Auditable quant core · adaptive outcome learner

Research-desk discipline.Machine-speed selectivity.

Northstar applies systematic-desk discipline to OKX cross-asset perpetuals: live finalized data, symmetric long/short rules, transparent factor weights, spread vetoes and risk-defined levels—with the evidence attached.

Read the research Audit the model

Up to 450 finalized bars200 book levels per sideNineteen score componentsNo black-box execution

Quant signal brief · illustrative

Format preview

Eligible instrument · 1H / 4H / 1D

Qualified swing brief

Decision record

Stored

Regime model

Causal HMM

MTF stack

1H / 4H / 1D

Book depth

5 / 25 / 200

Risk rule

Hard-gated

Regime

Observed

Liquidity

Checked

Outcome

Unresolved

Interface-format preview—not market data, a forecast or a live recommendation. Public live briefs appear below only when a stored signal passes every gate.

Northstar Learning

Research methods,explained without the hype.

Original notes on how quantitative evidence is designed, tested and governed. Each permanent article page includes authorship, sources, limitations and the automation disclosure.

Research note 04 · Live market anatomy

When leverage becomes the seller: reading a forced-deleveraging move.

A violent price move can contain both new information and mechanical position closure. The research problem is not to label every sell-off a liquidation cascade, but to test whether leverage, liquidity and price changed together in a way that alters continuation risk.

7 minute read09 Sep 20263 sources

Read the full research note

Research note 03 · Quant method explained

When a reversal is just the spread: testing the price behind the signal.

A short-term reversal can look predictable while remaining impossible to capture. Before improving the forecast, investigate what the recorded price actually represents.

6 minute read07 Sep 20262 sources

Read the full research note

Research note 02 · Research validation

The search is part of the backtest: how quants control false discoveries.

A rigorous model-selection process records every attempted strategy—not just the winner—because repeated searching can turn ordinary noise into an impressive result.

9 minute read04 Sep 20263 sources

Read the full research note

Research note 01 · Model engineering

Quant algorithm engineering: from hypothesis to auditable signal.

A practical look at how a trading idea becomes a controlled, testable and continuously monitored decision system.

7 minute read04 Sep 20263 sources

Read the full research note

General educational information—not investment advice. Automated publication is disclosed; named sources do not imply endorsement.

Browse the learning library

Public research desk

The thesis is public. The trade geometry is private.

These briefs are written from qualified, delivered OKX signals. Visitors can audit the technical regime and derivatives context; exact entry, invalidation and targets remain reserved for paid members.

“Fundamental” context here means observable market structure—open interest, funding and liquidity. Corporate earnings, balance sheets and macro releases are not yet inputs to the engine.

Oct 19, 2026, 3:13 AM UTC

OPENAI-USDT-SWAP

1h/4h/1d · Stock/ETF derivative multi-timeframe swing continuation

▼ SHORT

78/100 evidence

Technical thesis

1D bearish regime and 4H structure align while the 1H trigger is a controlled pullback. The causal HMM classifies the regime as bear (81% state confidence); directional book imbalance is 37%/27%/3% across 5/25/200 levels, microprice pressure is 0.66 bps, open interest expanded 0.0%, funding is 0.0000%, and short/long realized-volatility ratio is 1.04×.

HMM regime: BEAR · 88%
Horizon agreement: +99% / +0%
Book 5/25/200: +37% / +27% / +3%
Microprice: +0.66 bps
Open interest: +0%
Funding: 0.0000%
Volatility state: 1.04× short/long
Execution quality: 1.33 bps spread

Portfolio-aware selection

Signal-book selection: 12 exposures compared; maximum directional correlation 0.43.
Rolling market context: 178/445 measured contracts showed low absolute overlap across all three estimates. Historical price returns, not guaranteed strategy returns.

Recorded at selection. This research signal book is not a subscriber’s actual portfolio.

Adaptive context · shadow research

No funding crowding flag.
Regime uncertainty 26/100 (entropy, not failure odds).
No peer candidates in this scan; diversification unknown.

