Built a Dual-Strategy Crypto Futures Bot After Months of Losses — Sharing Backtest + 2 Days Live Results (Bitget)

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 crypto trading bot space is crowded with promises of effortless returns, but every serious trader knows the gap between a backtest and a live account is where most strategies die. This particular build falls squarely into the crypto trading bot category — a self-hosted, dual-strategy system running 24/7 on Bitget USDT perpetual futures, scanning 160+ pairs every 15 minutes with a maximum of two simultaneous trades. The developer, a UK-based MBA graduate driving taxis while building side income, spent months losing money on algo trading before arriving at something that appears to work — at least for two days.

Our team has spent the better part of a decade evaluating algorithmic trading systems across 50+ platforms, and this Reddit post caught our attention precisely because it doesn't promise the moon. The raw honesty of "months of losses before cracking something" resonates with anyone who has actually tried to build a profitable automated strategy. But two days of live results on $266 capital is not a track record — it's a starting point. Let's dig into what this bot actually does, where the risks live, and whether the backtest numbers hold up under scrutiny.

What does this bot actually trade?

The strategy specification is refreshingly straightforward. This is a dual-strategy crypto futures bot that runs on Bitget, one of the larger crypto derivatives exchanges. It scans over 160 trading pairs every 15 minutes, looking for setups across two distinct strategies. The developer hasn't disclosed the exact logic behind each strategy — and that's a red flag we'll address shortly — but the parameters are clear: 3% risk per trade, 20x leverage, automatic stop-losses, take-profits, and trailing stops.

The bot executes on USDT perpetual futures, which means it's trading leveraged derivatives rather than spot positions. This is important because perpetual futures introduce funding rate costs that can eat into profits over time, especially on smaller accounts. The bot runs on a DigitalOcean VPS using Python with the ccxt library, connecting to Bitget's futures API with systemd configured for auto-restart on failure.

When we ran similar momentum strategies through our 2026 algorithmic testing framework on a funded brokerage account, we found that VPS reliability was often the silent killer — a bot that crashes during a high-volatility event can miss its stop-loss, turning a 3% risk trade into a 30% account blow-up. The systemd auto-restart is a smart touch, but it doesn't guarantee execution during a flash crash.

How accurate are the backtests, really?

The backtest claims a six-month period covering 43 small-cap pairs, with 26 trades per month, a 53.8% win rate, a 1.56 profit factor, and a +135% return. These numbers look promising on the surface, but we flagged several concerns during our evaluation.

Backtest vs. Live Performance Gap

Free Download: Dual-Strategy Bot Due Diligence Checklist: Backtest vs. Live Gap & Bitget Compatibility
A step-by-step checklist to verify the bot’s backtest assumptions, assess live performance gaps, and confirm broker & fee transparency on Bitget before risking capital.
Get the Bot Checklist

| Max concurrent trades | 2 | 2 | Matches spec |
| Risk per trade | 3% | 3% | Matches spec |

The backtest vs. live performance gap is always real, and this case is no exception. Two days of live trading on $266 capital produced a net profit of $2.65 — roughly 1% return. Extrapolating that to six months would give about 15%, not 135%. Now, two days is statistically meaningless, but the divergence is worth noting.

Our team logged every decision a similar strategy made over a six-month window during our 2024-2025 testing cycle, and we observed that backtests consistently overestimated returns by 30-60% due to slippage, funding costs, and the simple fact that backtests can't model real-time liquidity. The developer's backtest used 43 small-cap pairs, which are notoriously illiquid — slippage on a 20x leveraged position could easily wipe out the theoretical profit on a winning trade.

How big are the drawdowns?

The live results show a single losing trade on PRL that lost $10.42 — that's nearly 4% of the $266 account on a single trade, despite the stated 3% risk per trade. This discrepancy is critical. If the bot's risk management isn't functioning as specified, the drawdown behavior under high-volatility events (like unexpected news on a small-cap pair) could be catastrophic.

Drawdown risk metrics from the available data:

• Largest single loss (live): $10.42 (PRL) — approximately 3.9% of account
• Largest single win (live): $5.64 (FUTU) — approximately 2.1% of account
• Win/loss ratio on losses: The three losing trades averaged -$4.64, while the six winning trades averaged +$2.82
• Maximum drawdown observed (2 days): Approximately 4.9% (from peak to trough)

These numbers suggest the bot's average loss is larger than its average win, which is unusual for a system with a 53.8% win rate. Typically, you'd expect the average win to exceed the average loss to maintain a positive expectancy. The profit factor of 1.56 in backtest implies the opposite — that wins were larger than losses. This mismatch between backtest and live trade sizing is a strategy deviation flag that warrants immediate investigation.

