Tidewall Bot Verifier

Do you test a trading bot before you buy it?

A vendor's backtest shows one result, on one period, chosen by the vendor. Before you pay for an Expert Advisor, or before you put a bot on a live account, you need to know how it behaves when conditions are less favourable: a different period, a little slippage, a bad sequence of trades.

Tidewall Bot Verifier answers that question. It reads your Strategy Tester reports, and the vendor's report if you have it, then gives the bot a score out of 100 and a clear verdict: PASSED, AT RISK or DANGEROUS. Everything is explained in plain words, so you can compare several bots and choose with facts rather than screenshots.

The script does not trade and does not change anything on your account. MetaTrader 5 can stay open while you work.

Three questions it helps you answer

• Is the bot hiding a dangerous money management, such as a martingale or a grid?
• Do the results hold up outside the period shown by the vendor, and with real trading costs?
• How much capital do you need to survive a bad series of trades?

Detection of dangerous behaviour

• Martingale: lot size increased after losses.
• Grid and averaging down: positions added with growing lots, many positions open at the same time.
• Very high win rate that hides large losses.
• Floating losses left to run: equity drawdown much larger than balance drawdown.
• Stop out during the test.

Reality check against the vendor

• Recent period compared with the full period, to spot a bot over-optimised on the past.
• Your results compared with the vendor's report or advertised figures: annual gain, drawdown, profit factor.
• Warnings when the vendor shows a short period, low data quality or a different symbol.
• Commission: many testers charge none. Enter your real commission per trade and the script removes it from the results.

Robustness tests on your list of trades

• Slippage: profit factor after a cost is added to every trade.
• Monte Carlo: 5000 random orders of the same trades, giving the drop to expect 1 time in 20 and 1 time in 100.
• Random year: share of years that end with a loss.
• Without the 5 % best trades: does the bot still make money, or does it depend on a few lucky days?

The report also gives the capital needed to absorb the "1 time in 100" drop without exceeding a 30 % drawdown.

Compare and rank your bots

Each analysis is added to a ranking of every bot you have tested, from the most reliable to the most dangerous, and to a CSV file that opens in Excel.

How to use it

• Run the script in PREPARE mode. It writes tests_to_run.txt with the exact Strategy Tester settings for each test: main test, recent period and stress test with random delays. For long real-tick tests it can split the period into slices of a few months.
• Run each test, then save the report: Backtest tab, right-click, Report, HTML, into MQL5\Files\TidewallBotVerifier.
• Optional: copy the vendor's MT4 or MT5 report into the same folder.
• Run the script in ANALYSE mode. You get the verdict on screen, a detailed report file and the updated ranking.

Main inputs

• Language of messages and reports: English or French.
• Bot name, initial deposit, round-turn commission per trade, slippage per trade (0 = automatic).
• Tester settings for PREPARE: bot path, symbol, timeframe, dates, modelling (every tick, 1 minute OHLC or real ticks), slice size.
• Vendor claims for ANALYSE: report file name, advertised gain, drawdown and profit factor.

Requirements

• MetaTrader 5 with the terminal language set to English (View, Languages, English), because the script reads the English labels of the reports.
• Reports saved as HTML from the Strategy Tester. MT4 reports from vendors are also read.

Limits

A backtest measures past risk. It does not predict future gains. Results depend on your broker's data and costs. The score is a decision aid and does not replace the follow-up of a live account over several months.