After my Octane bot hit a 42% drawdown, I realized it was no longer valuable for me to continue checking it minute-by-minute. So I deleted it from my phone entirely. Why let it get to me? I’ll take whatever it leaves me with and forego bots for good.
That’s it. I’m done. Not “taking a break” done. Not “reassessing my strategy” done. Actually done.
But first, let me tell you about the email that arrived the same week my account was cratering—because it perfectly encapsulates everything that’s broken about the algorithmic trading industry.
The Quote That Should Be Taped to Every Algo Bot User’s Monitor
“This was not manual intervention or a bug. This is exactly how the system is designed to behave. It’s risk management working as intended.”
That’s from an actual email sent by NURP to their community this week. Their algorithm activated “protective measures” ahead of high-impact AUD news—closing positions, placing hedges, taking controlled losses to prevent catastrophic ones.
I read it while watching my Octane bot bleed through a 42% drawdown.
Here’s what struck me: both systems were doing exactly what they were supposed to do. NURP’s users got an email explaining how sophisticated their risk management is. A peek at my Octane algo bot trading account showed me it was again struggling to stay liquid.
Both companies built smart systems. Both deployed machine learning algorithms. Both implemented protective measures and drawdown limits and automated safeguards.
Both groups of investors are learning the same brutal lesson: sophisticated risk management can still lose unsophisticated amounts of money.
Background: When “Working As Intended” Isn’t Working
I don’t run NURP bots anymore. I closed my last NURP account over nine months ago after three months of losses. But I’m still on their mailing list, still monitoring the industry, still watching what happens when retail investors meet institutional-grade algorithms.
This week’s email was unusually transparent about how modern algo systems operate under stress. Here’s what NURP explained:
Their algorithm has automated safeguards that trigger under two conditions: when individual subsystems approach their drawdown or exposure limits, and when high-impact economic events are scheduled. Last night they had both—some subsystems were already near thresholds, and major AUD news was imminent.
So the system closed positions. Placed hedges. Took small losses to prevent larger ones. Exactly as designed.
Meanwhile, my Octane bot—built by a completely different company, using different algorithms—was doing its own version of the same thing. Managing risk. Protecting capital. Following its programming.
And here I am, down 42%, deleting apps off my phone so I can sleep at night.
The Quote That Explains Everything (And Nothing)
Let me share the key paragraph from NURP’s email, because it’s the most honest thing I’ve ever read from a bot company:
“These systems leverage a form of AI called machine learning, which allows them to analyze market conditions and make dynamic decisions in real time. Unlike rigid rule-based systems that behave identically every time, machine learning enables the algorithm to weigh multiple factors (current drawdown levels, exposure, upcoming volatility events, and broader market context) and respond accordingly. This means protective actions won’t always look the same, because the system is adapting to the specific combination of conditions present at that moment.”
Read that again: “Protective actions won’t always look the same.”
Translation: We can’t predict what it’ll do, but trust us, it’s smart.
This is the paradox at the heart of modern algorithmic trading. The more intelligent the system becomes, the less comprehensible it is to the human funding it. Last time it held through news. This time it bailed early. The system is weighing probability distributions you can’t see, making optimal decisions you can’t verify, protecting capital in ways that somehow still result in 40%+ drawdowns.
When it makes money, you trust the process. When it loses money, you get emails about how the math is too sophisticated for you to understand but it’s definitely optimal.
I’m not too unsophisticated to understand. I’m 42% poorer…sort of. There’s a difference.
The Opportunity Angle (Why This Keeps Happening)
Here’s why the bot industry continues to grow despite regular drawdown disasters: the pitch is irresistible.
What they’re selling: Set-and-forget passive income. Professional-grade algorithms. Institutional risk management. Machine learning that adapts to market conditions. Returns that would make hedge funds jealous, available to anyone with capital and a dream.
What they’re actually delivering: Backtested performance that may or may not translate to live markets. Systems optimized for historical conditions that may not exist anymore. Risk management that prevents total catastrophe while allowing significant pain.
The opportunity is real—there are algorithms making consistent money in Forex. But they’re typically running on institutional infrastructure, with professional risk oversight, funded by capital that can absorb 40% drawdowns without blinking.
