK
APEX RIDGE vs SPY BUY-AND-HOLD · 2024-2026
/ does the methodology add alpha over passive?
26 MONTHS · LIVE

The fair comparison is the cheapest passive alternative: SPY buy-and-hold over the same period. Apex Ridge is reported on the all-grade portfolio (size-weighted by recommended risk budget) — pre-fee, pre-slippage, no leverage.

MetricApex Ridge (live)SPY buy-and-holdΔ vs SPY
Total return (Feb 2024 → Apr 2026)+182%+38%+144 pp
Annualised return+58%+15%+43 pp
Sharpe (annualised)2.100.94+1.16
Max drawdown12.0%17.4%−5.4 pp
Correlation to SPY0.311.00low corr · diversifying
Best month+19.2%+8.4%—
Worst month−4.1%−6.8%smaller losing month

Reading — the model returns ~3.8× SPY's compound, with a Sharpe over 2x higher and a smaller max drawdown. Most importantly, the 0.31 correlation to SPY means the model adds a return stream that doesn't move with the index — true diversification, not closet-beta.

DRIFT
BACKTEST → LIVE DELTA
/ per-metric drift attribution
4 METRICS

The drift between backtest and live is attributable. The largest contributor to higher live win rate is the 2024 macro overlay re-calibration after the rate-cut regime change; the largest contributor to lower live drawdown is tighter exclusion rules on equities around earnings (added in v1.3, March 2025).

MetricBacktestLiveΔAttribution
Win rate79.4%82.3%+2.9ppv1.3 macro re-cal
Sharpe1.852.10+0.25vol regime + grade discipline
Max DD15.2%12.0%−3.2ppv1.3 earnings filter
Profit factor2.412.68+0.27tighter A-grade threshold (v1.4)
08

RISK · POSITION-SIZING

/ how risk is bounded per position · per signal · per portfolio
— 3 LIMITS · PORTFOLIO LAYER —guidance, not execution

Apex Ridge publishes position blocks; it does not size them. Portfolio sizing is the subscriber's responsibility, but the methodology recommends three explicit risk limits that mirror the discipline applied in the operator's own book. They are guidance, not execution: the model cannot enforce them, but it is calibrated under their assumption.

— LIMIT 01 — PER-POSITION RISK

Per-position risk budget

1.0%OF PORTFOLIO PER GRADE-A POSITION

Recommended maximum loss per single position at the published stop. Grade B = 0.85%, C = 0.70%, D = 0.55%, E = 0.40%. Sizing is computed as risk-budget ÷ stop-distance.

— LIMIT 02 — CORRELATION CAP

Correlated-cluster cap

3.0%OF PORTFOLIO ON ANY CORRELATED CLUSTER

Sum of risk across positions in the same correlated cluster (e.g. USD-cross FX, large-cap US tech, oil-complex) capped at 3% — same as three independent A-grades. Cluster membership computed from rolling 90-session correlation > 0.5.

— LIMIT 03 — PORTFOLIO HEAT

Total portfolio heat

8.0%AGGREGATE OPEN RISK · HARD CEILING

Sum of all open per-position risk budgets capped at 8% of portfolio. New signals beyond this cap are deferred until existing positions close. Hard ceiling — never overridden, even at A-grade.

RISK · MODEL
CORRELATION CLUSTERS · CURRENT
/ which instruments are treated as one bet
5 CLUSTERS · ROLLING 90-SES

The correlation cap (Limit 02) is computed against the following clusters. Membership is re-tested every Sunday on rolling 90-session correlation. Any pair with ρ > 0.5 on the recent window is grouped.

ClusterMembers (current)Internal ρCombined cap
USD-major FXEUR/USD · GBP/USD · AUD/USD0.623.0%
JPY-cross FXUSD/JPY · GBP/JPY (off-universe)0.713.0%
US large-cap techAAPL · MSFT · GOOGL · AMZN · NVDA · META0.683.0%
US broad equitySPY · QQQ · DIA · IWM · XLE0.743.0%
Oil complexCL · XLE · USD/CAD (inverse)0.583.0%

Note — the cap is on aggregate risk, not aggregate notional. Three A-grades on EUR/USD, GBP/USD, AUD/USD with 1.0% per-position risk = 3.0% cluster heat = at the cap. A fourth signal in the same cluster is deferred.

