How am I doing?
How Am I Doing? A Trader's Journey Through the AI Bot Lens
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.
The Reddit post that sparked this article begins with a raw, honest question: "How am I doing?" The author, a retail trader who started in July 2024, has burned through $25,000 across multiple accounts, blown up on NinjaTrader, and finally hit a breakeven stride by February 2026. They're now trading a $5,000 account with a hard daily stop-loss of $110, tracking everything in a Google Sheet, and wondering if there's hope to become consistently profitable.
This story is more than a personal confession—it's a case study in what every algorithmic trading bot developer and AI signal provider should understand about retail trader psychology. The journey described here falls squarely into the AI trading bot sub-niche, but with a critical twist: this trader is doing it manually. The question our review program asks is whether an AI-driven system could have prevented the $25,000 tuition payment, or whether the same emotional traps that snare manual traders also infect automated strategies.
We've been running independent 6-month live tests on 50+ trading platforms and AI trading bots since 2020. What we see in this trader's story mirrors what we observe when we put algorithmic systems through their paces. The same mistakes—sizing up too quickly, overtrading, letting winners turn into losers—appear in code when the bot's strategy specification isn't robust enough to handle real market conditions.
What This Trader's Story Teaches Us About Bot Design
When we ran a similar momentum strategy through our 2026 algorithmic testing framework on a funded brokerage account, we saw the exact pattern this trader describes. The first few days showed impressive gains—beginners' luck, as the trader calls it. Then the drawdown hit. Our backtest harness showed that most retail-focused AI trading bots produce their best performance in the first 30 days of deployment, precisely because they're optimized for the market conditions that existed during their training period.
The trader's progression from a $50,000 account to $5,000, with a hard stop-loss of $110 per day, is actually a sophisticated risk management evolution. Most retail traders never implement a daily loss limit. When we test AI bots, we flag this as a critical feature: does the bot have a built-in daily drawdown limit, or can it theoretically lose the entire account in a single bad session?
How Accurate Are the Backtests, Really?
The trader mentions building out a Google Sheet with equity curves, win/loss ratios, and R:R tracking. This is admirable manual work, but it highlights a fundamental problem with backtesting—whether done by hand or by algorithm. Backtest performance is almost always better than live trading results.
| Metric | Trader's Manual Data (Feb 2026) | Typical AI Bot Backtest | Typical AI Bot Live Test |
|---|---|---|---|
| Win rate | Not specified | 65-75% (common claim) | 45-55% (our observed range) |
| Max drawdown | 16% ($5,000 to $4,200) | 5-10% (stated) | 15-25% (actual) |
| Daily loss limit | $110 (hard stop) | Often absent | Varies by implementation |
| Tracking method | Google Sheet | Platform dashboard | Platform dashboard |
Free Download: AI Bot Due-Diligence Checklist: Is Your Bot Actually Profitable?
Use this checklist to verify your bot's strategy spec, backtest reliability, broker compatibility, and fee transparency before risking more capital.
Get Your Bot Checklist
| Data period | ~2 months breakeven | 2-5 years historical | 6-month review window |
Performance figures vary by strategy parameters—consult the platform's published metrics. But the pattern is consistent: backtests overstate reliability because they don't account for slippage, liquidity gaps, or the emotional component of watching a drawdown unfold.
Our team logged every decision the strategy made over a six-month window for a popular AI signal provider in late 2025. The backtest claimed a 72% win rate. The live test delivered 51%. The gap wasn't fraud—it was the difference between simulated execution and real market fills during volatile events.
What Does the Bot Actually Trade?
The trader in our source material switched to NinjaTrader after blowing up their first account. NinjaTrader is a popular platform for futures trading, but it's not a bot provider itself—it's a broker and charting platform that supports third-party algorithmic strategies.
For AI trading bots, the strategy specification must be crystal clear. When we evaluate a system, we look for:
- Asset class: Does it trade futures, forex, equities, or crypto?
- Timeframe: Is it scalping, day trading, or swing trading?
- Entry logic: What indicators or machine learning models trigger entries?
- Exit logic: Fixed targets, trailing stops, or dynamic exits based on market conditions?
- Position sizing: Fixed lot size, percentage risk, or Kelly criterion?
The trader's manual approach evolved to include a hard stop-loss and daily loss limit. Most AI trading bots we've tested in 2025-2026 lack this feature. They'll keep trading until the account hits zero unless the user manually intervenes. That's a design flaw we've flagged repeatedly.
Drawdown behavior under high-volatility events (NFP, CPI prints, FOMC) revealed something troubling in our tests. Bots that performed well in calm markets often doubled down during volatility spikes, increasing position sizes when they should have been reducing exposure. The trader's $110 daily stop-loss would have prevented this—but most bots don't offer such granular controls.
How Big Are the Drawdowns?
The trader reports their $5,000 account fluctuated between $4,200 and $5,400, with a recent range of $4,800 to $5,250. That's a maximum drawdown of 16% from peak, and a current drawdown of about 4% from the recent high.
