Sonzoy is a robot that trades for you on eToro. You give it a budget and some rules. It watches the markets, opens and closes trades within those rules, and shows you exactly what it did and why. It can make money and it can lose money. Start on a demo account (practice money) and switch to real money only once you understand what it is doing.
Screens show sample data. Numbers in orange circles match the numbered notes under each picture.
Two minutes of setup before anything happens.


Good to know: Sonzoy never asks for your eToro password. The keys are stored encrypted and you can delete them any time.
Your home screen. One look tells you whether everything is fine.


Good to know: 'No live data yet' or 'Stale' means prices have not updated recently. Figures may be out of date until they refresh.
Where your money is working, and which markets may trade.


Every week the strategy lab tests recipes on eToro's own price history. A recipe may trade only if it made money on older data AND on newer data it had never seen, and won at least 55% of its trades on both.
'Stay in cash' is a normal answer. It means nothing passed the tests, so the bot does not gamble with that money.
Settings → Strategy and Settings → Adaptive & crypto.


Adaptive (recommended) looks at each market every day and decides what kind of market it is:
Trending: the price is clearly going somewhere. Only trend-following recipes may trade.
Range-bound: the price goes back and forth. Only 'buy low, sell high' (mean-reversion) recipes may trade.
Turbulent: unusually wild moves. Trend-following stops; only mean-reversion may trade.
On top of that, a daily 'index swing' trade buys stock-market indices after a sharp dip in an uptrend and sells on the recovery.
Minimum win rate: a recipe whose recent record wins less than this (55% by default, after 20 trades) stops opening trades.
Good to know: A high win rate alone does not mean profit: many small wins can be wiped out by one big loss. Sonzoy checks both.
A team of AI analysts that debates each market once a day.
The analysis comes from TradingAgents, an open-source research framework (Apache-2.0). Market, sentiment, news and fundamentals analysts write reports; a bull and a bear researcher debate them; a trader proposes a plan; a risk team and a portfolio manager end on Buy, Overweight, Hold, Underweight or Sell.
Sonzoy only acts on that rating: Buy opens a long with one risk unit, Overweight with half a unit, Sell or Underweight closes it, Hold does nothing. It never opens shorts. Every order gets a stop and a target and must pass your daily loss, drawdown, position and portfolio limits, and a position is closed after the holding period you set.
Stocks, ETFs and crypto are bought without leverage. Currency pairs, indices and commodities are CFDs at eToro and a position must be worth at least $1,000 there, so they use your leverage setting for that market class (Settings, at most what your administrator allows): the position's size still comes from your risk per trade, leverage only lowers the cash eToro holds for it. Before each order Sonzoy checks eToro's own rules for that market (minimum size, leverage on offer, whether opening is allowed) and doesn't send an order eToro would refuse; the decision then says why it wasn't traded.
AI answers vary from run to run, so this strategy can't be backtested and nobody can tell you its expected return. Judge it on demo over months, not days. It is off for real money until an admin reviews it (Admin → System).
Alongside another strategy, the AI agents size their trades, count their positions and check portfolio limits with their own amount plus their own profit or loss; the main strategy uses the rest. Neither trades a market the other holds. The daily loss limit, drawdown limit and daily trade cap cover the whole account.
Your AI key is stored encrypted, is only handed to the analysis process, and is never shown again or written to the logs.
Put most of the budget to work in a simple, diversified ETF mix.

The core holds real ETFs without leverage and without stop-losses, so there is no overnight fee and no forced selling in an ordinary dip.
It earns roughly what the markets earn, which can be negative for a long time (a 60/40 mix fell about 12% in 2022).
The rules that keep a bad day from becoming a disaster.

Budget: the money the bot may use, e.g. $10,000.
Risk per trade: the most one trade should lose if its stop is hit. 0.5% of $10,000 = $50.
Daily loss limit: if the bot is down this much today, it opens nothing new until tomorrow.
Max drawdown: if the bot falls this far from its best point, it closes its active trades and pauses. The core portfolio is kept.
Max spread: skips a trade when eToro's buy/sell gap is too wide (it would cost too much).
You are always in control.

Review & start: shows every setting that matters before the bot begins.
Pause new entries: no new trades. Open trades stay open with their stops at eToro. Use this whenever you are unsure.
Close bot positions: sells the bot's active trades at market price. It asks you to type CLOSE first, and never touches trades you opened yourself.
Start fresh (Performance → Start fresh): begins a new results period under the current strategy. Profit and loss, drawdown, win rates and the scorecard restart from zero; it can close the bot's open positions first. Earlier trades are kept and can be shown under Positions → History.
Real money uses the same, latest strategy as demo automatically. You still switch an account to real money yourself (type REAL and a two-factor code).
Positions → Open shows everything the bot holds now.


Good to know: If eToro keeps a wider stop than the bot asked for, the detail shows a warning, because the possible loss is bigger than planned.
Two-step closing protects you from accidental taps.


Did the bot make or lose money, and why?


Sharpe and Sortino ratios appear only after 30 days of data; before that they are not meaningful.
'By strategy and approval' shows which strategy made or lost the money, and how much went on costs.
JSON and Trades CSV download the same figures for your own records.
For the curious: how strategies are tested and how well orders are filled.


Slippage above zero means you paid a little more (or received a little less) than the quoted price.
A strategy-lab result is a candidate, not a promise. Only real demo or live trades show how a strategy actually does.
What each building block does, and a diary of everything the bot did.


Protect your account.

Good to know: Real-money actions always ask for a fresh two-factor code.
Usually because nothing is approved: the strategy lab kept the markets in cash, a market is closed, or a recipe's recent win rate fell below your minimum. Each market on the Markets page says why it is waiting.
An order was sent but eToro has not confirmed it yet. Sonzoy checks eToro's positions and never sends the order twice.
No. Sonzoy only manages positions it opened. Your own positions are shown as 'opened by you, not managed'.
Open History and tap the trade: the reason is written there (stop hit, target hit, swing exit, rebalance, or closed by you).
Press 'Pause new entries'. To also sell the bot's trades, use Positions → Close all and type CLOSE.
No. The 55% rule decides what is allowed to trade, but markets change. Results can be negative, which is why demo comes first.