Strategy Scanner + Backtest Engine

Your indicators cannot see
what liquidates you.

RSI and MACD are derived from price. Liquidations are produced by funding, open interest and stacked leverage. The Strategy Scanner tests all of it: 50,000 parameter combinations across 40 market fields, and returns only the strategies that survive out-of-sample validation.

3 days free on the Maxi plan. Cancel any time before the trial ends.

ScanningBTCUSDT · 4h · 720d
0combinations tested
312
Survived
48.2k
Overfit
1,4k
Below min trades
Leaderboardsorted by robustness
1
funding_pctl < 12 AND liquidations_skew > 0.7
return +184%sharpe 1.94
82
2
mvrv < 1.1 AND etf_flow_streak >= 3
return +141%sharpe 1.61
74
3
oi_change < -8 AND drawdown < -22
return +118%sharpe 1.38
68
4
rsi < 24 AND macd crosses_above 0
return +96%sharpe 0.71
OVERFIT
Row 4 has the highest raw return of the four and still fails. It only worked on one slice of history.
The blind spot

Four indicators reading the same number

RSI, MACD, moving averages and Stoch RSI are all transformations of closing price. Stacking them does not add information, it repeats it. Meanwhile the variables that actually decide where the market hunts stops are somewhere else entirely.

What a chart indicator sees

  • Closing price
  • Price, smoothed
  • Price, normalized 0 to 100
  • Price, two smoothings subtracted

One input. Four views. Every trader on the same chart is looking at the same thing at the same time, which is precisely what makes their stops predictable.

What actually produces the liquidation

  • Funding rate and its percentilehow expensive it is to stay long
  • Open interest changeleverage entering or leaving
  • Liquidation skew and streakswhich side is being cleared out
  • Crowding and long/short ratiohow one-sided positioning has become

The Scanner treats these as first-class testable fields, not as a side panel you glance at after the trade is already open.

How it works

Three steps, no code

You describe the hypothesis. The engine does the search, the validation and the ranking.

STEP 01

Pick up to 4 fields

Choose from 40 fields across 9 families. Mix a derivatives field with an on-chain one and a macro one if that is your thesis. The engine builds the parameter grid for every threshold of every field you selected.

Indicators
FUNDING PCTLLIQUIDATIONS SKEWMVRVM2 CHANGERSINUPLETF FLOW PCTLCYCLE PHASEFEAR GREED
4 selected · grid search · TP 10,20,30 · SL -5,-10
STEP 02

Run the scan

Up to 50,000 combinations are evaluated against BTCUSDT history in your browser. Grid or random search, configurable folds, rolling or anchored walk-forward, minimum trade count, capital and allocation.

ScanningBTCUSDT · 4h · 720d
0combinations tested
312
Survived
48.2k
Overfit
1,4k
Below min trades
STEP 03

Read what survived

Every result carries a robustness score built from consistency, decay and concentration, plus a deflated Sharpe that penalizes the fact you searched a large space. Results are benchmarked against Buy & Hold and DCA, so a strategy that merely tracked the market is visible as such.

Leaderboardsorted by robustness
1
funding_pctl < 12 AND liquidations_skew > 0.7
return +184%sharpe 1.94
82
2
mvrv < 1.1 AND etf_flow_streak >= 3
return +141%sharpe 1.61
74
3
oi_change < -8 AND drawdown < -22
return +118%sharpe 1.38
68
4
rsi < 24 AND macd crosses_above 0
return +96%sharpe 0.71
OVERFIT
Row 4 has the highest raw return of the four and still fails. It only worked on one slice of history.
Out-of-sample validation

Every strategy looks brilliant in hindsight

Search a large enough space and something will fit any history perfectly. That is not an edge, it is a coincidence you paid for. The Scanner fits parameters on one slice and judges them on the next one, then keeps a final holdout the search never touched.

Walk-forward
Rolling or anchored, with configurable folds.
Untouched holdout
A final slice the optimizer never saw.
Deflated Sharpe
Discounts the Sharpe by how many candidates you tried.
Robustness score
Consistency, decay and concentration in one number.
fit judged final holdout
fold 1
fold 2
fold 3
fold 4
fold 5
holdout
The data universe

40 fields. Only 5 of them are momentum.

That ratio is the entire argument. Most retail strategies are built out of the smallest family on this list.

Derivatives
15
funding ratefunding percentilelong/short ratioL/S changeOI changenet longs pctlnet shorts pctlliquidationsliquidations pctlliquidations skewlong liq pctlshort liq pctlliquidation streakL/S change pctlcrowding
On-chain
8
MVRVNUPLSOPRPuell multipleVDDhash rate pctlrealized P/L pctldominance
Momentum
5
RSIMACDStoch RSI %KStoch RSI %DRSI divergence
Cycle
3
cycle phasecycle progresscycle drawdown
Sentiment
2
fear & greedsearch trends
ETF flows
2
ETF flow pctlETF flow streak
Macro
2
M2 changeDXY change
Price
2
price changedrawdown
Volatility
1
realized volatility
Backtest Model Creator

When you already know the thesis

The Scanner searches for you. The Model Creator lets you write the rule yourself: nested AND / OR groups up to four levels deep, up to twenty conditions, entry and exit logic, take profit and stop loss. Then it runs the same backtest engine over it and shows the equity curve, the trade list and the full metric panel.

Open the Model Creator
Entry rulesIF
AND
funding_pctl < 12
OR
liquidations_skew > 0.7
mvrv < 1.1
TP 20%SL -8%up to 20 conditions
Market context

The worst time to build a system is when everything is going up

In a bull run every strategy looks like genius and nobody validates anything. Right now attention is gone, funding is compressed and positioning is thin. That is the environment where a tested hypothesis is worth something and a feeling is worth nothing.

Build and validate the system while it is quiet. You do not want to be assembling it in the middle of the next expansion.

Included with your account

A short course on using it properly

Five lessons walking through a real scan from hypothesis to validated result, including the mistakes that produce great looking numbers and no edge.

01

Why price-derived indicators repeat themselves

6 min
02

Reading funding, open interest and liquidation skew

9 min
03

Your first scan, field by field

12 min
04

Telling a real edge from an overfit one

10 min
05

Turning a validated result into a rule and an alert

8 min

Start with three days, decide after

Run real scans on the trial. If the results are not worth it to you, cancel before it ends.

3 days free

MAXI ACCOUNT

Everything you need to run and validate strategies.

$25/month
  • Strategy Scanner, full 40-field universe
  • Backtest Model Creator with rule trees
  • Walk-forward and robustness scoring
  • Liquidation Heatmap Pro
  • Saved scans and templates
Start 3-day free trial

NAKAMOTO PRO

For traders running this as their primary process.

$55/month
  • Everything in Maxi
  • Full historical depth on every field
  • CounterFlow contrarian signal
  • REST API and MCP server access
  • Position Signals and alerts
Choose Nakamoto Pro

Want to compare every plan? See full pricing

Test the hypothesis before the market tests you

Fifty thousand combinations, forty fields, one number telling you how much the result deserves your trust.

Start 3-day free trial