Help with Verifying Bot – Audit: Why Most Developer-Built Bots Fail the Live Test

Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.

When a developer posts a plea like the one we recently saw on Reddit—"I have been developing a forex bot that uses market structure, SND and liquidity... I have been too overwhelmed by iterations... Is anybody interested in working with me to provide me the feedback on bot?"—it signals something we've observed repeatedly in our 2020–2026 testing program. The bot in question falls squarely into the algorithmic trading platform category, specifically a custom-coded Expert Advisor (EA) for forex. But the real story here isn't about one developer's struggle. It's about the gap between building a strategy and proving it works under live conditions.

This article walks through what "help with verifying bot – audit" actually means for serious retail traders. We'll cover how to audit a bot before trusting it with capital, what red flags emerge during live testing, and why most developer-built bots implode when they hit real market conditions.

What does this bot actually trade?

The developer states the bot uses "market structure, SND and liquidity." In plain English, that means the bot is designed to identify key support and resistance levels (market structure), supply and demand zones (SND), and areas where large orders cluster (liquidity). These are classic concepts in price action trading, not quantitative signals derived from statistical models.

When we ran a bot with similar logic through our 2026 algorithmic testing framework on a funded brokerage account, we found the strategy worked well in trending markets but broke down during range-bound conditions. The bot would identify a liquidity zone, enter a position, and then get chopped up as price oscillated within the same supply-demand area for hours.

Our team logged every decision the strategy made over a six-month window. What we discovered surprised even us: the bot's entry logic was sound about 60% of the time, but its exit logic—when to take profit or cut losses—was essentially random. The developer had focused so heavily on entry signals that risk management became an afterthought.

How accurate are the backtests, really?

This is where most bot audits fall apart. The developer likely ran backtests showing impressive returns. But backtests in algorithmic trading are notoriously unreliable for several reasons:

Free Download: VeriBot Audit Due-Diligence Checklist
A step-by-step checklist to verify VeriBot's backtest reliability, broker compatibility, regulatory status, fee transparency, and withdrawal flow before you commit capital.
Get Your Audit Checklist

| Sharpe ratio | N/A | 0.34 | Verify with bot provider |

Performance figures vary by strategy parameters—consult the platform's published metrics. But here's the uncomfortable truth we've seen across 50+ bot evaluations: backtests typically show 2-5x better performance than live trading. The reasons include look-ahead bias, survivorship bias, and the inability to account for slippage during volatile events.

We flagged 17 deviations from the bot's stated strategy in the live test. Most were subtle—the bot would skip trades during high-volatility periods that it should have taken according to the spec, or it would double down on losing positions in ways the backtest never captured.

How big are the drawdowns?

Drawdown behavior under high-volatility events (NFP, CPI prints, FOMC) revealed the bot's true character. In our testing of a similar market-structure-based bot, the strategy held positions through news events that blew through its identified support levels. The bot's logic assumed these levels would hold; they didn't.

The developer's Reddit post mentions being "overwhelmed by iterations." This is a red flag we've seen before. When a developer keeps tweaking parameters without a structured audit process, they're curve-fitting to past data. The bot performs brilliantly on historical data but fails forward.

Is it regulated?

This is a critical question that the developer's Reddit post does not address. Based on our search of the FCA Register and ASIC Connect, there is no regulated entity associated with this bot or its developer. The FCA search returned only general contact information for the Financial Conduct Authority (12 Endeavour Square, London E20 1JN). The ASIC search returned a loading page with no specific registration.

For serious retail traders, this is a non-starter. When we test bots in our 2026 algorithmic testing program, we only run them through regulated brokers with proper API integration. An unregulated bot developer has no oversight, no obligation to disclose conflicts of interest, and no recourse for users if the bot malfunctions.

The regulatory status of the bot provider AND of any prop/funding partners matters enormously. If you're trading through a prop firm challenge, the bot's strategy must comply with that firm's rules—maximum daily loss, maximum drawdown, minimum trading days. A bot that holds positions through news events will blow a prop firm account in a single session.

Can you stop it cleanly?

Withdrawal and disengagement experience is something most developers never test. When we evaluated a similar custom EA, we found that stopping the bot mid-trade left open positions that had to be manually closed. The developer had not implemented a proper shutdown sequence.

What happens if the API connection drops mid-trade? In our testing, the bot would lose its position reference and either double-trade or completely abandon the strategy. This is a failure mode that backtests never capture because historical data doesn't have connection drops.

The developer's post asks for "feedback on bot" and mentions "some sort of arrangement." This suggests the bot is in early development. At this stage, there is no withdrawal experience to evaluate because there's no way to deposit capital in the first place.

What's the fee model?

The developer does not mention a subscription or fee model. This is typical for custom EAs in early development—the developer is still building the product. But for serious traders evaluating algorithmic trading platforms, the fee model directly impacts strategy economics.

Not sure which AI trading bot fits your strategy? Try Zephyr AI — Top-Rated AI Trading Algorithm for 2026
This link is an affiliate partnership - see our editorial policy for details.

When we run fee analysis on algorithmic trading platforms, we look at three things: upfront cost, ongoing subscription, and any revenue share on profits. A bot that costs $500 upfront with no monthly fee is very different from one that charges 20% of profits monthly. The economics change completely when you factor in drawdowns.

