Start here · SignalDesk v1.0.0

Learn SignalDesk in about 10 minutes

SignalDesk is a self-hosted crypto buy-setup scanner. It reads public Binance spot data, scores each pair in your watchlist from 0 to 100, and turns the result into a concrete trade plan with an entry, a stop and two targets.

It is a research and education tool. It never connects to your exchange account, never places orders, and does not need an API key.

Open dashboard Try the backtester Jump to the glossary

No API keys public Binance spot data Rule-based transparent 0–100 score No orders analysis only, never trades

What SignalDesk is

One sentence: it filters the market for possible buy setups and shows you the risk before you act.

SignalDesk watches a list of coins you choose. For every coin it combines a daily trend filter with 4-hour entry timing. If the evidence lines up, it produces a score, a verdict and a trade plan.

Step 1

Daily trend

Is the bigger trend healthy? Price above key daily averages is treated as a tailwind.

Step 2

4H entry timing

Is the shorter timeframe offering a sensible moment — RSI, support, volume and momentum?

Step 3

Risk plan

Where is the stop, where are the targets, and does the reward justify the risk?

What it is not: not financial advice, not a broker, not a guaranteed signal service. A high score is a reason to look closer, not a reason to skip your own research or risk limits.

60-second quick start

From blank dashboard to your first full analysis.

Open the dashboard
Go to /. The first automatic scan starts when the page loads.
Build a small watchlist
In the Watchlist panel, type BTC, eth/usdt or SOLUSDT and press Add. Start with 5–8 liquid pairs, not 50.
Press Rescan
A scan of a small watchlist usually takes a few seconds. The status dot turns green when it finishes.
Read the row
Start with Score, Signal, Entry / Stop and Target / R:R. Ignore every other column until those make sense.
Click the pair
A row opens the full-analysis page with the chart, evidence, score breakdown and every indicator used.
Optional: backtest it
Open the Backtester, choose the pair and run the defaults. Compare the strategy with simply buying and holding.
Monitor with Auto 5m
Turn on Auto 5m in the top bar if you want the dashboard to rescan every 5 minutes. Live prices refresh every 5 seconds either way.

Good first watchlist: BTCUSDT, ETHUSDT, SOLUSDT, BNBUSDT, XRPUSDT. Liquid pairs have cleaner candles and more reliable indicators.

Dashboard tour

What every area on the main screen is for.

Area	What it does
Top bar	Status, Auto 5m switch, Guide, Backtest and Rescan. The status dot shows idle, busy or error.
Stat cards	How many pairs are strong buys, buys or on watch, plus the average score across the watchlist.
Watchlist panel	Add or remove coins. Changes are saved to `config.json` outside Docker, or the `/data` volume in Docker.
Buy setups table	One row per pair with score, verdict, indicators and the trade plan.
Search & filter	Filter by symbol text or by verdict (Strong buy, Buy, Watch, Wait, Avoid).
Live pill	LIVE means the price feed is healthy. STALE means the feed could not be reached; analysis stays visible.
How the score is built	The six scoring components and the legend for every plan field.

Reading a result row

Column	Meaning
Pair	The coin and its quote currency, e.g. SOL/USDT.
Price / 24h	Live Binance spot price and the change over the last 24 hours. Refreshes every 5 seconds.
Score	0–100. Higher means more of the rule-based evidence agrees.
Signal	Verdict mapped from the score: Strong buy, Buy, Watch, Wait or Avoid.
RSI 4H	Momentum gauge from 0–100. Below 30 is oversold, above 70 is overbought.
Trend 1D	Daily trend structure: up, down or mixed.
Vol	Current volume compared with its 20-bar average. `2.0×` means twice normal.
Support / Resist.	Nearest active LuxAlgo swing support and resistance levels.
Entry / Stop	The plan entry at scan time and the stop-loss below it.
Target / R:R	First target and the reward-to-risk ratio. Higher R:R means more reward per unit of risk.

Strong buy75–100 Buy60–74 Watch45–59 Wait30–44 Avoid0–29

How a signal is made

Two timeframes, six checks and one honest rule: only closed candles count.

Timeframe	Job	Why it matters
1D — one day	Trend filter	Stops the scanner from buying into a strong downtrend.
4H — four hours	Entry timing	Catches the shorter-term setup: RSI, support, volume, momentum and divergence.