WORKED
SIZING WORKED EXAMPLE
/ from Grade-A signal to position size
EUR/USD · CASE 01
StepQuantityValue
PortfolioAccount equity$100,000
GradePer-position risk budget1.0% = $1,000
EntryPublished level1.0875
StopPublished level1.0790
Stop distanceEntry − Stop · in pips85 pips
Pip valueEUR/USD standard lot$10 / pip / lot
Position sizeRisk ÷ (stop × pip value)$1,000 ÷ (85 × $10) = 1.18 lots

Cluster check — EUR/USD belongs to USD-major FX cluster. If GBP/USD or AUD/USD positions are open, total cluster risk must remain ≤ 3.0%. If at cap, the new EUR/USD signal is deferred, not down-sized.

09

LIMITS · KNOWN

/ where the model under-performs · documented failure modes
— 4 FAILURE MODES —documented

Apex Ridge is a trend-following + structural-support model. There are environments where the design is structurally weak. We document them rather than hide them. The exclusion filters drop these conditions where mechanically possible; the rest is communicated to subscribers in the daily session note.

— L01 —
Regime changes (early)

First 1-3 weeks of a new trend. The trend regime sub-system is a 30/200 EMA construction; it is structurally lagging. New trends generate quant scores that hover near the 0.55 threshold and either chop in and out of advancement or get downgraded to grade D/E.

Mitigation: grade D/E published with reduced size guidance · subscribers notified of regime-change risk in session note
— L02 —
Event-driven gaps

Earnings, central bank surprises, geopolitical shocks. Apex Ridge does not predict event outcomes. A position open into an event is exposed to gap risk that the structural-support stop cannot defend.

Mitigation: earnings 48h blackout · central-bank T-1 skip on macro pair · geopolitical halt = drop
— L03 —
Thin-liquidity instruments

Spread > 3× 30-day median. Wider spreads erode the structural-support edge — published entry can no longer be approximated by execution. Slippage attribution shows ~38% of expected edge lost when this trigger fires.

Mitigation: spread filter drops instrument from session · auto re-test next session
— L04 —
Crypto weekend gaps

BTC / ETH / SOL / BNB / XRP only. 24/7 tape means the 06:00 UTC session-window snapshot can be dislocated by Saturday-Sunday flow before Monday cross-asset session re-aligns. Session window mitigates but does not eliminate.

Mitigation: Monday signals on crypto are systematically grade-capped at C · subscribers notified
10

EVOLUTION · VERSION HISTORY

/ v1.0 → v1.4 · what changed and why
— 5 RELEASES · ~26 MONTHS —v1.4 current

The model is versioned. Every release ships with a changelog, a before/after performance comparison, and a documented motivation. Version is part of the position-block timestamp — subscribers can always tell which rule set produced a historical signal.

v1.0
FEB 2024
Initial release

Quant engine + macro overlay + A-E grade scaffolding. Ten initial instruments (FX + indices). Grade thresholds set from 2020-2023 backtest. Win rate ~79%.

v1.1
JUN 2024
Universe expansion

Added Futures (6 instruments) and Crypto (top-5 by mcap). Universe count 10 → 21. Macro overlay extended with crude/positioning factor for FX-commodity pairs.

v1.2
OCT 2024
Equities tier

Added 9 large-cap US equities with earnings-blackout filter (48h either side). Universe 21 → 30. Polymarket discretionary set added on biweekly cadence.

v1.3
MAR 2025
Macro re-calibration

Post rate-cut regime change. Macro overlay re-weighted: rate-cycle factor up, positioning factor down. Win rate +2.1pp; max DD −2.4pp. Cryptographic timestamp signing added at publish stage.

v1.4
FEB 2026
Grade discipline

A-grade threshold raised: TR ≥ 0.75 · MS ≥ 1.5 (from 0.65 / 1.3). Daily A-grade count ~9 → ~6. Profit factor on A signals 2.41 → 2.68. REST API + WebSocket institutional delivery channels added at publish stage.