Retail versions of these systems face challenges the backtests never capture:
Slippage and execution quality. Your broker’s execution during volatile news events isn’t the same as the backtest’s theoretical fills.
Infrastructure fragility. Professional systems have redundant connections, backup servers, failover protocols. Your bot runs on infrastructure that might hiccup at the worst possible moment.
Capital constraints. Institutions can handle drawdowns because they have deep reserves. Retail traders hit psychological breaking points (or actual margin calls) long before the system’s theoretical recovery period completes.
Regime changes. Markets shift character. Strategies that worked brilliantly in trending conditions fail in choppy markets. The NURP email revealed something crucial: “Some subsystems were already near their thresholds” before the news even hit. These aren’t systems failing—these are systems discovering in real-time that market conditions don’t match their training data.
The email was remarkably honest about this: “Each unique strategy-pair combination (subsystem) has a max risk limit/stop loss where it makes sense to close the trades instead of adding more exposure and risking further drawdown.”
That’s the sales pitch in technical language: we’ve backtested this so thoroughly that we know exactly when to cut losses. What they don’t mention: those loss-cutting triggers can fire repeatedly in choppy markets, slowly bleeding your account through a thousand small “protective measures.”
The Risk Angle (What They’re Not Telling You)
Let’s talk about what “very small chance of catastrophic loss” actually means in practice.
Risk #1: The Optimization Trap
Every bot has been backtested extensively. Parameters tweaked. Entry conditions refined. Stop losses optimized. By the time a system reaches market, it’s been tuned to perform beautifully on historical data.
The problem? Over-optimization creates systems that are exquisitely adapted to past conditions and fragile when facing anything new. You’ve built a Formula 1 race car optimized for a specific track, and now you’re racing it in changing conditions.
The NURP email mentioned their subsystems “approaching drawdown limits”—meaning the real market was pushing boundaries the backtest said were unlikely. When “unlikely” becomes “actual,” you discover the difference between theoretical risk and realized risk.
Risk #2: The Machine Learning Black Box
NURP’s explanation perfectly captures why ML-based systems are particularly problematic for retail investors: you can’t learn from them.
When I make a bad discretionary trade, I can analyze what went wrong. Market condition changed. I misread the setup. I let emotion override logic. I learn. I adapt.
When an ML algorithm makes a bad trade, I get an explanation that it “weighed multiple factors” and made the optimal decision based on probability distributions. I can’t replicate the logic. I can’t verify the reasoning. I just have to trust that somewhere in that neural network, good decisions are being made.
Until they’re not. And then I get emails about how it was actually still optimal, just unlucky.
The NURP email says protective actions “won’t always look the same” because the system adapts to “the specific combination of conditions present at that moment.” That sounds sophisticated. In practice, it means you’re funding a black box that can’t explain its own decisions in terms you can understand or predict.
Risk #3: Emotional Capital Depletion
Here’s the risk nobody discusses: mental and emotional tolerance.
I preach risk management. I allocated only money I could afford to lose. I understood drawdowns were possible. I followed every rule about position sizing and capital preservation.
And still, a 42% drawdown is emotionally devastating in ways the risk calculator never captured.
You can afford to lose money and still not be able to afford the sleepless nights. The constant account checking. The mental math about recovery time. The slow erosion of confidence in your own judgment.
The bots are managing position risk, drawdown limits, and exposure parameters. Nobody’s managing the risk that you’ll lose faith in the entire approach while still having money on the line.
I deleted the app not because I couldn’t afford the financial loss. I deleted it because I couldn’t afford to keep paying attention.
Risk #4: The Fee Burden Reality
Here’s the math that matters:
On a $200,000 account with $200/month in subscription fees, you need just 1.2% annual return to cover costs. That’s manageable, right?
Except that’s not how drawdowns work.
When you’re down 42%, your account is now worth $116,000. Those same $200/month fees now require 2.1% annual return just to break even. The smaller your account gets during drawdown, the higher the percentage return required to justify the fixed costs.
And that’s before trading costs, slippage, and spreads. Every time the system takes those “protective measures”—closing positions, placing hedges, cutting exposure—you’re paying the spread on both sides of the trade.
