Methodology

How it actually works

No magic, no opaque "AI", no insider edge. Just real Binance market data, a strategy we can describe in one paragraph, and a simulator that tried every sensible combination of parameters before we ever placed a real order.

Step 1

We start with the same data the pros use.

Every price the bot sees, every candle the simulator replays, every backtest stat on this site — they all come from Binance's public market-data API. No third-party feeds, no aggregators, no cleaned-up "marketing data". The exact same endpoint your favourite charting site is pulling from.

For backtesting we download two full years of 5-minute candles per coin — roughly 210,000 data points each. The live bot calls the same endpoint every minute to get the most recent candles before deciding whether to trade.

The endpoint is documented and rate-limited like any public API — you can hit it yourself right now from a browser.

Binance public REST
Endpoint
GET https://api.binance.com/api/v3/klines
Example call (DOGE, 5-min candles)
?symbol=DOGEUSDT
&interval=5m
&limit=100
What it returns

For every 5-minute window: open price, high, low, close, trading volume, and the candle's timestamp. That's all we need.

View Binance's official docs

Parameter search grid

864 combinations
  • Buy threshold (1h % gain) 1.0, 1.25, 1.5%
  • Take profit 3, 5, 7, 9%
  • Fixed stop loss -1, -2, -3, -4%
  • Trailing stop -0.5, -1, -2%
  • Short SMA period 20, 30, 40 candles
  • Trend SMA period 50, 100 candles

Every combination is replayed candle-by-candle against the full 2-year history. We rank by total P&L but require at least 10 closed trades to avoid lucky outliers.

Step 2

We try every sensible setting — then pick the one that worked.

A trading strategy is only as good as its parameters. Pick a take-profit that's too tight and you exit too early; too loose and winners turn into losers. So we don't pick by hand. The simulator runs the strategy across a grid of 864 parameter combinations against two years of real candles, scoring each combo on closed P&L, win rate, and trade count.

Crucially, the simulator uses the same trading rules as the live bot — same entry condition, same exit hierarchy, same Binance fee (0.075% per side), same slippage assumption (0.1%). What you see in the backtest is what you'd actually have netted.

Once the winning combo is found for a coin, those exact numbers are written into the bot's config row and the live bot picks them up within sixty seconds — no redeploy needed.

2 yrs
of 5-min candles
864
param combos
~210k
candles per coin
Step 3

The strategy, in plain English.

The bot is looking for a coin that has just started climbing while the broader trend is still up. It rides the move with a tight trailing stop and a hard cap on losses. That's it.

Entry (BUY)

Two conditions must both be true on the latest candle:

  1. 1 Short-term momentum — price is up more than the buy threshold over the past hour. (sma_short − price_1h_ago) / price_1h_ago × 100 > threshold
  2. 2 Trend confirmation — the short SMA is above the long trend SMA, so we only buy into uptrends. sma_short > sma_trend

Exit (SELL)

Three independent triggers, checked in this priority order:

  • TP Take profit — price is up by the take-profit % since entry. Lock in the win.
  • TRAIL Trailing stop — armed only once the trade is up +1%. After that, if price drops by the trailing-stop % from its peak, sell. Lets winners run without giving back the gain.
  • SL Fixed stop loss — if price drops past the stop-loss % from entry, cut the loss immediately. No averaging down, no hoping.

Position sizing

When the entry condition fires, trade size is computed dynamically from the bot's pool, so as the pool grows from profits and new deposits, trade size grows too:

quantity = (total_pool_usdt × trade_allocation_percent) ÷ current_price

A buffer (the remaining 1 − allocation share, typically 20%) stays uninvested as a cushion against drawdowns. Quantity is rounded down to Binance's LOT_SIZE step for the symbol.

Step 4

The exact numbers we're trading with — right now.

These values are read live from the bots themselves. Anything an admin changes is reflected here within a minute.

Bot Candle SMA short / trend BUY threshold Take profit Stop loss Trailing Alloc %
ADA Bot
ADAUSDT
5m 20 / 50 +0.5% +3% -4% -0.5% 80%
DOGE Bot
DOGEUSDT
5m 20 / 50 +0.5% +3% -4% -0.5% 90%
ETH Bot
ETHUSDT
5m 20 / 50 +0.75% +3% -1% -1.5% 80%
LTC Bot
LTCUSDT
5m 20 / 50 +0.5% +3% -4% -0.5% 80%
SOL Bot
SOLUSDT
5m 20 / 50 +0.5% +3% -4% -0.5% 80%
SUI Bot
SUIUSDT
5m 20 / 50 +0.5% +5% -3% -0.5% 80%
Trend SMA is the longer moving average that defines the underlying direction. Short SMA is the faster one that triggers the entry. All numbers above are reloaded from the database every 60 seconds by the live bots.
Step 5

What we will not pretend.

Backtests are useful, but they are not the future. Here are the things we want every user to read before depositing a single dollar.

Past ≠ future

A strategy that won across 864 combinations on two years of historical data has been tuned to those exact two years. Future market behaviour can differ. A losing month is not just possible — it is expected.

Crypto is volatile

Even with a hard stop loss in place, a sudden gap-down between candles can fill below the stop and turn a planned -2% loss into something larger. We size positions conservatively because of this.

No leverage, no shorts

The bots only ever buy and sell spot — they never borrow, never short, never use derivatives. The worst case for any single trade is a stop loss, not a margin call.

Pool, not principal-guaranteed

Your balance moves with the bot pool. When the bot has a winning month, your share grows. When it has a losing month, your share shrinks. We do not — and cannot — guarantee returns.

Ready to put it to work?

Browse the bots, see their live performance, then allocate when you're comfortable. You can withdraw your share once the 4-week lockup is over.