Recorded at generation, not a live feed. Experimental observations do not change this signal’s score or qualification.

Leading evidence

multi-timeframe trendsetup trenddaily regimebreakout structure

Paid execution layer

Entry reference, stop/invalidation and all three targets are intentionally redacted from this public brief.

Unlock the complete signal

Live OKX record · V4.1.0 · general market information only

Oct 18, 2026, 7:35 AM UTC

ANTHROPIC-USDT-SWAP

1h/4h/1d · Stock/ETF derivative multi-timeframe swing continuation

▼ SHORT

74/100 evidence

Technical thesis

1D bearish regime and 4H structure align while the 1H trigger is a controlled pullback. The causal HMM classifies the regime as bear (49% state confidence); directional book imbalance is -87%/-64%/4% across 5/25/200 levels, microprice pressure is -0.26 bps, open interest expanded 0.0%, funding is 0.0000%, and short/long realized-volatility ratio is 2.00×.

HMM regime: BEAR · 50%
Horizon agreement: +100% / +7%
Book 5/25/200: -87% / -64% / +4%
Microprice: -0.26 bps
Open interest: +0%
Funding: 0.0000%
Volatility state: 2.00× short/long
Execution quality: 0.52 bps spread

Portfolio-aware selection

Signal-book selection: 15 exposures compared; maximum directional correlation 0.61.
Rolling market context: 290/447 measured contracts showed low absolute overlap across all three estimates. Historical price returns, not guaranteed strategy returns.

Recorded at selection. This research signal book is not a subscriber’s actual portfolio.

Adaptive context · shadow research

No funding crowding flag.
Regime uncertainty 77/100 (entropy, not failure odds).
No peer candidates in this scan; diversification unknown.

Recorded at generation, not a live feed. Experimental observations do not change this signal’s score or qualification.

Leading evidence

multi-timeframe trendsetup trenddaily regimeregime persistence

Paid execution layer

Entry reference, stop/invalidation and all three targets are intentionally redacted from this public brief.

Unlock the complete signal

Live OKX record · V4.1.0 · general market information only

Oct 18, 2026, 7:18 AM UTC

WCT-USDT-SWAP

1h/4h/1d · Crypto perpetual multi-timeframe swing continuation

▼ SHORT

79/100 evidence

Technical thesis

1D bearish regime and 4H structure align while the 1H trigger is supportive. The causal HMM classifies the regime as bear (44% state confidence); directional book imbalance is 72%/12%/-3% across 5/25/200 levels, microprice pressure is 1.23 bps, open interest contracted 0.0%, funding is 0.0002%, and short/long realized-volatility ratio is 1.31×.

HMM regime: BEAR · 58%
Horizon agreement: +94% / +7%
Book 5/25/200: +72% / +12% / -3%
Microprice: +1.23 bps
Open interest: +0%
Funding: 0.0002%
Volatility state: 1.31× short/long
Execution quality: 2.83 bps spread

Portfolio-aware selection

Signal-book selection: 14 exposures compared; maximum directional correlation 0.71.
Rolling market context: 157/440 measured contracts showed low absolute overlap across all three estimates. Historical price returns, not guaranteed strategy returns.

Recorded at selection. This research signal book is not a subscriber’s actual portfolio.

Adaptive context · shadow research

No funding crowding flag.
Regime uncertainty 56/100 (entropy, not failure odds).
No peer candidates in this scan; diversification unknown.

Recorded at generation, not a live feed. Experimental observations do not change this signal’s score or qualification.

Leading evidence

multi-timeframe trendsetup trendbreakout structuredaily regime

Paid execution layer

Entry reference, stop/invalidation and all three targets are intentionally redacted from this public brief.

Unlock the complete signal

Live OKX record · V4.1.0 · general market information only

Institutional-style process

Deterministic at the core. Adaptive at the edge.