Closed candle rule: a 4H candle is “closed” when its four hours have finished. Signal maths ignores the candle currently forming. The live price can still tick between scans; it is shown separately and does not rewrite the score.

The six score components

Component	Max points	What earns points
Daily trend	25	Price above the daily 200-EMA, 50-EMA above 200-EMA, daily RSI healthy.
RSI (4H)	20	Oversold or cooling RSI that has started turning up.
Support	20	Price close to a tested swing support, measured in ATR.
Volume	15	Volume above average on a green candle, plus buyer pressure.
Momentum	12	Breakout pressure, aligned EMAs and volume-confirmed resistance breaks.
Divergence	8	Regular bullish RSI divergence: price lower low, RSI higher low.

Penalties reduce the score for overbought RSI, an overextended 24h move, thin volume, price far from support, price below the 4H 50-EMA and poor reward/risk. Two hard caps keep it honest:

Bearish daily structure caps the score at 44, no matter how good the 4H chart looks.
Very low volatility (ATR below 0.35% of price — usually stablecoins or pegged pairs) caps the score at 30.

From score to verdict

Score	Verdict	How to read it
75–100	Strong buy	Trend, timing and volume all agree.
60–74	Buy	Actionable setup; check the plan and the evidence.
45–59	Watch	Interesting, but wait for confirmation.
30–44	Wait	Weak or conflicting evidence.
0–29	Avoid	No edge or bearish structure.

Trade plans, without the jargon

Every signal answers three questions: where in, where wrong, where out.

Entry

Where the plan starts

The current price when the scan ran. The live price may have moved since then.

Stop

Where the idea is wrong

Placed below the nearest support, with an ATR buffer. If price reaches it, the setup has failed.

First target

The nearest resistance. Often the more realistic target and the default backtest exit.

Second target

The next resistance above T1. Further away, so fewer trades reach it.

Reward-to-risk ratio (R:R)

R:R compares the money you could make with the money you are risking. The formula is:

R:R = (target − entry) ÷ (entry − stop)

Example: entry $100, stop $95, target $110. Risk is $5, reward is $10, so R:R is 2.0. A 2R trade earns twice what it risks. The scanner penalises plans below 1R because you would risk more than you stand to gain.

Plan snapshot vs live price: the trade plan is calculated once per scan from the last closed candle. If the live price moves away from the plan entry, the displayed entry does not chase it. Re-run the scan when you want a fresh plan.

The full-analysis page and chart

Click any pair to see the complete reasoning behind its score.

The analysis page repeats the important parts of the dashboard, then adds the complete indicator readout, the evidence list, the score breakdown, break history and divergence detail.

What the chart contains

Element	Meaning
Candles	Green means the candle closed above its open; red means it closed below.
EMA 20 / EMA 50	Fast and medium exponential moving averages. Price above them is generally constructive.
Lux S / Lux R	Active LuxAlgo swing support and resistance levels, held until a new pivot replaces them.
Entry / Stop / T1 / T2	Dashed horizontal lines showing the current trade plan.
Volume panel	How much was traded in each 4H candle.
RSI 14 panel	Momentum line with 30 / 50 / 70 reference levels.
Break markers	LuxAlgo volume-confirmed break labels above or below candles.

Chart controls

Action	Result
Hover / move the pointer	Crosshair and a tooltip with OHLC, change, volume, EMA values and RSI.
Drag left or right	Pan through older or newer candles.
Mouse wheel	Zoom in or out around the pointer.
Two-finger pinch	Zoom on touch devices.
Double-click	Reset to the last 96 bars.
Toolbar + / − / Reset	Zoom in, zoom out or reset without the mouse wheel.
Keyboard ← / →	Pan the chart when it has focus.
Keyboard + / − / Home	Zoom in, zoom out or reset.

Summary cards

Fast read on trend, RSI, volume, ATR, support and resistance.

Evidence

Plain-English reasons the score is high or low, plus warnings.

Score breakdown

Exactly how many of the 100 points came from each component.

Indicator readout

Every input used by the scanner, grouped and labelled.

The backtester

A flight simulator for the scanner’s rules — not a crystal ball.

The backtester rewinds history and replays the same scoring engine used by the live scanner. It does not invent a separate strategy. For every closed 4H candle it scores the setup and simulates what would have happened if a signal had been taken.