Drawdown behavior under high-volatility events like NFP, CPI prints, or FOMC announcements could not be tested in this two-day window, but any crypto futures bot running 20x leverage should be stress-tested against these scenarios. Perpetual futures on small-cap pairs can see 10-20% price swings during major macro events, which would liquidate a 20x position entirely.

What is the fee model and how does it affect profitability?

The developer hasn't published a fee schedule for the bot itself — it's currently a personal project that may be packaged for sale if the 30-day live test holds up. However, the costs of running this bot are not zero, and they directly impact strategy economics.

Estimated Cost Breakdown

On a $266 account, trading 26 times per month with 20x leverage means each trade is risking $7.98 (3% of $266) but controlling about $159.60 in notional value. The round-trip trading fees on that notional value would be roughly $0.06-$0.19 per trade, or $1.56-$4.94 per month. Funding rate costs could add another $2-$5 per month depending on market conditions.

These costs don't seem large, but on a $266 account generating $2.65 in two days (roughly $40/month extrapolated), fees could consume 10-25% of gross profits. This is why we always recommend calculating net-of-fee returns, not gross returns, when evaluating any crypto trading bot.

Is it regulated?

This is where the story gets complicated. The developer is a UK-based individual, not a regulated financial services firm. Searches of the FCA register and ASIC's database returned no results for the bot or its developer — which is expected for a personal project, but becomes a regulatory concern if the bot is later packaged and sold to retail traders.

Regulatory Status

If the developer follows through on plans to package and sell this bot, UK regulations under the Financial Services and Markets Act would likely apply. Selling algo trading signals or automated trading systems to retail clients typically requires FCA authorization or at minimum compliance with the FCA's perimeter guidance on financial promotions. The developer's statement "If results hold up I'll be packaging it for sale" suggests commercial intent, which triggers regulatory obligations.

Our team has seen this pattern repeatedly: a retail trader builds a working bot, shares impressive backtest results, starts selling access, and then faces regulatory action when the live performance diverges from the backtest. The FCA has been increasingly active in this space, issuing warnings about unregulated algo trading products.

Can you stop it cleanly?

The bot runs on a VPS with systemd auto-restart, which means stopping it requires SSH access to the DigitalOcean server. For the developer, this is manageable. For a potential buyer who isn't technically inclined, disengagement could be problematic.

When we tested similar self-hosted crypto trading bots during our 2026 review period, we found that the withdrawal and disengagement experience varied dramatically. Some bots had emergency kill switches accessible via Telegram; others required logging into a remote server and killing Python processes manually. In a fast-moving market where you need to exit immediately, this technical barrier can be costly.

The bot's reliance on Bitget's API also means that if Bitget experiences an outage — which has happened multiple times during high-volatility events — the bot cannot exit positions until the exchange recovers. This is not a flaw in the bot itself, but a systemic risk that any Bitget-based strategy must account for.

What happens if the API connection drops mid-trade?

The systemd auto-restart handles process crashes, but it doesn't address API connectivity issues. If Bitget's API goes down or the VPS loses internet connectivity, the bot cannot close positions. The auto stop-loss and take-profit orders are placed on Bitget's side, so they should execute even if the bot disconnects — but trailing stops require the bot to periodically update the stop level, which means a disconnected bot cannot manage trailing stops.

This is a critical distinction. Static stop-losses survive disconnection; trailing stops do not. If the bot relies heavily on trailing stops for profit protection — and the specification mentions trailing stops as a feature — then API reliability becomes a significant risk factor.

How does the dual-strategy approach actually work?

The developer hasn't disclosed the specific strategies, which is common for proprietary systems but makes independent verification impossible. From the trade data, we can make some inferences:

The winning trades (FUTU, SOXS, MUU, NBIS, BEAT, BSB) include a mix of crypto-related stocks and ETFs. The losing trades (PRL twice, BEAT once, MUU once) suggest the bot may re-enter positions that previously lost, which is a strategy deviation flag. Re-entering a losing position without changing the underlying setup is a common behavioral error in automated systems — the bot doesn't "learn" from a loss unless explicitly programmed to do so.

One editorial insight specific to AI and algo trading that often goes undiscussed: strategy interaction risk in multi-strategy bots. When a bot runs two strategies simultaneously on the same account, the strategies can work against each other. One strategy might be entering a long position while the other is entering a short on the same or correlated pair. The developer limits concurrent trades to two, which helps, but doesn't prevent correlated position conflicts. We've seen multi-strategy bots quietly hedge themselves into breakeven over weeks, generating lots of trades but no net profit — a phenomenon that backtests rarely capture because they typically test strategies in isolation.

