Auto-execution and subscriptions are closed to the public. The algorithms below currently run only for funds and managed accounts. Everything else on this page is unchanged: the strategies, the trade log, the execution stack. Funds: contact us →

Not a signal group - a quant desk that builds, tests, and kills its own algorithms.

100+ hypotheses tested, most killed - read the research log →

Four strategies.
Managed capital only.

Four uncorrelated algorithms watch 250 Binance markets around the clock. Since 15 October 2026 they execute only for funds and managed accounts: public auto-execution and subscriptions are closed while the desk rebuilds its edge research. The strategies, the unedited trade log and the execution stack below stay public.

For funds: contact the desk →Open the terminal →
Verified backtest Self-custody Read-only optional
OPERATIONS · LIVE
250 MARKETS 4 ALGOS <3s EVENT→ORDER --:--:-- UTC
MARKET DATAprice · volume · derivatives FILTERS0 rejected SIGNAL ENGINESNEVA · PHOENIX · CATALYST · VENUE EXECUTEyour account · hard stop
NEVA PHOENIX CATALYST VENUE
EXECUTION LOG NEVA · 30-day window
loading live log…
- trades · -W/-L · WR -%
EQUITY · 30-DAY WINDOW · MODEL ACCOUNT - - full history ↓
HISTORICAL MONTHLY RANGES NEVA +8-70% PHOENIX +5-70% CATALYST +5-30% VENUE +15-80%few trades/mo past performance ≠ guarantee
Every line is a real recorded trade - wins and stops, unedited. Win rate is ~51%, but winners run far past the −10% stop (PLAY +71%, LAB +30%): the edge is asymmetry, not hit-rate.

Four non-overlapping edges, each independently validated.

Universe
Binance Futures · USDT-M
$50M-$300M 24h volume
NEVA
Asymmetric
compression
LONG · medium-frequency · 8h hold
ClassAsymmetric long. Identifies a small number of coins per week matching our proprietary, undisclosed conditions.
ThesisA secret in-house model that acts on structural conditions our research identified. What it looks for, and why it works, is proprietary and deliberately undisclosed. There is no off-the-shelf analogue.
DisclosureNEVA is the firm's proprietary in-house quant + AI research. Inputs, model, weights, conviction logic - all kept private by design. Subscribers see entries and exits, not the recipe. This is the engine we will not open.
CATALYST
Event-driven
fade
SHORT · 4h hold · sub-3s latency
ClassPublic-event short. The catalyst is on the official Binance announcement page - the only question is who reacts first.
ThesisWhen Binance flags a coin for monitoring, the path is statistically one direction. The trade is not finding the edge - it is arriving at it before the human clicks. CATALYST is the algorithm we are open about, because the edge is execution, not secrecy.
Our edgeDirect CMS polling and a pre-warmed Binance API session. From announcement to filled position in under three seconds - faster than a Telegram channel can repost it. We do not predict; we react at machine speed.
VENUE
Liquidity-event
reversion
SHORT · 4h hold · post-pump
ClassLiquidity-event short. Trades the structural overhang that follows almost every memecoin listing on a major venue.
ThesisNew listings carry an unavoidable seller cohort - airdrop recipients, early allocators, locked-token holders - meeting retail FOMO into thin liquidity. The peak is a setup, not a market.
DisclosureEligibility filters and the exhaustion-detection logic are not published. What we will say: we pass on most listings. We only enter when the structure confirms.
PHOENIX
Proprietary
reversion
LONG · event-driven · uncorrelated to NEVA
ClassLiquidity-event long. Trades the structural exhaustion that follows forced-selling events on mid-cap futures pairs.
ThesisA proprietary in-house model that takes the other side of specific, mechanical market conditions, the class of engine leading market makers run internally, with no off-the-shelf analogue. Which events qualify, at what size, and when: that is the part we keep.
DisclosureThe exact size / volume / venue thresholds are firm IP. Public: 90-day OOS validation showed strong mean-reversion behaviour across all tier groups. Live execution is gated by per-user opt-in.

Equity curve from a $1,000 base.