Before you press Run

Control	What it changes
Pair	Which watchlist coin to replay.
Lookback	How many 4H bars to simulate. More bars = more trades and a more meaningful sample.
Min score	Only signals at or above this score are taken. Higher = fewer, more selective trades.
Risk / trade %	Percentage of current equity risked on each trade. Default 1%.
Exit target	T1 exits at the first target; T2 waits for the second and usually has a lower win rate.
Max hold	Close the trade after this many 4H bars even if neither stop nor target was hit.
Fee / side %	Exchange fee charged on entry and on exit.
Slippage / side %	Assumed price worsening on entry and exit. Real slippage can be worse.

How one simulated trade works

Signal on a closed candle
The engine scores candle i. If the score is high enough and no trade is open, the plan becomes pending.
Fill on the next candle’s open
This prevents lookahead. The engine cannot buy at a price it did not know yet.
Size the position
Risk a fixed percentage of equity, capped at available cash. No leverage.
Manage the exit
Check stop, then target, then max hold. If stop and target are both touched inside one candle, the stop is assumed first.
Record the result
Fees and slippage are applied, the trade is logged, and the equity curve is updated at every close.

Reading the result

Statistic	Meaning
Total return	How much the simulated account grew or shrank.
Buy & hold	What would have happened by simply holding the coin for the same window.
Max drawdown	The worst fall from an equity peak. Smaller is easier to live with.
Trades	How many simulated trades happened. Fewer than ~20 is a small sample.
Win rate	Percentage of trades that ended profitable. It says nothing about size of wins versus losses.
Profit factor	Gross profits ÷ gross losses. Above 1.0 is profitable before other costs; higher is better.
Average R	Average trade result measured in risk units. `+1R` means you made the amount risked.
Exposure	Percentage of bars with an open position. Low exposure means the strategy was mostly in cash.
Best / worst trade	The largest percentage winner and loser.

What R means

R is your risk unit. If you risk $100 on a trade, then +2R means you made $200 and −1R means you lost $100. R lets you compare trades fairly regardless of coin price.

Example from a 180-bar SOLUSDT run: $10,000 → $9,719.37 (−2.81%), buy & hold −21.65%, max drawdown −4.31%, 14 trades, 42.9% win rate, profit factor 0.70, average −0.07R, exposure 26.7%. The rules avoided most of the fall but still lost slightly. Fourteen trades is not enough evidence to trust the result.

What it does not model: the exact intrabar path, partial fills, funding, taxes, liquidity, exchange outages, news gaps and real-world slippage spikes. Results are a sanity check, not a promise.

How to experiment without fooling yourself

Change one input at a time so you know what caused the difference.
Compare every run with buy & hold, not just with zero.
Prefer more trades and longer lookbacks over a perfect short run.
Watch max drawdown as closely as total return.
Beware overfitting: tuning settings until one past period looks perfect usually fails in the future.

Glossary and abbreviations

Every shorthand used across the dashboard, chart and backtester.

Market data and timeframes

1D: One-day timeframe. Used for the trend filter.
4H: Four-hour timeframe. Used for entry timing and the chart.
OHLC: Open, High, Low, Close — the four prices that define a candle.
Candle: A fixed time block of price action. A green candle closed above its open; a red candle closed below.
Closed candle: A candle whose time period has finished. Indicators only use these to avoid unstable readings.
Spot: Buying the actual asset on Binance’s spot market, not a futures contract.
USDT / USDC / FDUSD: Dollar-pegged quote assets. A pair like SOL/USDT means SOL priced in USDT.
Base / quote: In SOL/USDT, SOL is the base asset and USDT is the quote asset.
Ticker: A small live snapshot containing last price, 24h change and 24h volume.
Live price: The latest spot price, refreshed every 5 seconds. It is not the same as the scan-time plan entry.
Watchlist: The coins SignalDesk scans. Stored in config.json or the Docker /data volume.
Scan: One full analysis pass over the watchlist. Usually takes a few seconds.
Rescan: Force a new scan, bypassing the short server cache.
Cache / TTL: Temporary stored results reused for a short time. TTL means “time to live” before the cache expires.
API: Application Programming Interface — the JSON endpoints the page uses to fetch data.
API key: A private exchange credential. SignalDesk does not need one because it only reads public data.

