Live Trading Signals

CAFEa market structure engine BETA

Crypto ATR Fractal Engine by Cryptanov measures how far price usually travels after a phase change and shows where the market has an edge. Every signal is recorded in real time before the outcome — a forward test, not a backtest. Access from $9.99 per month.

Free access to the CAFE group: 95 of 100 places left — then subscription $9.99/month
***Closed
LONG
Entry0.000000
Opened00.00.0000 00:00
Unlock All 30+ Coins
***Closed
LONG
Entry0.000000
Opened00.00.0000 00:00
Unlock All 30+ Coins
***Closed
LONG
Entry0.000000
Opened00.00.0000 00:00
Unlock All 30+ Coins
***Closed
LONG
Entry0.000000
Opened00.00.0000 00:00
Unlock All 30+ Coins
XRPUSDT Closed
LONG
Entry1.236500
Exit1.244200
Opened29.10.2026 23:43
Closed30.10.2026 06:45
+0.62%
***Closed
LONG
Entry0.000000
Opened00.00.0000 00:00
Unlock All 30+ Coins

VIP / Premium Signals

All coins and scales, market-regime filter and strategy map.

500+Signals with outcomes
5Years of historical testing
34+Coins + the whole market
Demonstration data slices

Strategies D, D1, D2, D3 — what they show

D stands for Dynamic — dynamic entries. These four are not trading strategies. They are demonstration slices of the forward test: each shows how the same entry points behave with a different exit and risk logic. They can be used, but we do not recommend trading them as they are — and if you do, longs only.

First version · asymmetric

Strategy D

The very first variant of entry selection from the stats collector. The expected take can be smaller than the stop — the asymmetric logic: frequent small wins, rarer larger losses.
Symmetric · short stop

Strategy D1

Take is larger than the stop, stops never exceed ~4% — built for ×20 leverage. The flip side: short stops are hit by ordinary noise very often; shorts even more so.
Same entries · stop 10%

Strategy D2

The same stats-collector entry points, but with a fixed 10% stop — for clarity: how the entries behave with a wide, simple risk limit.
Symmetric · beyond the fractal window

Strategy D3

Large expected take, the trade is held beyond the fractal window until price returns to the signal point. Win rate drops: rare takes, frequent stops. We measured the inversion too — the result is moderate.

Why we keep them public: they show honestly what happens with different exits on the same data. The working logic is described below.

Shorts — honestly

We measured shorts on all the same data. Most shorts do not work after fees. Only a narrow selection works — roughly 15% of what longs give — and only with our own rules, which we do not publish.

We also tried the opposite: inverting losing shorts into long trend trades. Before fees there is a small edge, after fees — about zero: the losses of shorts come from rare extreme moves that a mirrored trade cannot capture.

Trend strategies — honestly

In every trend test we ran — 90 variants of moving averages and channel breakouts on 5 years, plus trend versions on the whole market — the yearly drawdown is large and none of the variants is positive in every year.

Why: strong trends are rare and look the same at entry as thousands of ordinary moves that reverse. On the way to a level price reverses more often than it continues — the very property CAFE measures. A trend system pays fees on many false starts and lives on a few big moves, so between them it drags through long losing streaks.

Key tool

Trend filter: it forbids, it does not point

The trend filter watches the whole market and switches on a ban: when the market is in a one-sided move, it forbids entries against it. It does not show where to enter and is not a trend signal.

Important: do not try to trade "with the trend" by the filter. Above we showed why trend trading gives large drawdowns. The filter only removes the most dangerous entries from the CAFE data.

The filter state is published for subscribers in Telegram: when longs are forbidden, when shorts are forbidden, when both are allowed.

What we recommend

Deep zones, the long slice and the trend filter

This is not a "long system". CAFE measures the market skew — and on our data the skew works for longs on deep pullbacks. That is exactly what the long slices of the statistics show.

  1. Open the overview or statistics and switch the filter to LONG — this is the long slice of the forward test.
  2. Look at levels 4–8%: the deeper the level, the higher the expected value after fees. Levels 0.5–2% are negative after fees.
  3. Enter only when the trend filter does not forbid longs (state in Telegram for subscribers).
  4. Limit risk in advance: isolated margin, a stop on every position, no averaging down.
  5. Open the long slice
Overview page with the LONG filter enabled
Overview → LONG: the long slice of all strategies and levels of the forward test.

Build your own strategy

The data lets you assemble an approach for yourself: choose coins, entry levels, how many coins to trade in parallel and what share of the deposit to give each position.

Fractals differ: on the small ones (27m, 50m) trades are short, and losses do not drag on for long in any case; on the large ones (2h, 4h) moves are bigger but take longer. Combining them gives an even more flexible strategy.

You make all decisions on your own account yourself. The site provides measurement data and statistics, not investment advice; past results do not guarantee future ones.

Our trading bot

Unibot — CAFE rules on 34 coins

Unibot is our own executor: the CAFE engine plus a set of our own rules for entry, position size and protection, on 34 liquid coins. It is the only rule set tested on 5 years of history with costs measured against the live exchange account.