Every setup still passes reproducible hard gates. A deployed online learner now scores accepted setups against their later outcomes in shadow mode; it cannot alter eligibility, levels or risk controls, and it cannot go active until live chronological calibration earns manual promotion.

Core decision · deterministicML role · outcome calibration / shadow

450
Maximum finalized bars / horizon: 46
Versioned adaptive features: 19
Weighted score contributors: 04
Qualification pipeline stages

P | Z E V A R + < F W B S - > G X C T 0 N H Y D U 1 P J Z E V A R K < F W B S - > G X C T 0 / H Y D U 1 \ J Z E V A R K < F W B S L > G X C T M / H Y D U 1 \ J Z E V A | K < F W B + L > G X C T M / H Y D U N \ J Z E V P | K < F W B + L > GP | Z E V A R + < F W B S - > G X C T 0 N H Y D U 1 P J Z E V A R K < F W B S - > G X C T 0 / H Y D U 1 \ J Z E V A R K < F W B S L > G X C T M / H Y D U 1 \ J Z E V A | K < F W B + L > G X C T M / H Y D U N \ J Z E V P | K < F W B + L > GT 1 R - N J < G Y E A S 0 P K > H Z F B T 1 R - / J < G Y U A S 0 \ K > H Z V B T 1 R L / J < G C U A S M \ K > H D V B T 1 | L / J < W C U A + M \ K > X D V B T N | L / J E W C U P + M \ K F X D V B - N | L / Y E W C 0 P + M \ Z F X D V RT 1 R - N J < G Y E A S 0 P K > H Z F B T 1 R - / J < G Y U A S 0 \ K > H Z V B T 1 R L / J < G C U A S M \ K > H D V B T 1 | L / J < W C U A + M \ K > X D V B T N | L / J E W C U P + M \ K F X D V B - N | L / Y E W C 0 P + M \ Z F X D V RG Z G D W D W T A T A - P - \ L \ L H < H < X E X E B U B U R 0 R M | M | > J > J F Y F Y V C V C 1 S 1 + N + N K / K / Z G Z G D W D W T A T P - P - \ L \ L H < H < X E X E B U B 0 R 0 R M | M | > J > J F Y F Y V C V S 1 S 1 + N + N K / KG Z G D W D W T A T A - P - \ L \ L H < H < X E X E B U B U R 0 R M | M | > J > J F Y F Y V C V C 1 S 1 + N + N K / K / Z G Z G D W D W T A T P - P - \ L \ L H < H < X E X E B U B 0 R 0 R M | M | > J > J F Y F Y V C V S 1 S 1 + N + N K / K+ P M + P M + P M + P M + P M + P 0 + P 0 + P 0 + P 0 + P 0 + P 0 + P 0 S P 0 S P 0 S P 0 S P 0 S P 0 S P 0 S A 0 S A 0 S A 0 S A 0 S A 0 S A 0 S A U S A U S A U S A U S A U S A U S A U C A U C A U C A U C A U C A U C A U C W U C W U C W U+ P M + P M + P M + P M + P M + P 0 + P 0 + P 0 + P 0 + P 0 + P 0 + P 0 S P 0 S P 0 S P 0 S P 0 S P 0 S P 0 S A 0 S A 0 S A 0 S A 0 S A 0 S A 0 S A U S A U S A U S A U S A U S A U S A U C A U C A U C A U C A U C A U C A U C W U C W U C W UV U D Y X G F < K J \ N M - + R A 1 U D C X G F < Z J \ / M - + R P 1 U T C X W F < Z J H / M L + R P 1 0 T C B W F E Z J H / > L + | P 1 0 T S B W V E Z Y H / > L K | P N 0 T S B A V E D Y H G > L K | \ N 0 - S B A V U D Y X G > < K | \ NV U D Y X G F < K J \ N M - + R A 1 U D C X G F < Z J \ / M - + R P 1 U T C X W F < Z J H / M L + R P 1 0 T C B W F E Z J H / > L + | P 1 0 T S B W V E Z Y H / > L K | P N 0 T S B A V E D Y H G > L K | \ N 0 - S B A V U D Y X G > < K | \ NY J | | R R B B X H H \ \ P P A W W G G / N N 1 1 V V F > > M M 0 0 U E E < < L - - T T D D Z K K + + S S C Y Y J J | R R B B X X H \ \ P P A A W G G / / N 1 1 V V F F > M M 0 0 U U E < < L L - T T D D Z Z K + + S S C C Y J J | | R B B X XY J | | R R B B X H H \ \ P P A W W G G / N