HISTORICAL MODEL CHECK

Simple Backtest Lab

See whether the trend model actually worked on past market data. The page finds the strongest weight mix, then checks it again on unseen data.

Timeframe

Start

End

The test automatically uses the first 70% of available history for training and the final 30% for validation. If a coin or exchange market did not exist at the selected date, testing begins from the earliest history actually available. Older halving presets are for comparison; only the 2024 → Today cycle is eligible to become the live Current Halving model.

Halving cycle

Ready0%

Last run details

Ready.

Universe 30 coins Historical source Bitget Optimizer Walk-forward Selection Top 3 uptrends

Historical Radar Test

Freeze the Radar on a past date, reconstruct the full market state using only data known at that time, rank the strongest signals in either direction, then reveal their later returns.

Timeframe

As of date

Choose a past date and run the historical Radar.

Candidate ranking uses no data after the selected date. Future prices are read only after the TOP 10 list is frozen. Liquidity is recorded only as research metadata and does not affect selection or ranking. HIGH / MEDIUM / LOW are relative liquidity bands based on the latest 20 selected-timeframe candles. Direction strength is shown from 0–100.

Export Backtest Excel

Choose one timeframe and one date range. Only the selected timeframe will be calculated and exported, so the download is much faster.

Timeframe

Start

End

Sampling

Rows

Ready.

Start / End / Sampling / Rows work exactly as before. Only the selected timeframe is downloaded and calculated. Export includes candle open/close, signal confirmation, entry time/price, New Episode, and Research Signal fields.

Backtest summary

Run the backtest to see the result.

The result will compare the current model with the optimized model using unseen validation data.

Best weight mix

Run the backtest to calculate optimized weights.

Validation result

No validation result yet.

Current halving model

Trend—

Momentum—

BTC relative—

Volume—

Breakout—

Choose a halving-cycle preset above, then run the backtest. The 2024 → Today result can be saved as the live current-cycle model.

Robust weight for Crypto Radar

Chosen to stay competitive across multiple halving cycles instead of winning only one market period.

Not calculated

Trend—

Momentum—

BTC relative—

Volume—

Breakout—

Run the backtest. If enough cycle data exists, the robust weights will be saved automatically and used by Crypto Radar in this browser.

Best weights by Bitcoin halving cycle

Each cycle is optimized separately using only data available inside that cycle.

Run the backtest to calculate halving-cycle weights.

2016 → 2020

2020 → 2024

2024 → Current

How this result is made

1 · Historical dataUses only the market history actually available for each coin.

2 · TrainingThe first 70% of available history is used to find better weight combinations.

3 · ValidationThe final 30% is kept unseen until the model is tested.

4 · Halving cyclesThe same optimizer is also run separately inside each Bitcoin halving cycle when enough data exists.

Advanced details

Current model vs optimized model

Model	Avg return	BTC alpha	Hit rate	Signals
Current 30/25/20/15/10	—	—	—	—
Optimized	—	—	—	—

Signals—

Worst signal—

Turnover—

Forward horizons

Horizon	Avg return	BTC alpha	Hit rate
Run a backtest first.

Run log

Ready.