+148%per year on average (model, 5 years, stake 1/450)
+118%worst year
−12%worst drawdown on open positions

This is modelling of the past with real costs, not a promise. Since 24.10.2026 the current version runs without manual changes; the details are in the technical passport.

Faster, but riskier

CAFE screener — the whole market

The same engine on all liquid Binance futures — about 120 coins, without Unibot's 34. More signals and faster turnover, but small coins are riskier: sharp delistings, thin order books, manipulation. So the screener trades with a half stake and extra protections.

  • Forward test since 20.10.2026: ≈ +0.4% per trade after fees
  • Live trading with a half stake since 29.10.2026
  • Unlike Unibot: the whole market instead of 34 coins, more trades, higher risk; the history is still short

Bot connection and access to the rules

On the top plan the bot can be connected to your own account — an exclusive service, discussed individually.

Access to the Unibot and screener rules is possible on request for those who really pay for the bot's work and want to connect it themselves. These are our own rules and our edge, so access is only by personal agreement.

Discuss in Telegram

CAFE ATR engine

How the structural ATR differs from the classic one and what the map of depths shows — on a separate page.

Open the ATR engine page →

For investors

Separate terms: from partnership to the sale of the strategy. Terms are discussed personally. Technical passport — results, methodology, risks.

Technical passport →

Absolute Transparency. No Look-Ahead Bias.

Most systems curve-fit past data. We play by strict rules: every signal is published beforehand and cannot be faked.

Zero Look-Ahead Bias

The algorithm cannot see the future. It trades blindly strictly on current mathematical realities.

No Curve-Fitting

We don't over-optimize for past performance. The math works dynamically in any market phase.

100% Verifiable

Every signal is logged before execution. You can audit its performance in real-time on the blockchain.

No Backtesting Magic

This is NOT a Backtest — It's a Forward Test!

Cryptanov is a full forward simulation (forward test) with complete verification of every signal. All data is collected in real-time on the live market, with no curve-fitting to historical data.

Typical Backtest

  • Analyzes past data with full hindsight
  • Results are curve-fit to look perfect
  • No way to verify — just trust the screenshot
  • Often breaks in live trading

Looks great on paper. Falls apart in reality.

LIVE

Cryptanov Forward Test

  • Bot places a LONG or SHORT signal in real-time
  • Signal is logged and timestamped before outcome
  • Market confirms: profit or loss — no editing
  • Every result is publicly verifiable

All results are from a real-time forward test

How It Actually Works

Algorithm detects pattern
Bot opens LONG/SHORT
Market moves
Result confirmed
Every signal is a forward test: the system cannot see the future. Results are confirmed by the market, not fitted to history.
Patience is your ultimate tool

Waiting for the Best Entry Point

The system patiently waits for the best entry point instead of jumping into a trade at current prices like most novice traders do.

The truth is, if you wait just a couple of candles, you almost always get a better price than if you entered in a hurry. This is one of the main secrets of the strategy.

We prefer the "rare but accurate" approach. The main goal is not the number of trades, but achieving a high average Expected Value (EV > 1). There is no need to rush.

Rule #1: Protect your deposit

Sideways Bounce and Trend Anomalies

In 70-80% of cases, the price moves in a range and bounces off the level. But there are also strong non-pullback trends — anomalies that can wipe out a month's worth of profit.

This is normal market mechanics: first, it feeds you and lets you earn, and then it takes everything away at once and even liquidates your deposit. By trying to wait out an anomaly, you lose everything.

That is why risk must be limited in advance: a stop on every position, isolated margin, no averaging down and no adding margin to a losing trade.

The main idea is not how to earn, but how not to lose what you have already earned. To do this, you can use our tables with smart filters for valid coins, or apply a "bad days" filter. The platform provides raw data — it is the trader's job to build their own strategy.

From Thousands of Hypotheses to One Working Strategy

A system born from rigorous testing and real losses.

My name is Alex. I have been doing algorithmic trading for three years. Along the way there were hundreds of strategies, several websites and thousands of tests — and most ideas broke against reality: either the algorithm hid look-ahead bias, or over the long run the result slid to 50/50 and exchange fees slowly ate the deposit.

The market is full of scammers trading "on intuition" and passing off wishful thinking as reality. My goal was to find a true mathematical edge—a statistically valid strategy that works transparently on large datasets.

In spring 2026 CAFE — Crypto ATR Fractal Engine — grew out of that experience. Since 25 May 2026 every signal has been recorded in a real-time forward test; the engine has been tested on 5 years of history and trades a live account. I do not guess — I test, and I show the results, losses included.

Self-Learning Entry Points

Our system continuously analyzes each coin's behavior and calculates optimal entry and stop-loss levels in real-time.

View Live Levels

Levels update automatically — no static values

VIP / Premium Signals

All coins and scales, market-regime filter and strategy map.

95 spots remaining

Join Telegram Group

The Illusion of Fast Money

Why trying to turn $100 into $150 in a day leads to liquidation, and how the system solves it

The Problem of Unrealistic Expectations

Even strong funds consider 15–30% a year an excellent result. Yet traders with a $100 deposit want to make $50 in a day. This forces them into massive leverage and high risks, mathematically guaranteeing a blown account.