N 1 1 V V F > > M M 0 0 U E E < < L - - T T D D Z K K + + S S C Y Y J J | R R B B X X H \ \ P P A A W G G / / N 1 1 V V F F > M M 0 0 U U E < < L L - T T D D Z Z K + + S S C C Y J J | | R B B X XM 1 A X Y K L 0 1 W X J K - 0 V G H | + T U F G \ | S T E F / \ R S D < > N P B C Z < M N A B Y Z L M 1 A X J K - 0 V W H J + - U V G H | + T U F / \ R S D E > / P R C D < > N P B C Z L M 1 A X Y K L 0 1 W X J K - 0 V W H | + T U F G \ | SM 1 A X Y K L 0 1 W X J K - 0 V G H | + T U F G \ | S T E F / \ R S D < > N P B C Z < M N A B Y Z L M 1 A X J K - 0 V W H J + - U V G H | + T U F / \ R S D E > / P R C D < > N P B C Z L M 1 A X Y K L 0 1 W X J K - 0 V W H | + T U F G \ | SB D > \ S U G J - 1 X Z M P S E G | - V X K M A C < / R T F H + 0 A Y < N R D F \ + U W J L 1 B Z > P S U G J - 1 X Z M P C E / | T V H K 0 A C < / R T F H + 0 W Y L N B D > \ S U W J L 1 B Z > P S E G | - V X K M A C E / | T V H K 0 A Y \ S U G J - 1 X Z M P S E G | - V X K M A C < / R T F H + 0 A Y < N R D F \ + U W J L 1 B Z > P S U G J - 1 X Z M P C E / | T V H K 0 A C < / R T F H + 0 W Y L N B D > \ S U W J L 1 B Z > P S E G | - V X K M A C E / | T V H K 0 A Y << \ - W Z N S F J M B E \ - A Z / S V J M R E H - A Z / + V Y M R U H L A D / + 1 Y > R U H L P D G + 1 C > | U X L P T G K 1 C > | 0 X R U H L P D G + 1 C > | U X L P T G K 1 C > | 0 X + W < | V Z P U Y N T X M S W L R F K P E J N D H 0 C G - B F + A E | 1 Z \ 0 Y / - X > S W < R V Z P U Y N T H M S G L R F K A E J 1 D H 0 C G - B > + A < | 1 Z \\ U Y / T X M S W L R V K P U J N D H M C G L B F + A E | 1 D \ 0 C / - X > + W < | V Z P U Y N T X M S W L R F K P E J N D H 0 C G - B F + A E | 1 Z \ 0 Y / - X > S W < R V Z P U Y N T H M S G L R F K A E J 1 D H 0 C G - B > + A < | 1 Z \T Y P V < S X / U Z | W > - C \ 1 E + B G 0 D J A F L C H N E K R W M T Y P V < S X / U Z | W > - Y \ 1 E + B G 0 D J A F L C H N E K R G M T Y P V < S X / U Z | W > - Y \ 1 < + B G 0 D J A F L C H N E K R G M T J P V < S X / U Z | W > - YT Y P V < S X / U Z | W > - C \ 1 E + B G 0 D J A F L C H N E K R W M T Y P V < S X / U Z | W > - Y \ 1 E + B G 0 D J A F L C H N E K R G M T Y P V < S X / U Z | W > - Y \ 1 < + B G 0 D J A F L C H N E K R G M T J P V < S X / U Z | W > - YG N V M U < - Z + C J B H P W N V > U < - D + C J R H P W / V > 0 E - D K C J R X P W / 1 > 0 E L D K S Y R X \ W / 1 F 0 E L T K S Y | X \ A G 1 F M E L T Z S Y | B \ A G N F M U < T Z + Y | B H A G N V M U < - Z + C J B H P G N V > U < -G N V M U < - Z + C J B H P W N V > U < - D + C J R H P W / V > 0 E - D K C J R X P W / 1 > 0 E L D K S Y R X \ W / 1 F 0 E L T K S Y | X \ A G 1 F M E L T Z S Y | B \ A G N F M U < T Z + Y | B H A G N V M U < - Z + C J B H P G N V > U < -+ D L U > V / A H R J S Z - E 0 F N W P X | C K D L U > 1 / A H R Y S Z - E 0 F N W \ X | C K T L U > 1 G A H R Y S Z - E M F N W \ B | C K T 1 G A H R Y + Z - E M V N W \ B J C K T 1 G P H R Y + D - E M V / W \ B J C K T < 0 >+ D L U > V / A H R J S Z - E 0 F N W P X | C K D L U > 1 / A H R Y S Z - E 0 F N W \ X | C K T L U > 1 G A H R Y S Z - E M F N W \ B | C K T 1 G A H R Y + Z - E M V N W \ B J C K T 1 G P H R Y + D - E M V / W \ B J C K T < 0 >1 H C L V \ C L F P Y - F P Y T > A J T > W | D M W | Z 0 G R Z 0 G B K U / B K E N X + E N H S < 1 H S < V \ C L V \ Y - F P Y - > A J T > A J D M W | D M G R Z 0 G R K U / B K U / X + E N X + < 1 H S < 1 \ C L V \ C L F P Y - F P J T > A1 H C L V \ C L F P Y - F P Y T > A J T > W | D M W | Z 0 G R Z 0 G B K U / B K E N X + E N H S < 1 H S < V \ C L V \ Y - F P Y - > A J T > A J D M W | D M G R Z 0 G R K U / B K U / X + E N X + < 1 H S < 1 \ C L V \ C L F P Y - F P J T > AY 0 H T / Y 0 H T / C M X - G C M X - G S > B L G S > B L W + F R < W + F R < A K V | < A K V | E P Z 1 J E P Z 1 J U \ D N J U \ D N Y 0 H T / Y 0 H T / C M X - / C M X - G S > B L G S > B L W + F R L W + F R < A K V | < A K V | E P Z 1 |Y 0 H T / Y 0 H T / C M X - G C M X - G S > B L G S > B L W + F R < W + F R < A K V | < A K V | E P Z 1 J E P Z 1 J U \ D N J U \ D N Y 0 H T / Y 0 H T / C M X - / C M X - G S > B L G S > B L W + F R L W + F R < A K V | < A K V | E P Z 1 |M Y 1 K W - H U | F S / D P < X 0 J V + G T \ E R > C N Z A L X 0 J F S / D P C N Z A L H U | F S / D P C N Z A L X 0 J V + G T \ E R > C N K W - H U | F S / DM Y 1 K W - H U | F S / D P < X 0 J V + G T \ E R > C N Z A L X 0 J F S / D P C N Z A L H U | F S / D P C N Z A L X 0 J V + G T \ E R > C N K W - H U | F S / DR / E S \ F T | G U + H V - J W 0 K X 1 L Y A M Z B N < C P > D R / E S \ F T | G U + H V - J W 0 K X 1 L Y A M Z B N < C P > D R / E S \ F T | G U + H V - J W 0 K X 1 L Y A M Z B N < C P > D R / E S \ F T | G U + H V - J W 0 K X 1 L Y A MR / E S \ F T | G U + H V - J W 0 K X 1 L Y A M Z B N < C P > D R / E S \ F T | G U + H V - J W 0 K X 1 L Y A M Z B N < C P > D R / E S \ F T | G U + H V - J W 0 K X 1 L Y A M Z B N < C P > D R / E S \ F T | G U + H V - J W 0 K X 1 L Y A ME T + J X A N > E T + J X A / F U - K Y B P / F U - K Y B P / F U L Z C R \ G V 0 L Z C R \ G V 0 L Z C | H W 1 M < D S | H W 1 M < D S | H W N > E T + J X A N > E T + J X A N > E - K Y B P / F U - K Y B P / F U - K Y R \ G V 0 L Z C R \ GE T + J X A N > E T + J X A / F U - K Y B P / F U - K Y B P / F U L Z C R \ G V 0 L Z C R \ G V 0 L Z C | H W 1 M < D S | H W 1 M < D S | H W N > E T + J X A N > E T + J X A N > E - K Y B P / F U - K Y B P / F U - K Y R \ G V 0 L Z C R \ G\ H X B R J Y C S + K Z D T L < E U 0 M > F V 1 / G W A P \ H X B | J Y C S + K Z D T L < E U 0 M > F V N / G W A P \ H X B | J Y C S + K Z D - L < E U 0 M > F V N / G W A P \ H X R | J Y C S + K Z D - L < E U 0 M > F 1 N / G W A P \ H X R\ H X B R J Y C S + K Z D T L < E U 0 M > F V 1 / G W A P \ H X B | J Y C S + K Z D T L < E U 0 M > F V N / G W A P \ H X B | J Y C S + K Z D - L < E U 0 M > F V N / G W A P \ H X R | J Y C S + K Z D - L < E U 0 M > F 1 N / G W A P \ H X RT 0 / H Y D U 1 \ J Z E V A | K < F W B S L > G X C T M / H Y D U N \ J Z E V A | K < F W B + L > G X C - M / H Y D U N \ J Z E V P | K < F W R + L > G X C - M / H Y D 0 N \ J Z E 1 P | K < F W R + L > G X S - M / H Y T 0 N \ J Z E 1 P | KT 0 / H Y D U 1 \ J Z E V A | K < F W B S L > G X C T M / H Y D U N \ J Z E V A | K < F W B + L > G X C - M / H Y D U N \ J Z E V P | K < F W R + L > G X C - M / H Y D 0 N \ J Z E 1 P | K < F W R + L > G X S - M / H Y T 0 N \ J Z E 1 P | K