The fees seem reasonable on the account size you started with. They become punishing on the account size you’re left with after the algorithm “protects your capital.”
Risk #5: Survivorship Bias in Marketing
Every bot vendor shows you their best results. The accounts that started at optimal times. The backtests run on the cleanest data. The client testimonials from people who got lucky with timing.
What you don’t see: the distribution of actual outcomes. The accounts that started during unfavorable conditions. The systems that worked for three years and then forgot how markets work. The investors who followed every rule and still got wrecked.
I’m not uniquely unlucky. The NURP email proves it—their users are experiencing the same thing. “Some subsystems had already started taking controlled losses,” they wrote. That’s corporate-speak for “accounts are bleeding.”
I’m just willing to say out loud what a lot of people are experiencing: these systems aren’t as robust as advertised, and the gap between “protective measures working as intended” and “my account not being down 42%” is where your retirement dreams go to die.
The Bottom Line
After this experience—watching Octane bleed while reading NURP’s email about optimal risk management—I’ve reached a few conclusions:
The systems aren’t fraudulent. They’re sophisticated, built by smart people, and probably doing exactly what they’re designed to do. The problem is that “what they’re designed to do” includes losing substantial amounts of your money while technically preventing total catastrophe.
“Working as intended” is a meaningless phrase. When NURP says their protective measures worked as intended, they mean the algorithm followed its programming. When I say my account is down 42%, I mean the outcome was catastrophic regardless of whether the process was correct. Both statements are true. Only one matters to my retirement account.
Backtesting is a sales document. Not because it’s fake, but because it shows you results without the experience of living through them. A 30% drawdown in a backtest is a number. A 30% drawdown in your account is three months of questioning every decision you’ve made.
Machine learning is a double-edged sword. The adaptability that makes these systems “smart” also makes them unpredictable. You’re not learning alongside the bot—you’re a spectator to decisions you can’t anticipate, understand, or replicate. When it works, that’s fine. When it doesn’t, you’re left with losses and an explanation you can’t use to make better decisions next time.
Risk management protects capital, not conviction. The bots can manage position size and drawdown limits. They can’t manage the psychological cost of watching your account shrink while receiving emails about how sophisticated the underlying algorithms are.
The industry runs on hope and complexity. Every bot vendor shows winning accounts and optimal backtests. Nobody shows the full distribution of outcomes. Nobody admits that timing matters enormously and most retail investors enter at exactly the wrong time. And when accounts struggle, the response is always the same: it’s working as designed, trust the process, the math is too complex for you to understand.
So what am I doing?
I deleted the app. I’m letting Octane finish its run—whatever that means—without watching every tick. When it’s done, I’m done.
Will I try bots again? Maybe. A token amount—$500 or a little more if I am feeling generous—just enough to keep monitoring the industry and warning people about what I’ve learned. While I’m there, I’ll tell anyone who’ll listen: “If you haven’t lost money yet, you just haven’t been made the fool yet.”
Because here’s the real lesson: sophisticated systems operated by smart people can still lose your money. Risk management working exactly as intended can still be emotionally intolerable. And “protective measures” that technically succeed while your account drops 42% aren’t protecting the thing that actually matters—your ability to stay in the game.
Thanks for the education, EFX. Thanks for showing me what “risk management working as intended” actually looks like from the investor’s side.
And to NURP: thanks for the honest email. At least you told your users the truth—the algorithm did exactly what it was supposed to do. It’s not the algorithm’s fault that what it’s supposed to do still hurts a lot.
The party’s over. I’m going home.
Disclaimer: This is not financial, tax, or legal advice. This is one person’s experience with algorithmic trading systems, risk management frameworks, and the gap between technical optimization and practical results. The author has experience with multiple bot platforms including EFX Algo’s Octane system and previously held a NURP account. Past performance—whether in backtests or live trading—does not guarantee future results. Algorithmic trading involves substantial risk of loss. Only invest capital you can afford to lose completely, and understand that “afford to lose” includes both financial and emotional tolerance for significant drawdowns. Consult qualified professionals before risking capital in automated trading systems.