Loss Mechanics & Black Swans

We naturally risk more than we earn. Taking 10x leverage on $100 creates a $1,000 position. Earning 10% means the price must move 10%, which is rare. A single "black swan" can wipe out a month of gains—this is just market mechanics.

The Strategy D1 Solution

Strategy D1 is designed to offset these inevitable losses through a high Expected Value (0.2-0.3%+ per trade). The cumulative profit allows you to stay in the green long-term, neutralizing drawdowns.

ATR Engine · Liquidity Window

One Chart. One Algorithm. No Noise.

99% of systems pile on indicators to predict the unpredictable. Cryptanov does the opposite — strip everything down to pure price structure and measure only what is mathematically real.

What 99% of Systems Use

Dozens of indicators. One opinion. Zero statistical edge.

RSI / MACD / Stochastic
Bollinger Bands & EMA clouds
News feeds & sentiment scoring
Elliott Waves & pattern recognition
Macro calendar & event-driven entries

None of these predict the future. They describe the past — with delay.

PURE MATH

Cryptanov Engine

The signal is a mathematical statement, not a prediction.

Liquidity Window Algorithm — identifies the exact zones where stop-hunts and liquidation cascades statistically cluster
Smart ATR Engine — measures the true average volatility range per coin and timeframe, dynamically recalibrated
Raw price structure only — no lagging indicators, no human interpretation, no curve-fitting
ATR Volatility Band — the only thing the algorithm reads
Read the ATR Method in detail
Risk-Reward Architecture

Symmetric vs Asymmetric Risk-Reward

The classic 1:3 RR rule looks perfect on paper. In a live, leveraged market it often works against you. Two validated models solve this from opposite angles.

Why Classic 1:3 RR Fails Small Accounts

Waiting for a 3× take-profit in a volatile market means your stop gets triggered first — statistically, more often than not. The wider your target, the lower your win rate. Lower win rate with leverage is a slow death spiral. Large funds can hold for months. A $500 account at ×10 leverage cannot. Cryptanov offers two architectures that work within real-world constraints.

Take > stop · results unstable

D1–D3 — symmetric models (experimental)

  • ATR-calibrated entry: stop is placed at the statistical edge of normal volatility — where price statistically does not go
  • Take-profit always larger than stop-loss — positive Expected Value per trade by design
  • Occasionally catches large momentum moves (+15–30%) when market aligns
  • A short stop allows higher leverage, but market noise hits it more often

Catches rare large moves, but because of short stops and long waits results are not yet stable. Kept in the forward test for observation.

Working zone 4–8% · tested on 5 years

ATR levels — asymmetric model (core)

  • Position breathes through normal volatility — no premature exits on market noise
  • In roughly two cases out of three price returns after a temporary pullback
  • Exchange-side stop on every position; liquidation lies beyond the stop
  • High frequency of small wins, rare but meaningful losses — net positive EV over time

The deeper the level, the higher the expected value after fees. The core of our own trading module.

ATR Level Strategies (Asymmetric)

The working levels are 4–8%: the deeper the level, the higher the expected value after fees. Shallow levels of 0.5–2% are near zero before fees and negative after — better not to trade them.

Not recommended

Levels 0.5–2%

Shallow pullbacks: near zero before fees and negative after fees over 5 years of history. The data is in the tables, but we do not recommend trading them.

Asymmetric · ATR 4%

Level 4%

Leverage ×10

Lower edge of the working zone: the most trades; near zero after fees without a filter, positive with the regime filter.
Asymmetric · ATR 6%

Level 6%

Leverage ×10

Middle of the working zone: consistently positive expected value after fees.
Asymmetric · ATR 8%

Level 8%

Leverage ×10

The deepest level: the highest expected value after fees, but fewer trades and more tail moves.

The Truth About Stop-Losses

Why stops only lock in losses and cut profits

  • A stop-loss is a tool for locking in losses. Any stop — whether 1%, 2%, 3%, 4%, or 5% — simply cuts your position at a loss. A stop doesn't protect from losses — it guarantees them.

  • For asymmetric strategies (ATR levels) the 5-year historical check showed that an 8% stop on the exchange with ×10 leverage beats tight stops — a tight stop knocks out trades that would have recovered. In the symmetric Strategy D1 the stop is placed by levels and the take-profit is always larger than the stop.

  • This is why the best potential lies in asymmetric strategies (ATR levels) on deep pullbacks: about 65–70% win rate and positive expectancy after fees. Look at expectancy, not just win rate.

  • Conclusion: don't use tight stops — they only cut profits. Look at asymmetric strategies with the highest EV potential.

Example: Strategy D3 is a symmetric strategy expecting greater profit than stop. Take-profit > stop-loss. However, when you expect larger profits, stops trigger more frequently, even with considerable stop-take distances. The strategy produces "small" losses but compensates with large gains. This is simply an example of how asymmetry works.