Coinvera AI Strategy Lab

What it is

A browser backtester for one kind of question: does this indicator rule, on this coin, on this timeframe, still make money after fees? You type the rule in words — the page's own example is "BTC 4h: buy when RSI crosses above 30. 3% stop, 6% take profit. Last 12 months" — and the Lab turns it into conditions and shows them back as a form to confirm before anything runs. The publisher's guide says anything it could not parse is listed and left out of the test, so the form is the rule that was actually tested, and reading it is not optional.

The form is also where the limits are visible. One coin from a list of 26 USDT pairs (BTC, ETH, SOL and the usual large caps, down to TON and SUI), a timeframe of 1h, 4h or 1d, long or short, entry conditions that must all hold on the same closed bar (or within the last three or five bars), a stop in percent or ATR, a take-profit, an optional maximum holding time, and a test length of 3, 6, 12, 24 or 36 months. The examples cover RSI, EMA and SMA crosses, MACD, Bollinger bands, volume breakouts, Supertrend, Stoch RSI and ADX; no full list of supported indicators is published.

The interface runs in English, Turkish, Arabic and Russian. The same tests are available through a Telegram bot, @CoinveraAjanBot, which replies with the headline numbers and a link back to the Lab.

The publisher's terms and privacy policy name Batuhan Vera, trading as "Coinvera AI", in Istanbul, Türkiye, as seller and data controller. No company, registration number or street address is given; contact is a Gmail address and the Telegram account @Kriptoajani.

Pricing

Read on 2026-11-06. Historical tests in the Lab and the Telegram bot are free and need no account; the Lab states no limit, and the bot limits tests per group per day without saying how many. The one Lab feature behind a paywall is forward tracking: freezing a rule and having each new bar checked against it, with a phone notification when its entry conditions are met. That needs an account and PRO, the plan of the publisher's wider Coinvera AI app.

PRO has no published price. The site says the price and duration are sent in writing on Telegram before payment, PRO is added to the account by hand once the money arrives, and nothing renews automatically. The terms add a seven-day full refund on request and name Shopier and iyzico, Turkish payment providers, as the card route. A new account gets six hours of PRO on the web or 48 on Android, once per device, no card needed.

Data & coverage

Binance USDT-M futures candles — the perpetual, not spot — and the result header states the range and bar count of each run (one BTC 4h test read on 2026-11-06: Nov 6, 2025 – Nov 6, 2026, 2,190 bars). The privacy policy says the history is pulled from the reader's own device, from Binance's and Bybit's public endpoints, and that strategies, results and saved tests stay in the browser's storage and are not sent to Coinvera's server. A share link carries the rule, not the result.

The method is stated on the page, and it is more careful than most free backtesters:

• Signals are read on closed bars only and fill at the next bar's open, so the test cannot see the bar it trades on.
• Stop and target are checked against each bar's high and low; if both fall inside one bar the stop is assumed, and a gap fills at the open.
• 0.05% fee plus 0.03% slippage on each side, a round trip of 0.16%. Funding is not included.
• One position at a time, results as a percentage of a 1x position; no leverage is applied.
• Missing bars are not filled, and their number is shown.

A run reports trade count, win rate, total return, maximum drawdown, average trade, profit factor, time in market and the return of simply holding the coin over the same window, plus the trade list. Two checks sit on top of the numbers. The run is split into halves and the page says when the halves went in opposite directions; and a run under 10 trades is labelled unreliable, 10 to 29 "indicative only", and only 30 or more is tested for an edge. A button reruns the same rule on the large caps — BTC, ETH, SOL, BNB and XRP on the publisher's strategy pages.

Limitations

• No funding. On Binance perpetuals a position held across funding windows pays or receives funding every eight hours or more often, and the Lab charges none of it. For a 1h scalp that is noise; for a rule that holds for a week on a crowded long it can be the whole result.
• One coin, one position, one direction per test. No portfolio, no position sizing, no pyramiding, no long-and-short in one rule, no leverage, and nothing below the 1h bar.
• The parser decides what was tested. A condition it did not understand is dropped, not refused; the confirmation form is the only record of what ran.
• No export and no code. Results cannot be downloaded, and there is no API or scripting layer behind the form; a rule that does not fit the form cannot be tested here at all.
• Three years at most, and only for pairs Binance has listed that long.
• The rest of the product is a different product. The Lab's account prompt offers a PRO trial with "the AGENT scan and detailed coin analysis". In the publisher's own words the AGENT "ranks 20 coins under long, short and avoid", and the app's coin pages show plans with "entry, stop and targets". The publisher says Coinvera AI "is not a signal service" and its terms rule out investment advice; this catalogue judged only the backtester, and lists the app nowhere.

Alternatives

Backtesting.py and Jesse are the code route to the same question, and the only way to add funding yourself. QuantConnect Cloud ships Binance funding-rate datasets and a free plan with unlimited cloud backtests, at the cost of writing Python or C#. Coinrule has a no-code rule builder too, but its free plan runs no backtests, the $19.99 plan five in a lifetime, and the product is the live bot the rule turns into.