Indicators and chart concepts

EMA: Exponential Moving Average. A smoothed average that gives more weight to recent prices. The scanner uses 20, 50 and 200 periods.
RSI: Relative Strength Index. A 0–100 momentum gauge. Below 30 is often called oversold, above 70 overbought.
RMA / Wilder: The smoothing method used by TradingView’s RSI. SignalDesk uses the same maths.
ATR: Average True Range. A measure of how much price normally moves; used to size stops and judge volatility.
S/R: Support and Resistance. Price zones where buyers or sellers have previously stepped in.
Support: A level below price where demand has appeared before.
Resistance: A level above price where selling has appeared before.
Pivot / swing high / swing low: A local turning point. A pivot high has lower highs on both sides; a pivot low has higher lows on both sides.
LuxAlgo / Lux S / Lux R: A TradingView support-and-resistance indicator reproduced by SignalDesk. Lux S is its active support; Lux R is its active resistance.
Volume oscillator: An EMA-based measure of whether volume is expanding or fading. Used to confirm break labels.
Divergence: Price and RSI disagreeing. Bullish divergence: price makes a lower low while RSI makes a higher low. Bearish is the opposite.
Regular divergence: The standard reversal-style divergence described above, as opposed to hidden/continuation divergence.
Bullish / bearish: Bullish means leaning toward higher prices; bearish means leaning toward lower prices.
B / W / Bull Wick / Bear Wick: LuxAlgo break labels. B is a volume-confirmed close through a level; a wick label means the candle had a dominant wick.
Overbought / oversold: Informal RSI zones: overbought above 70, oversold below 30. They describe momentum, not a guaranteed reversal.
Momentum: The speed and strength of a move, often measured by breakouts and EMA alignment.
Trend filter: The daily-timeframe check that prevents buying into a clearly bearish market.

Signal and trade-plan terms

Score: The 0–100 total from the six components, minus penalties and subject to hard caps.
Verdict / signal: The label mapped from the score: Strong buy, Buy, Watch, Wait or Avoid.
Min score: The lowest score the backtester will act on. The live scanner always displays all scores.
Setup: A potential trade opportunity that matches the rules. It is not an instruction to buy.
Entry: The planned starting price for a trade, taken at scan time.
Stop / stop-loss / SL: The price at which the idea is considered wrong and the position should be closed.
T1 / T2: First and second profit targets, based on resistance levels.
R:R / risk-reward: Reward divided by risk. A 2R:R plan risks $1 to target $2.
R / R multiple: A trade result measured in risk units: +2R made twice the risk, −1R lost one risk unit.
Plan snapshot: The entry, stop and targets calculated at scan time. Live price can move away from it.

Backtest and performance terms

Backtest: A historical simulation of the strategy on past candles.
Bars / lookback: How many candles are simulated. On a 4H chart, 180 bars is about 30 days.
Fill: The price at which a simulated order is executed.
Next-open fill: A signal from a closed candle is filled at the next candle’s open, preventing lookahead.
Lookahead bias: Accidentally using information that was not available at the time. The backtester explicitly avoids it.
Fixed-fractional risk: Risking a fixed percentage of current equity on every trade, e.g. 1%.
Position size / notional: How many coins are bought and what that position is worth. Capped at available cash; no leverage.
Fee: Exchange commission charged when entering and exiting.
Slippage: The difference between the expected price and the actual fill. Modelled as a small percentage per side.
Max hold: The maximum number of bars a trade may stay open before it is closed at the market.
P&L: Profit and loss, in dollars or percent.
Equity / equity curve: Cash plus the value of any open position, plotted over time.
B&H / buy & hold: The baseline of buying at the start of the window and doing nothing.
Win rate: Winning trades divided by total trades, as a percentage.
Profit factor / PF: Gross profits divided by gross losses. Above 1.0 means the winners outweigh the losers.
Max drawdown / Max DD: The largest peak-to-trough fall in account equity, as a percentage.
Exposure: The percentage of bars the strategy had money in the market.
Sample size: How much evidence a result contains. A handful of trades is noisy; more trades are more reliable.
Overfitting: Tuning a strategy until it fits one historical period perfectly. It usually performs poorly on new data.

System and deployment terms

Flask: The Python web framework that serves the dashboard, pages and JSON APIs.
Waitress: The production-style web server used inside Docker.
Docker: A way to run SignalDesk in a self-contained container with its own Python environment.
CSV: Comma-separated values. The CLI can export scan results as CSV.
JSON: A plain-text data format used by the APIs and config.json.
Health check: /api/health returns a small OK response so Docker can tell the app is alive.