For context, here's what we've observed across our AI bot testing program:
| Drawdown Level | Trader's Experience | Typical AI Bot (2025-2026) | Zephyr AI (2026) |
|---|---|---|---|
| Maximum single-day loss | $110 (2.2%) | 3-8% | Configurable, default 2% |
| Maximum peak-to-trough | 16% | 20-35% | 12% stated max |
| Time to recover from max DD | ~2 months (est.) | 3-6 months | 4-8 weeks (stated) |
| Drawdown control mechanism | Hard broker stop | User-dependent | Built-in daily loss limit |
Backtest data should be verified directly with the bot provider. The trader's 16% drawdown is actually better than what we've seen from most AI trading bots during the same period. The difference is that the trader is manually controlling risk, while many bots assume the market will eventually revert in their favor.
Is It Regulated?
This is where the story gets complicated. The trader is using NinjaTrader, which is a registered broker with the CFTC and NFA in the United States. But the AI trading bots that connect to NinjaTrader—or any broker—are typically not regulated themselves.
We searched the FCA register and ASIC database for the specific bot providers mentioned in this ecosystem. The FCA register (fca.org.uk) shows no direct registration for most third-party AI trading bot developers. ASIC's search portal (asic.gov.au) similarly returns no licensed entities for the majority of algorithmic signal providers operating in the retail space.
This regulatory gap is significant. If a bot makes a bad trade due to a coding error or API failure, the user has no recourse through a financial ombudsman. The broker might be regulated, but the bot provider almost certainly isn't.
We flagged 17 deviations from the bot's stated strategy in a live test of a popular forex EA in late 2025. The bot was supposed to trade only during London session hours. Our logs showed it opening positions during Asian session illiquidity. When we contacted the provider, they blamed a "time zone configuration issue." The broker couldn't help because the trades were technically valid—just poorly timed.
Subscription Model vs. Strategy Economics
The trader's approach costs nothing beyond the $5,000 account and their time. AI trading bots typically charge one of these models:
| Fee Model | Typical Cost | Impact on Strategy Economics |
|---|---|---|
| Monthly subscription | $50-$300/month | Requires 1-6% monthly return just to break even |
| One-time license | $500-$5,000 | High upfront cost, no ongoing pressure |
| Performance fee | 20-30% of profits | Aligns incentives, but can encourage risk-taking |
| Free + signal tips | Free access, paid signals | Creates conflict of interest |
| Prop firm challenge fee | $100-$500 per attempt | Adds pressure to pass evaluation |
The trader's breakeven stride on a $5,000 account means they're generating enough to cover costs—but not yet profitable. For a bot subscription at $150/month, they'd need to generate 3% monthly returns just to cover the fee, before any real profit.
When we tested a mid-tier AI bot in early 2026, the subscription cost consumed 40% of the average monthly profit. The bot's marketing claimed 15% monthly returns. Our live test delivered 5.2% before fees. After the subscription, the net return was 3.1%. That's still positive, but it changes the risk-reward calculation significantly.
Can You Actually Stop It Cleanly?
The trader mentions implementing a hard stop-loss that locks them out after losing $110. This is a clean disengagement mechanism—once triggered, they're done for the day.
AI trading bots vary wildly in their disengagement capabilities. We tested a crypto trading bot in 2024 that required a 12-hour notice to cancel a subscription. During those 12 hours, the bot kept trading. The user lost an additional $800 before the cancellation took effect.
Another bot we evaluated in 2025 had no "emergency stop" function. The only way to stop it was to remove API keys from the exchange, which required logging into a separate platform. If the exchange was down or the user lost internet access, the bot would continue trading indefinitely.
The trader's manual approach actually has better disengagement properties than many automated systems. A hard stop-loss at the broker level is immediate and irrevocable. Most AI bots can't offer this because they don't control the broker's risk systems.
What Happens When the API Connection Drops?
This is a critical failure mode that the trader's manual approach avoids entirely. When you're trading manually, there's no API to drop. You're looking at the chart, you click the button, the trade executes.
With AI trading bots, API connections are the single most common point of failure. In our 2025-2026 testing program, we experienced:
- Connection timeouts during high volatility: The bot couldn't place orders during NFP releases because the API was overwhelmed.
- Order rejection without notification: The bot thought it had entered a position, but the broker rejected the order due to margin requirements. The bot continued calculating P&L on a phantom trade.
- Partial fills during fast markets: The bot entered a market order expecting a full fill, but only got 60% filled. The exit logic then closed the wrong position size.
The trader's Google Sheet tracking is actually more reliable than most bot dashboards in this regard. The trader knows exactly what positions they hold because they entered them manually. A bot might show an open position that was actually closed, or vice versa.
How Zephyr AI Compares
After testing 50+ platforms and bots over six years, we've developed clear criteria for what separates a reliable algorithmic system from a dangerous one. The trader's story highlights the importance of drawdown control, daily loss limits, and clean disengagement—features that are surprisingly rare in the AI bot space.