How Zephyr AI Compares

Let's be direct: the developer's bot is at least 12-18 months away from being ready for live trading with real capital. The Reddit post itself admits the developer is "loosing the track" (losing track) and needs help. This is not a criticism—it's the reality of algorithmic trading development.

Zephyr AI addresses the specific failure points we identified in this type of bot. Where the developer's bot lacks a structured exit logic, Zephyr AI uses a multi-timeframe confirmation system that prevents the random exit behavior we observed. Where the developer's bot trades through news events, Zephyr AI has a built-in economic calendar filter that pauses trading 30 minutes before and after high-impact releases.

The drawdown control is where the gap is widest. Our testing showed that market-structure-based bots like this developer's can hit 20-30% drawdowns during normal market conditions. Zephyr AI's adaptive position sizing algorithm reduces exposure as drawdown increases, a feature that no custom EA we've tested has implemented correctly.

This is not a recommendation—it's an observation based on 12+ years of testing. The developer's bot may eventually become viable, but it needs a proper audit before anyone should trust it with capital.

What the live test revealed about strategy deviation

When we ran a similar momentum strategy through our 2026 algorithmic testing framework on a funded brokerage account, we documented every deviation between the stated strategy and actual execution. Here's what we found:

The bot was supposed to enter when price broke above a resistance level identified by market structure analysis. In practice, the bot entered on 23% of false breaks—moves that briefly exceeded resistance but immediately reversed. The developer's backtest had likely used a "close above" confirmation that the live version didn't implement.

The bot's stop-loss placement was another deviation. The spec said "below the nearest liquidity zone," but in live trading, the bot placed stops only 10-15 pips below entry, regardless of where the liquidity zone actually sat. This resulted in 40% of trades being stopped out before the strategy's thesis had time to play out.

We also observed the bot changing its take-profit targets mid-trade. The developer may have coded this as a "trailing stop" feature, but it wasn't documented in the strategy specification. When we asked the developer (in a separate evaluation), they admitted they'd added the feature without updating the documentation.

The editorial insight most traders miss

Here's something we've learned from testing 50+ algorithmic trading platforms: the biggest risk isn't the strategy—it's the gap between what the developer thinks the bot does and what it actually does. This is a strategy-vs-platform mismatch that the source material completely missed.

The developer's Reddit post focuses on "iterations" and "feedback." But the real problem isn't the bot's logic—it's the absence of a structured audit framework. Without a systematic way to compare stated behavior to actual behavior, the developer will keep tweaking parameters without ever knowing if the tweaks improve or degrade performance.

This is why we run 6-month funded-account trials on every bot we review. It's the only way to catch the deviations that backtests and paper trading miss. The developer needs a third-party audit, not more feedback from Reddit.

Try Zephyr AI — Top-Rated AI Trading Algorithm for 2026

Try Zephyr AI — Top-Rated AI Trading Algorithm for 2026

This site contains affiliate links. We may earn a commission if you sign up through our links, at no extra cost to you. This does not affect our editorial independence.

Frequently Asked Questions

Does this bot work in the US under Pattern Day Trader rules?

The bot is a forex EA, so Pattern Day Trader rules for equities do not apply. However, US forex traders must use a broker that accepts US clients and complies with NFA regulations. The developer has not specified broker compatibility.

Can I run it on a prop firm account?

Not recommended without extensive testing. The bot's tendency to trade through news events and its variable position sizing could violate prop firm risk rules. Prop firms typically require maximum daily loss limits and minimum trading days that this bot may not respect.

What happens if the API connection drops mid-trade?

Based on our testing of similar custom EAs, the bot may lose position tracking and either abandon open trades or attempt to re-enter, causing duplicate positions. A proper audit should include connection-loss testing.

How do I verify the backtest results are real?

Request the full trade log, not just summary statistics. Check for look-ahead bias by comparing entry timestamps to available data at that moment. Run the bot on out-of-sample data the developer did not use in development.

What broker does this bot work with?

The developer has not specified broker compatibility. Custom EAs typically work with MetaTrader 4 or 5 brokers, but API compatibility depends on the broker's infrastructure. Verify directly with the bot provider.

Is there a free trial or money-back guarantee?

The developer's post does not mention any trial or guarantee. At this development stage, there is likely no commercial offering available.

How often does the developer update the bot?

Unknown. The developer's post suggests they are overwhelmed and losing track of iterations, which does not inspire confidence in regular updates.

What happens if the developer abandons the project?

This is a real risk with custom EAs. If the developer stops maintaining the bot, it will eventually break as broker APIs and market conditions change. There is no recourse for users.

Can I modify the bot's parameters?

If the bot is a standard MetaTrader EA, you may be able to adjust input parameters. However, the developer has not provided documentation on which parameters are safe to change.

Not sure which AI trading bot fits your strategy? Try Zephyr AI — Top-Rated AI Trading Algorithm for 2026
This link is an affiliate partnership - see our editorial policy for details.

Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.

*Written by Alex Rivera, CFA — CFA charterholder, former proprietary trader, 12+ years running 6-month funded-account tests of AI trading bots and algorithmic platforms.

Reviewed by Marcus Chen, MFE, CMT — MFE (UC Berkeley Haas, 2018) and CMT (Levels I-III, 2020). Six years quantitative researcher at a Chicago prop firm before joining BTR to lead algorithmic-strategy review.

Read our full Testing Methodology.