$1,000 base
5% margin per trade
Slippage + fees + funding
March 2026
April 2026
May 2026
June 2026 · MTD
Final balance · May 2026 (live)
$1,542.62
Return
+54.26%
Trades
65
Win rate
41.5%
Max DD
−2.10%
May 01 May 15 Jun 01

Every closed position from the verified 30-day window.

All 39 closed trades
30-day NEVA window
Strategy
Side
Symbol
Entry → Exit
Hold
Net %
PnL

What the model gets right - asymmetric upside.

Method
Path-dependent replay
1m kline resolution · slippage + fees applied

NEVA is the firm's proprietary in-house quant + AI research. We don't publish how it scores - only what it produced. Three of NEVA's strongest live trades from the last 30 days on a real Binance Futures account, including a winner from this week.

币安人生 · MAY 31, 2026
+1,004%
What happened: NEVA flagged 币安人生 (Binance Life) in the morning before a parabolic memecoin pump. The position rode the full leg without an SL touch.
Why we publish only the result: the pre-trade fingerprint that pre-empts this kind of asymmetric move is firm IP. Subscribers received the signal in real time.
Sidelong
Statusclosed · live
Net result+1,004%
Realised P&L+$4,564
PLAY · MAY 31, 2026
+71.0%
What happened: PLAY moved into NEVA's candidate pool during a quiet Saturday session. The model's conviction crossed the entry gate, the breakout followed.
Why we publish only the result: the pre-trade signature that makes PLAY a candidate is the model's edge. The trade is the proof.
Sidelong
Statusclosed · live
Net result+71.0%
Realised P&L+$324
WLD · JUN 3, 2026
+15.0%
What happened: WLD (Worldcoin) was flagged early Tuesday morning UTC. NEVA caught the move cleanly - clean entry, clean exit, well inside the model's hold window.
Why we publish only the result: WLD shows the model still produces winners on majors, not just memecoin tails. The conviction-gate logic remains firm IP - published trades are the proof, not the recipe.
Sidelong
Statusclosed · live · 4d ago
Net result+15.0%
Realised P&L+$79
Real Binance USDT-M Futures trades from the firm's live trading account, May-June 2026. P&L includes leverage and is net of fees and funding. Past trades shown for transparency only - they are not a forecast and not all trades are winners (full WR and aggregate stats below). NEVA inputs, weights, and conviction logic remain proprietary.
01
16 AlgorithmsTested · killed · validated - full lab notebook
→ 02
Crypto Quant Pro30-lesson course · $499 · production methodology
→ 03
Daily Quant BlogLong-form essays · zero filler · published every day
→ 04
Strategic FundSpot equity in vetted projects - USDT entry
→ 06
Funding Rate ScannerBinance · Bybit · Hyperliquid - live APY ranking
→ 07
90-Day Track Record$1000 → $2129 · path-dependent · verified
→

From event to your position - in under 3 seconds.

Average end-to-end
2.4 seconds
p99 < 5 seconds
T+0ms
Event fires
Binance announcement published, aggregate-trade signature detected, or our proprietary conditions are met.
T+150ms
Validation
Multi-factor filters score the candidate. Sub-threshold signals discarded silently. BTC-regime + position-guard checks.
T+400ms
Risk sizing
Position size computed from your account balance × your risk preference × leverage. Pre-existing positions on the same symbol are left untouched - manual trades stay yours.
T+800ms
Order placed
Market order via your own Binance API. ISOLATED margin. Stop-loss attached as separate algo-order. No round-trip through any third-party broker.
T+2400ms
Confirmed
Position open. Telegram alert sent. Live in your dashboard. Average human trader is still reading the news.

From raw market data to your position - in under three seconds.

STEP 01
Observe
Four independent engines monitor the full Binance Futures universe in parallel - price action, derivatives state, exchange-level events, and information-flow signals.
STEP 02
Validate
Each candidate is scored against multi-factor filters validated on rolling historical windows. Sub-threshold signals are silently discarded.
STEP 03
Deliver
Telegram message with side, entry zone, recommended size, expected hold, and stop. Single-tap copy. Average from detection to delivery: 2.4s.
STEP 04
Execute
Manual on your terms - or connect your Binance API and the engine opens the position automatically with your sizing and SL settings.