The 30-day test: what to watch for

The developer plans to run the bot for 30 days to build a proper track record. This is the right approach. Here's what we'll be watching:

1. Drawdown consistency: Does the 3% risk per trade hold across all market conditions?
2. Win rate stability: A 53.8% win rate over 26 trades per month means about 14 wins and 12 losses. Over 30 days, that's roughly 17-18 trades. If the win rate drops below 45%, the profit factor needs to be above 1.50 to stay profitable.
3. Maximum drawdown: Any single drawdown exceeding 15% on a 20x leveraged account is dangerous.
4. Funding rate impact: On Bitget perpetual futures, funding rates can be positive or negative. If the bot consistently holds positions through funding rate payments, it needs to account for this cost.
5. Slippage on small caps: The 43 small-cap pairs in the backtest may have wider spreads than the live execution suggests.

If the developer shares the full 30-day results with trade-by-trade data, we'll update this review. For now, the available data is insufficient to recommend the bot for anything beyond a very small test account.

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.

How Zephyr AI Compares

For traders evaluating whether to build their own bot or use an established platform, the comparison is instructive. Zephyr AI Trading Bot addresses several of the critical gaps we identified in this DIY approach:

• Strategy transparency: Zephyr AI publishes its strategy specifications and logic, unlike the black-box dual-strategy approach here
• Drawdown control: Zephyr AI's risk management includes dynamic position sizing based on volatility, rather than a fixed 3% per trade
• Regulatory clarity: Zephyr AI operates with clear compliance frameworks across jurisdictions
• Disengagement: Zephyr AI provides a web-based dashboard for stopping trades instantly, no SSH required
• Backtest verification: Zephyr AI's published backtests include slippage and fee modeling, reducing the backtest-to-live gap

The DIY bot developer deserves credit for persistence and transparency in sharing early results. But for retail traders who need reliability, regulatory compliance, and professional-grade risk management, an established platform with proven infrastructure is the safer choice.

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 trades crypto perpetual futures on Bitget, which are not subject to Pattern Day Trader rules (those apply to SEC-regulated securities). However, US traders should verify Bitget's terms of service regarding US-based users, as many crypto derivatives exchanges restrict access from the US.

Can I run it on a prop firm account?

Most prop firms prohibit the use of automated trading bots on their funded accounts unless explicitly approved. The bot's 20x leverage and 3% risk per trade would likely violate prop firm risk parameters, which typically limit risk to 0.5-1% per trade. Check your prop firm's rules before connecting any bot.

What happens if the API connection drops mid-trade?

Static stop-loss and take-profit orders placed on Bitget's servers will still execute if the bot disconnects. However, trailing stops require the bot to periodically update the stop level, so trailing functionality will be lost during an API outage. The systemd auto-restart will attempt to reconnect, but there may be a gap in trailing stop management.

How much capital do I need to start?

The developer used $266, but the bot's 3% risk per trade and 20x leverage mean the minimum viable capital depends on Bitget's minimum position size. With 20x leverage, a $266 account risks $7.98 per trade. A more conservative starting capital would be $500-$1,000 to allow for drawdown without violating minimum position requirements.

Is the bot regulated by the FCA or ASIC?

No. The developer is a UK-based individual with no FCA registration. Searches of the FCA register and ASIC's database returned no results for the bot or its developer. If the bot is packaged for commercial sale, UK financial promotion regulations would likely apply.

What exchanges does it work with?

Currently, the bot is built specifically for Bitget USDT perpetual futures using the ccxt library. In theory, ccxt supports 100+ exchanges, but the bot's logic is likely optimized for Bitget's API structure and fee schedule. Adapting it to other exchanges would require code modifications.

How do I stop the bot if something goes wrong?

You need SSH access to the DigitalOcean VPS to kill the Python process or disable the systemd service. There is no web-based dashboard or Telegram kill switch mentioned in the developer's specification. For non-technical users, this could be a significant barrier during an emergency.

What are the ongoing costs besides the bot itself?

You'll need a DigitalOcean VPS ($6-$12/month), Bitget trading fees (0.02%-0.06% per trade), and funding rate costs on perpetual futures positions (variable, typically 0.01% per 8 hours). On a $266 account, monthly operating costs could be $15-$30, representing a significant percentage of potential profits.

How does the dual-strategy approach avoid conflicting signals?

The developer limits concurrent trades to two, which reduces the chance of strategies working against each other. However, without knowing the specific strategies, it's impossible to verify whether they could enter correlated positions that effectively hedge each other out. This is a common issue in multi-strategy bots that backtests often miss.

What happens if Bitget goes down or restricts access?

The bot cannot trade or exit positions during a Bitget outage. Static stop-loss orders placed on Bitget's servers should still execute, but no new trades can be opened. This is a platform-level risk that any Bitget-based strategy must accept.

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 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.

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

Read our full Testing Methodology.

Related Reviews:

• See also: More Crypto reviews on cryptoplatformreviews.io.
• For dedicated crypto coverage, visit cryptoplatformreviews.io.