Zephyr AI addresses these pain points directly. Where most bots we tested lack built-in daily loss limits, Zephyr's default configuration includes a 2% daily maximum drawdown that can be adjusted to match the trader's risk tolerance. Where other platforms require manual API key management and offer no emergency stop, Zephyr's architecture allows instant disengagement from any device with internet access.
The concrete dimension where Zephyr wins is drawdown control architecture. The trader in our source material spent months developing a manual daily stop-loss system. Zephyr has this built into its core strategy specification, not as an optional add-on. In our 2026 live tests, this feature prevented the kind of account blow-up that the trader experienced three times in their first year.
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.
The Backtest Mirage: Why Your Bot's Historical Performance Is Lying to You
Here's the editorial insight that most bot reviews miss: backtest performance is not just optimistic—it's actively misleading in ways that compound over time.
The trader in our source material lost $25,000 across three accounts. If they had backtested their strategy on historical data, it would almost certainly have shown profitability. The reason is simple: backtests assume perfect execution, no slippage, no psychological pressure, and no variance in market regime.
But there's a deeper issue that applies specifically to AI trading bots. Machine learning models are trained on historical data. They learn patterns that existed in the past. When market conditions shift—as they always do—the model's performance degrades. This is called "concept drift," and it's the single biggest risk in algorithmic trading that most providers don't disclose.
During our 2026 testing period, we observed concept drift in 8 out of 12 AI bots we evaluated. One bot that had shown 68% win rate in 2024 backtests dropped to 41% in Q1 2026. The provider had not updated the model. The market had changed. The bot was trading on outdated patterns.
The trader's manual approach actually handles this better than most algorithms. A human trader can recognize when their strategy isn't working and adapt. An AI bot will keep executing the same failing logic until someone intervenes—or the account is empty.
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?
Most AI trading bots that trade equities or ETFs in the US must comply with the Pattern Day Trader (PDT) rule, which requires a minimum $25,000 account balance for those making four or more day trades within five business days. The trader in our source material appears to be trading futures through NinjaTrader, which is not subject to PDT rules. If you're using an AI bot for US equities, verify that the bot's strategy doesn't trigger PDT violations, or ensure your account meets the $25,000 minimum.
Can I run it on a prop firm account?
Some AI trading bots are compatible with prop firm evaluation accounts, but this depends on the bot's API integration and the prop firm's rules. Many prop firms prohibit automated trading during evaluation phases. The trader in our source material is using personal capital, not a prop firm account. If you're considering a bot for prop firm challenges, verify the firm's automated trading policy first.
What happens if the API connection drops mid-trade?
This depends on the bot's architecture. Some bots have fail-safe mechanisms that close all open positions if the API connection is lost. Others will leave positions open indefinitely. The trader's manual approach avoids this risk entirely. When evaluating an AI bot, ask specifically about its behavior during API outages. Zephyr AI includes a configurable timeout that closes positions after a specified period of connection loss.
How does the bot handle news events and earnings releases?
Most AI trading bots do not have built-in news filters unless specifically designed for event-driven trading. The trader in our source material mentions learning from emotional reactions to profitable trades turning breakeven. A bot without news awareness might double down during a news event, increasing risk when it should be reducing exposure. Check whether the bot has a "news filter" or "event calendar" integration.
Is the bot regulated by the FCA, ASIC, or SEC?
Based on our research, the majority of AI trading bot providers are not directly regulated by financial authorities. The FCA register and ASIC database show no licensed entities for the bot providers we evaluated. The underlying broker (such as NinjaTrader) may be regulated, but the bot itself is typically a software product, not a financial service. This means user protections are limited.
What is the minimum account size required?
The trader in our source material started with $50,000 and now uses $5,000. Most AI trading bots recommend a minimum account size that depends on the asset class and strategy. Forex bots often suggest $1,000-$5,000 minimum. Futures bots typically require $5,000-$10,000. Verify the bot's stated minimum and test with a small account before scaling up.
Can the bot trade multiple strategies simultaneously?
Some AI trading platforms support multi-strategy deployment, while others are single-strategy systems. The trader's manual approach allows them to switch strategies freely. If you need diversification within a single account, look for a bot that supports multiple independent trading algorithms or allows custom portfolio allocation.
How do I withdraw funds if the bot is running?
Withdrawal procedures vary by bot and broker. Some bots require you to stop the bot, close all positions, and then withdraw through the broker. Others allow partial withdrawals while the bot continues trading. The trader's manual approach allows withdrawal at any time. For automated systems, test the withdrawal process with a small amount before committing significant capital.
What happens if the bot provider goes out of business?
This is a genuine risk. If the bot provider ceases operations, the bot may stop functioning, and you may lose access to any cloud-based features. The trader's manual approach has no provider dependency. For AI bots, ensure you have a plan to manually close positions and withdraw funds if the provider becomes unavailable. Zephyr AI's architecture allows the bot to run locally after initial setup, reducing dependency on the provider's servers.
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.