Qualification engine

Every hard gate must agree.

The system can reject a setup at any stage. Only aligned evidence, acceptable execution conditions and a locked audit record produce a subscriber alert.

Sequential gates: 04
Silent revisions: 00
Locked record: 01

Inspect the qualification model

Demonstration telemetry · not a live recommendation

Model pipeline / demonstration

BTCUSDT1H / 4H / 1DIllustrative

01Hard gate

Timeframe consensus

The 1D regime, 4H setup structure and 1H trigger must resolve into one side-aware swing thesis.

Pass

1D regime: Aligned
4H setup: Qualified
1H trigger: Supportive

02Evidence gate

Participation confirmation

Volume and derivatives participation test whether the move has enough support to qualify.

Pass

Rel. volume: 1.34×
OI delta: +2.8%
Flow: Supportive

03Blocker gate

Data and spread veto

Invalid market data, an invalid ATR or an excessive spread block publication before levels are generated.

Pass

Market data: Valid
Spread: 3.2 bps
ATR: Valid

04Audit gate

Publication lock

The signal fingerprint, original levels and delivery record are fixed before the alert leaves the system.

Ready

Signal ID: NS-7F21
Levels: Frozen
Delivery: Queued

Audit state

Original levels preserved · fingerprint recorded · delivery traceable

4 / 4 gates passed

Publisher transparency

Independent research, with the process in view.

Northstar Signals separates public education from paid signal access. Research notes explain methods, cite relevant primary work where available, disclose automation and limitations, and keep commercial relationships visible at the point they matter.

About the publisher Editorial standards Contact

Decision intelligence, packaged

The signal brief a systematic desk would want.