11

ATTRIBUTION · OPERATOR

/ the discretionary half · macro overlay reviewed daily
— STAGE 02 —D. O'NEILL
— OPERATOR · STAGE 02 —

Darren O'Neill

Founder · Vector Ridge
YearReturnMax DDSharpe
2023178%14%2.57
202494%19%1.44
2025168%12%2.10
The quant engine is the easy half — it's mechanical, the rules are written down, anyone can read them. The macro overlay is where the discretionary edge lives. That's the half that's my work, every session, against the framework. Everything published carries my name on it because that stage carries my judgement.

O'Neill is the 2023 Trading World Champion (tradingworldchampion.com), with audited returns of 178% in 2023 (14% max drawdown · 2.57 Sharpe), 94% in 2024, and 168% in 2025. He took 1st place in the Oct 2025 Monthly Forex division of the WTC and finished 4th overall in the Annual Forex with a 168% return.

The macro overlay framework — five factors: rate cycle, growth/inflation, risk-on/off, positioning, event risk — is the same framework applied in his own discretionary book. Vector Ridge subscribers see the same overlay output O'Neill applies to his own positions, with the same timestamp.

12

GLOSSARY · TERMS

/ every jargon term used on this drawing · defined
— ~20 TERMS —plain definitions

If a term in this document is not standard market vocabulary, it is defined here. Items marked with an acronym tag are referenced in formula notation; others are referenced in prose.

Trend RegimeTR
Composite score on [-1, +1] combining EMA spread, ATR-normalised slope, and HH/LL count. Above |0.55| the instrument is considered to be in a confirmed trend; below it is considered chop.
Momentum SlowdownMS
Composite score reading short-window vs long-window momentum (RSI, ROC, ADX percentile bands). The model looks for cooling within the trend, not reversals.
Structural SupportSS
Swing low (uptrend) or swing high (downtrend) within a 60-session window confirmed by ≥2 prior touches within 0.25·ATR. The structural read invalidates if breached.
EMA spreadEMA₅₀ − EMA₂₀₀
Difference between the 50-session and 200-session exponential moving averages, normalised by ATR. Reads as trend strength + direction.
ATRATR₁₄
Average True Range over 14 sessions — Wilder (1978). Used as the natural volatility unit for normalisation across instruments and regimes.
RSI / ROC / ADXPCT BANDS
Momentum oscillators: Relative Strength Index, Rate Of Change, Average Directional Index. Used as percentile bands, not raw values, for cross-instrument comparability.
HH / LL countHH₃₀ − LL₃₀
Count of higher-highs minus lower-lows over 30 sessions. Positive = structurally bullish; negative = structurally bearish; zero = consolidation.
Precision dip
A pullback from price into the structural-support level within a confirmed trend. The model's central pattern — entry is taken at the support level on the dip.
Macro overlay
The discretionary stage. Each surviving quant signal is reviewed against five macro factors and rated Aligned · Neutral · Opposed. Reviewed daily by the operator.
Quad framework
The growth × inflation regime classifier (Q1: G+/I-, Q2: G+/I+, Q3: G-/I+, Q4: G-/I-). Drives the growth/inflation factor in the macro overlay.
CFTC positioning
Weekly Commitment of Traders report. Used as the positioning factor input — extreme positioning mean-reverts, neutral positioning is non-informative.
Grade A → E
Conviction band assigned at entry. A = highest (TR ≥ 0.75, MS ≥ 1.5, macro aligned-strong). E = lowest (TR ≥ 0.55, MS ≥ 1.2, macro marginal). Permanent.
Down-band
When the macro verdict is Neutral, the grade earned by the quant pass is reduced by one band before publication (e.g. quant-A becomes published-B).
Position block
The standardised six-field publication unit: instrument · direction · entry · target · grade · timestamp. Cryptographically signed at the publish stage.
Cryptographic timestamp
A signed hash recorded at the publish stage that proves the position was published at a specific moment, verifiable after the fact against any third-party tape.
Walk-forward
A backtest discipline that re-fits parameters on rolling in-sample windows and tests on the next out-of-sample window. Closer to live operation than a single static fit.
Pass-rate
The fraction of inputs that survive a given gate. The model has stage-level pass-rates (~73% at quant, ~80% at macro) and a session-level publish-rate of ~57% (17/30).
Calibration
The match between published grade and observed live hit-rate. A model is well-calibrated when A-grade hit-rate > B > C > D > E with monotone drop. See §03 calibration chart.
Profit factor
Sum of winning trades ÷ sum of losing trades. Above 2.0 is considered a strong edge. Apex Ridge live profit factor: 2.68 (2024-2026).
Sharpe ratio
Excess return ÷ return volatility, annualised. Above 2.0 is institutional-grade. Apex Ridge live Sharpe: 2.10 (2025).
13

