Six steps through the setup. All eight bot types use the same layout — only the middle block with the strategy's own settings differs. Once you have done it, you can do it for every type.
Via Trading Bots in the top bar you get to bot management, there click “New bot”. You see eight cards: Grid, Dynamic Grid, DCA, Confluence Engine, Scout Engine, Donchian Channel, Quorum Engine and Trend Guard.
Which one suits you is not written here but in the bot overview — there each has a fact sheet with strengths and with the market phases in which it loses. Read the second part. It is the more important one.
| Field | What you enter |
|---|---|
| Coin | The market the bot should work on |
| Period | From–to for the backtest. The longer, the more market phases it contains |
| Candle TF (backtest) | On which timeframe it calculates |
| Timeframe (entry) | On which level the bot will later decide its entries |
| Name | Freely selectable — take something you will still recognise in three weeks |
For the period, do not take only the last four weeks. A bot that has only seen one single market phase is tuned to exactly that phase — and it is the next one to end.
| Field | Meaning |
|---|---|
| Amount (USDT) | What the bot works with |
| Leverage | Multiplies profit, loss and the risk of liquidation |
| Maker fee (%) | Charged when a limit order sits in the book and gets filled |
| Taker fee (%) | Charged when trading immediately at the market price |
| Slippage (%) | Surcharge for the inaccuracy of execution |
| Take profit / stop loss (%) | Target and stop per trade |
The setup shows you two helper figures: Safe leverage — which leverage would have survived the worst price drop in the chosen period — and Required starting capital. Both are estimates from the past, no guarantee, but they keep you from the obvious mistakes.
One click, and the setting runs against the chosen period. Under Backtest result you find the result after fees, hit rate, profit factor, number of trades and MaxDD% — the deepest drawdown.
Look at MaxDD% and the count first. A result of +40 % is worthless if half the account was gone in between — in reality you would have switched it off before. And a hit rate of 80 % from twelve trades is an anecdote, not statistics.
With Show individual trades you expand the list. It is worth it: often the whole result hinges on two or three outliers — and then you know that the number above means nothing.
Instead of trying by hand, the optimizer calculates the combinations systematically and presents the best fifty as a table. One click takes a row over into the setup.
Depending on market and period, the run takes from a few minutes to almost an hour. It keeps running on the server even if you close the tab.
The one cross-check that is really worth it: take your setting and let it run on three or four other markets. If it roughly holds there, you have probably found something real. If it collapses everywhere except on that one coin, it has memorised that coin's past. In the mobile version this step is called basket evaluation.
Trades with play money on real live prices. The mechanics are the same as live, only without money. This is where you start — and for several weeks, not for an afternoon.
Places real orders on your own MEXC account. Needs access keys stored there. That is real money with real risk of loss — and explicitly your decision.
Once created, it appears as a card in the bots panel on the right: status, result, number of trades and a rolling log of its latest decisions. Stopping and editing works directly on the card.
Not by the profit after three days. The questions that really tell you something:
FoundLux