Every accepted setup arrives as the same compact, reviewable decision package—enough context to assess the trade without surrendering control of your account.

Side-symmetric setup classification
Entry reference and expiry
Hard invalidation level
Three R-based targets
Explainable evidence score
HMM state and transition risk
5/25/200-level liquidity and microprice
OI, funding, spread and data freshness
Model version, drivers and watch factors

Inbox preview10:02 UTC

QUALIFIED SIGNAL · MEMBER FORMAT

Multi-timeframe swing continuation · 1H / 4H / 1D

Entry referencePaid members only

InvalidationPaid members only

Target 1–3Paid members only

Signal expiryPaid members only

HMM regime · state probability and transition risk

Horizon agreement · deterministic and HMM evidence

Liquidity · 5/25/200-level book measurements

Positioning · open interest and funding context

Execution · spread and source-data freshness

Execution venue directory

Choose the exchange yourself.

Northstar never connects to your account or routes orders. Each button opens a fixed, server-validated destination, and every commercial relationship is labeled at the point of click. Signal generation continues to use OKX public production data only.

BNPaid referral

Binance

Global derivatives venue with broad perpetual-market coverage.

View exchange

Northstar may earn a commission if you open an account or trade through this link.

BYPaid referral

Bybit

Derivatives-focused venue with unified account tooling.

View exchange

Northstar may earn a commission if you open an account or trade through this link.

OKPaid referral

OKX

Multi-product venue supporting perpetuals and portfolio tools.

View exchange

Northstar may earn a commission if you open an account or trade through this link.

BGOfficial link

Bitget

Futures venue with web and mobile execution interfaces.

View exchange

Commercial disclosure: Links marked “Paid referral” are compensated links. If you follow one and later open an account or trade, the exchange may pay Northstar Signals a commission. This creates a financial incentive to feature the exchange.

Venue risk: Exchange availability, products and eligibility vary by jurisdiction. Futures, tokenized exposure and cryptoasset trading are high risk; independently review the venue, product structure, its terms and local restrictions before opening an account. OKX stock/ETF and commodity products are derivatives or tokenized exposure—not direct ownership of shares or physical commodities.

Independent research desk access

Signal access, native to Telegram.

Subscribe inside Telegram and receive qualified Northstar signal posts directly in the private Pro channel.

Telegram-native access

Northstar Signals Pro channel

Receive qualified signal posts inside the paid Telegram channel for 3,550 Stars every 30 days. Telegram handles access and recurring Stars billing.

Channel-only access. It does not include the website member desk, email delivery, or private bot messages.

Subscribe in Telegram

Subscription provides general market information, not personalized investment advice or account management. No performance is guaranteed. Confirm whether this service is permitted in your jurisdiction before subscribing.

Free verified weekly brief

See how we think before you subscribe.

A double-opt-in weekly note covering qualified live setups, resolved outcomes and adaptive-model governance. No paid signal levels.

Good questions, answered plainly.

Does this place trades for me?

No. Northstar never asks for exchange credentials and does not place orders. You decide whether and how to use each signal.

How often are signals sent?

The engine evaluates configured markets after finalized bars, but sends only when every hard gate and the minimum quality threshold pass. Some periods may produce no signals.

Is Northstar an AI model?

It uses a hybrid design: deterministic qualification plus an online FTRL outcome learner. The learner begins in shadow mode, trains only on causally resolved live signals, and never overrides hard gates or published risk levels.

Is the quality score a win probability?

No. It is a normalized evidence score. A separate learned score is shown only after at least 250 live resolved outcomes, class-balance, calibration, drift and baseline-improvement checks pass.

What markets does it cover?

The default scope includes crypto, stocks & ETFs, commodities, forex and bonds where OKX lists eligible live USDT perpetuals. The Markets page shows the current configured catalogue by official category, including empty categories. Bounded rotating batches use a 1H trigger, 4H setup and UTC-aligned 1D regime. Listing does not imply a qualified signal: insufficient history or failed quality checks still reject the instrument.