REFERENCES · LINEAGE

/ intellectual provenance · what this work is built on
— 9 SOURCES —cited & acknowledged

The Swing Trade Model (internal designation: Apex Ridge Model™) is a synthesis, not an invention. Each component below is acknowledged: where the idea originated, what the canonical reference is, and how the implementation departs from or extends it.

[01]
Wilder, J. W. (1978). New Concepts in Technical Trading Systems. Trend Research. — Source of ATR, RSI, and ADX. Apex Ridge uses Wilder smoothing as the canonical method for all three; thresholds and percentile-banding are Apex Ridge's own.
[02]
Wyckoff, R. D. (c. 1930s). The Richard D. Wyckoff Method of Trading and Investing in Stocks. — Origin of the structural-support concept (cause-and-effect, accumulation/distribution). Apex Ridge formalises this as swing-low + ≥2 prior touches within 0.25·ATR.
[03]
Pardo, R. (2008). The Evaluation and Optimization of Trading Strategies (2nd ed.). Wiley. — Canonical reference for walk-forward analysis. Apex Ridge backtests are walk-forward on a rolling 36-month / 6-month split, 2020-2023, before going live in 2024.
[04]
Mandelbrot, B. (1963). The Variation of Certain Speculative Prices. Journal of Business, 36(4). — The work that established fat-tailed return distributions. Apex Ridge uses ATR-normalised stops, not σ-normalised stops, in deference to this finding.
[05]
Sharpe, W. F. (1966). Mutual Fund Performance. Journal of Business, 39(1). — The Sharpe ratio. Apex Ridge live Sharpe is reported on annualised daily returns (not weekly or monthly), per Lo (2002) recommendation.
[06]
Lo, A. W. (2002). The Statistics of Sharpe Ratios. Financial Analysts Journal, 58(4). — Standard reference for Sharpe-ratio confidence intervals and the bias from autocorrelated returns. Apex Ridge applies the Newey-West correction.
[07]
Hedgeye Risk Management. The Quad Framework. — Public influence on the growth/inflation factor of the macro overlay. Apex Ridge uses a Q1-Q4 classifier as one of five macro inputs; weights and thresholds are Apex Ridge's own.
[08]
Commodity Futures Trading Commission. Weekly Commitment of Traders Report. — Public source of the positioning factor. Apex Ridge ranks each instrument's net positioning at the 252-week percentile and reads extremes as mean-reverting.
[09]
O'Neill, D. (2024-2026). Vector Ridge live publication archive. — The only authoritative source for Apex Ridge published positions: cryptographically timestamped at session, immutable thereafter, reconciled annually by AuditedTrader.
14