Routes you can type

URL	What it opens
`/`	Dashboard and watchlist
`/coin/SOLUSDT`	Full analysis for SOL/USDT
`/backtest`	Strategy backtester
`/guide`	This page
`/api/prices`	Live JSON price feed
`/api/backtest/SOLUSDT`	Backtest JSON for one pair

Beginner FAQ

Short answers to the questions new users ask first.

Does SignalDesk place trades for me?: No. It has no exchange account, no order execution and no API keys. You stay in full control.
Why is the price different from the Binance app?: Make sure both show the same spot pair. Futures, Binance.US and regional apps can differ. Tiny last-tick differences are normal.
Why is the live price different from the plan entry?: The plan is a snapshot from the last scan. Live price keeps moving. Press Rescan for a fresh plan.
Why is the score low even though the candle is green?: One green candle is not enough. The score needs trend, RSI, support, volume, momentum and divergence to agree, and caps can override a good-looking 4H chart.
Why do I see no Buy or Strong buy signals?: Corrective markets often produce Wait or Avoid scores. The scanner refuses weak setups by design. Check the evidence list instead of forcing a trade.
What do B and W mean?: LuxAlgo break labels: a volume-confirmed break and its wick variant. They add or subtract momentum points.
Why did the backtest take a few seconds?: It replays the full scoring engine on every candle. Longer lookbacks need more computation; identical runs are cached for 120 seconds.
Why are backtest and live results different?: Backtests use historical fills, fixed costs and one pair at a time. Live results depend on when you act, real slippage and the current market.
Which coins can I add?: Liquid Binance spot pairs quoted in USDT, USDC or FDUSD, with at least 60 daily candles of history. Stablecoin and pegged pairs are rejected.
Can it run without internet?: No. It needs public Binance candle and ticker data.
Is this financial advice?: No. It is rule-based technical analysis for research and education. You are responsible for your decisions and risk.

Limits, safe use and next steps

How to use the tool without pretending it is magic.

Know the limits

Technical analysis only. News, fundamentals, regulation and liquidity shocks are not modelled.
Indicators lag. RSI, EMAs and swing levels are based on past prices; they can be late or wrong.
Data can fail. Binance endpoints can be blocked, rate-limited or delayed. The live pill turns STALE when prices fail.
Backtests simplify reality. Intrabar path, partial fills, funding, taxes and real slippage are not fully modelled.
One pair at a time. The backtester does not simulate a portfolio with shared capital or correlation risk.

Safe-use checklist

Decide your maximum risk per trade before looking at signals.
Use a fixed, small percentage — the backtester defaults to 1%.
Never move a stop further away to avoid being wrong.
Treat a high score as a prompt to research, not a command to buy.
Prefer liquid pairs and avoid trading around major news if you do not understand the risk.
Paper-trade or small-size first while you learn how the signals behave.

Risk warning: crypto prices are volatile. A stop can gap, a level can fail, and a backtest can look good while the next month looks nothing like it. Never risk money you cannot afford to lose.

Learn SignalDesk in about 10 minutes

What SignalDesk is

Daily trend

4H entry timing

Risk plan

60-second quick start

Open the dashboard

Build a small watchlist

Press Rescan

Read the row

Click the pair

Optional: backtest it

Monitor with Auto 5m

Dashboard tour

Reading a result row

How a signal is made

The six score components

From score to verdict

Trade plans, without the jargon

Where the plan starts

Where the idea is wrong

First target

Second target

Reward-to-risk ratio (R:R)

The full-analysis page and chart

What the chart contains

Chart controls

Summary cards

Evidence

Score breakdown

Indicator readout

The backtester

Before you press Run

How one simulated trade works

Signal on a closed candle

Fill on the next candle’s open

Size the position

Manage the exit

Record the result

Reading the result

What R means

How to experiment without fooling yourself

Glossary and abbreviations

Market data and timeframes

Indicators and chart concepts

Signal and trade-plan terms

Backtest and performance terms

System and deployment terms

Routes you can type

Beginner FAQ

Limits, safe use and next steps

Know the limits

Safe-use checklist

Your next steps

Build your watchlist

Study one pair

Replay the strategy