🧪 THE EXPERIMENT CONTINUES:
We’re continuing to test our bot on Binance Spot, and we came across a finding that significantly changes how we look at scalping:
👉 Getting the direction right isn’t enough.
A bot can correctly detect that the price is starting to rise and still lose money.
Why?
Because there’s a silent barrier that many strategies ignore:
fees + spread + slippage.
In our experiment, the price needed to move approximately +0.19% just to cover the friction of entering and exiting.
So what happened?
The algorithm found small upward impulses:
+0.03%
+0.06%
+0.09%
They looked like good signals, but economically they were useless.
The market moved in the right direction, but not enough to cover the cost of the trade.
That forced us to change the question.
Before, the bot asked:
“Is the price going up?”
Now we want it to ask:
🔥 “Is there enough strength behind this move to make entering worthwhile?”
To answer that, we started studying aggressive buying volume, effective bid movement, the imbalance between buyers and sellers, order book depth, spread, and flow continuity.
And we made another discovery:
When the bid simply held steady, many signals turned out to be false impulses.
When the bid actively moved up, the subsequent price expansion was much greater.
Visually, it’s a small difference, but for an algorithm it can mean the difference between buying at the end of a move and entering while buyers are still pushing the market.
Now the next stage begins: checking whether this new filter works on completely new market data, without reusing the data it was designed with.
Because optimizing based on the past is relatively easy.
The difficult question is:
Does it work when the market hasn’t happened yet?
To be continued.
$BTC $BNB $USDC #Trading #Scalping #TradingBot #Binance #bot