QUESTIONS · COMMON

/ what subscribers and reviewers ask
— 10 ANSWERS —schema-registered
— Q01 —How does the Swing Trade Model decide which trades to publish?+
The Swing Trade Model runs a four-stage pipeline every session: a deterministic quant engine scores instruments on trend regime, momentum slowdown, and structural support; a discretionary macro overlay reviews each surviving signal against a five-factor macro framework; an A-E grade is assigned based on quant–macro alignment; and the standardised six-field position block is published with a cryptographic timestamp.
— Q02 —What is the A-E grade system?+
Grades A through E rate conviction at entry. Grade A requires Trend Regime score ≥ 0.75, Momentum Slowdown score ≥ 1.5, and an aligned macro view. Grade E is the lowest band where the macro view is marginal. Grade is permanent once assigned and does not change in flight.
— Q03 —How many signals are published each session?+
Approximately 17 signals per session out of 30 instruments scanned. Roughly 8 are dropped at the quant stage as chop or no-momentum, and roughly 5 are dropped at the macro stage as opposed-to-thesis. The remaining 17 are graded A through E and published at 07:30 UTC.
— Q04 —Is the model fully automated?+
No. Stage 01 (the quant engine) is fully deterministic; the same input produces the same output. Stage 02 (macro overlay) is discretionary — reviewed daily by Darren O'Neill against the five-factor macro framework. Apex Ridge is a quant–discretionary hybrid, not a pure black box.
— Q05 —What is the daily session timing?+
The session starts at 06:00 UTC with data ingest, quant scoring at 06:05, macro overlay at 06:25, grade assignment at 06:55, and publication at 07:30 UTC. Total session length is approximately 90 minutes.
— Q06 —Has the model been backtested?+
Yes. Apex Ridge was walk-forward backtested across 2020-2023 before going live in 2024. Live performance has tracked the backtest within 4 percentage points on win rate and within 0.3 on Sharpe — the methodology has held up out-of-sample. See §05 for the full drift attribution.
— Q07 —What is the Apex Ridge model NOT good at?+
Apex Ridge is a trend-following + structural-support model. It under-performs in regime changes (first 1-3 weeks of a new trend), in event-driven gaps (earnings, central bank surprises), and on instruments with thin liquidity or wide spreads. The exclusion filters drop these conditions where possible — see §09 for the full failure-mode log.
— Q08 —How is position size determined?+
The model publishes position blocks, not sizes. The recommended methodology sizes per-position risk by grade: A = 1.0% of portfolio, B = 0.85%, C = 0.70%, D = 0.55%, E = 0.40%. Position size = risk-budget ÷ (entry − stop) ÷ pip-value. A correlated-cluster cap of 3.0% and a portfolio-heat ceiling of 8.0% are also recommended. See §08 Risk for the worked example.
— Q09 —How does the model handle correlated positions?+
Instruments are grouped into correlation clusters (re-tested weekly on rolling 90-session correlation > 0.5): USD-major FX, JPY-cross FX, US large-cap tech, US broad equity, and the oil complex. Total open risk inside any cluster is capped at 3.0% of portfolio — the equivalent of three independent A-grades. New signals beyond cluster cap are deferred, not down-sized.
— Q10 —How can I verify a published position after the fact?+
Every position block carries a cryptographic timestamp signed at the publish stage (07:30 UTC). The signature can be verified against any third-party tape — instrument, direction, entry, target, grade, and timestamp must all match. Annual reconciliation is performed by AuditedTrader against the public price tape, and the published archive is immutable (no retroactive edits).
— READY · TRY APEX RIDGE —

The methodology is on the page.
Now read the live tape.

Every paragraph above is what you'll see in the live signal stream. Same engine, same overlay, same grades. Free trial includes the All Models bundle for seven days; the 240-page methodology book is free to anyone, no signup.

Start 7-Day Free Trial → See Today's Signals Apex Ridge Live Page Download Free Book
7-DAY FREE TRIAL · $50/MONTH ALL SIGNALS & RESEARCH
VR-METH-001 · APEX RIDGE · v1.4 · DRAWING REV 2026.04.30 · 14 SHEETS · 1 FIG · 3 CHARTS SCALE 1:1 · UNITS SI · STATUS LIVE
↑