Most beginners set up an AI trading bot the wrong way: maximum leverage, large position size, and the fastest timeframe available. Then they are surprised when the account drops quickly. A bot does not fix bad settings. It amplifies them. The best settings for a beginner are boring on purpose. They keep you in the game long enough to learn what works, and they limit the damage while you are still learning.

There is no single perfect configuration. The right settings depend on your account size, your pair, and your strategy. But there are safe starting points that work for most beginners, and you can adjust them gradually as you gather real results. Start conservative, measure everything, and change one thing at a time.

Risk per trade: start at 1 percent or less

Risk per trade is the most important setting on any bot. It is the amount you are willing to lose if the trade hits its stop loss. For beginners, 1 percent of the account per trade is a sensible maximum, and 0.5 percent is even safer while you are testing. With 1 percent risk, ten losses in a row cost about 10 percent of the account. With 5 percent risk, the same losing streak costs half the account. That is the difference between a bad week and a blown account. Set a daily loss limit too, for example 3 percent. When the bot hits it, it stops for the day. No exceptions.

Position size: let risk decide, not gut feeling

Position size should follow from your risk per trade and your stop loss distance, not from a round number that feels right. If your account is 1,000 dollars, your risk is 1 percent, and your stop loss is 2 percent away from entry, your position size is 500 dollars. Many bots calculate this for you if you enter the risk percentage and the stop distance. Use that feature instead of typing a fixed amount. Avoid fixed large sizes on small accounts. A 500 dollar position on a 1,000 dollar account with no stop loss is not a trade. It is a gamble. Also keep leverage low at first. Two to three times leverage is enough to learn. Ten times leverage turns a normal pullback into a liquidation.

Trade frequency: fewer, better trades win

Beginners often choose the fastest settings because more trades feel like more opportunity. In practice, high frequency means more fees, more slippage, and more exposure to random noise. A bot trading every few minutes on a 1-minute chart will pay a large share of its profits in fees. Start with a higher timeframe, such as 1-hour or 4-hour charts, and a strategy that takes a few trades a day at most. This gives each trade room to develop, reduces fee drag, and makes the results easier to review. If you want more activity, add a second pair rather than speeding up the same strategy. Track the number of trades per day and the average profit per trade after fees. If the average is close to zero, the bot is trading too often for its edge.

Stop loss and take profit: always set both

Never run a bot without a stop loss. Some beginners disable it because they believe the price will always come back. Sometimes it does. Sometimes it falls 30 percent and never recovers. A stop loss is what keeps a single bad trade from ending your account. A good starting point is a risk-to-reward ratio of at least 1 to 1.5, meaning you risk 1 dollar to make 1.50 dollars. That way you can be right less than half the time and still come out ahead. Use a take profit rather than letting trades run forever, and consider trailing stops only after you understand how they behave in choppy markets. Test the stop and target distances in a backtest before going live.

A safe starter configuration to copy

Here is a conservative starting setup for a beginner with a small account: trade one major pair such as BTC/USDT, use the 1-hour timeframe, risk 0.5 to 1 percent per trade, use 2x leverage maximum, set a daily loss limit of 3 percent, and allow no more than 3 to 5 trades per day. Run this on a demo account or with minimum size for at least two weeks. Review the trades every few days. Ask which trades won, which lost, and whether the losses were normal or caused by a setting problem. Only then adjust one setting — for example, the take-profit distance — and test again. Keep a simple spreadsheet of each change and its result. That record is what turns random tweaking into real learning.