Your Bot Works For 3 Hours, Then Fails Silently
You built an EA. Backtests look great—47% annual return, 2.3 Sharpe ratio. You deploy it live on Monday morning. By 11 AM, you notice orders aren't executing on time. By 2 PM, you get request timeouts. By 3 PM, your account has 15 open positions instead of 3. You panic. You blame the code.
You rebuild the EA. Optimize the order logic. Cache more data locally. Deploy again. Same problem. This time you blame your data source. You switch to a different price feed. Deploy again. Still failing.
Here's what you don't know: the broker rate-limited you 90 minutes into the live session. Your EA is fine. Your code is fine. The broker just doesn't want retail traders running this volume.
What API Throttling Actually Is
API throttling is when a broker intentionally slows down or rejects your requests after you exceed a limit. Not a hard crash. Not a timeout error message. Just... silent failure. Request 101 gets rejected. Request 102 queues. Your EA doesn't know it's being throttled—it thinks the broker is just slow.
So it waits. Your trade window closes. Meanwhile, the next signal is already gone.
Most retail traders never find out they're being throttled. They assume their EA is broken. They rebuild it four times. They optimize it to death. Then they give up and pay for someone else to build it.
Retail vs Pro: The Infrastructure Difference Nobody Talks About
Here's the real split:
- Retail accounts: 100–200 requests per minute, shared connection pool with thousands of other traders
- Pro accounts: 1,000–5,000 requests per minute, dedicated server infrastructure
The difference isn't the API itself. It's the infrastructure behind it. Retail accounts share a single connection pool with everyone else on the broker's platform. Pro accounts get isolated lanes.
When the shared pool fills up (which happens constantly during high-volatility windows), retail requests queue. When the queue backs up, requests drop. Silent drops. No error message. Just gone.
Investopedia's guide to APIs explains the technical mechanics, but brokers never publish their actual rate limit architecture. You have to discover it through trial and failure—which means losing money on live trades while you figure it out.
Why 100 Requests Per Minute Isn't Enough For Any Serious Strategy
Let's do the math. A baseline EA needs:
- 1 request to fetch current price
- 2 requests to place an order and confirm it
- 1 request to check position status
- 1 request to update stop loss
- 1 request to adjust take profit
That's 6 requests per trade. If you're executing 20 trades per minute (modest for a scalping bot), you need 120 requests per minute. You're already over the retail limit.
Add risk management checks, sentiment filters, or multi-timeframe analysis? You're at 200+ requests per minute. The broker is throttling you before you even know you started.
Professional traders know this going in. They don't build EAs that work inside the limit. They build EAs that require a higher limit, then they upgrade their account tier to accommodate it. Completely different approach.
The Compounding Cost Of Throttling (In Dollars)
This is where retail traders go broke without realizing why. They don't understand the throttling. So they blame the EA and rebuild it. Each rebuild adds complexity. More indicators. More data checks. More API calls.
Each additional API call increases throttling. The EA slows down. Trades miss their window. You add more optimization. Now you're spending 40 hours rebuilding something that can't work at your broker's limits.
Opportunity cost: $0 in code costs, but $800–$2,000 in missed trades. Psychological cost: you lose confidence in automation entirely.
Meanwhile, a pro account trader with the exact same EA is executing 200 trades per day at scale. Same broker. Different account tier. Different outcome.
How Professional Traders Escape The Death Spiral
Pros do one of three things:
- Upgrade to a pro account: $5,000–$15,000 per month. Instant 5–10x rate limit increase. Your EA runs unthrottled.
- Use multiple sub-accounts: Spread volume across 5–10 retail accounts. Each gets 100 req/min. Together you get 500–1,000 req/min. Harder to manage, but it works temporarily.
- Host on the broker's VPS: Request processing is faster. Latency drops. The broker gives you higher limits because you're "on their hardware." Not cheating—just using the system the way it's designed.
Retail traders don't know about option 3. They either pay for a pro account upgrade or accept the ceiling and build inside it.
Can You Code Your Way Around This?
This is where most retail developers waste time. They think better code beats broker limits.
You can optimize. Batch requests. Cache data locally. Use WebSocket instead of REST (lower overhead). Use MetaTrader 5's API documentation to understand latency reduction. But you hit the ceiling eventually.
If the broker limits you to 100 req/min and your strategy needs 180, optimization buys maybe 15% more throughput. Then you're stuck again. You can't code your way past a rate limit.
The objection traders raise: "But I'll just make my EA smarter—fewer API calls per trade."
Here's the thing: if fewer API calls made better returns, professional traders would do that. They don't. They do the opposite. They add more frequency, more redundancy checks, more data. Volume compounds returns. Fewer API calls means fewer signals. Fewer signals means fewer opportunities. You're not solving throttling with better code. You're solving it by upgrading your infrastructure.
What Actually Works When Your Broker Throttles You
Option A: Accept the limit. Build your EA to work inside 100 req/min. Slower bots can still be profitable—just not at scale. This is viable if you're running 5–10 trades per day, not 100.
Option B: Switch brokers. Some brokers publish higher rate limits for retail accounts, though most don't. Depends on your platform—MT4 vs MT5 vs TradingView vs cTrader have different limit structures.
Option C: Upgrade to a professional account. This is what serious traders do. $5,000–$15,000 per month unlocks 5,000 req/min. If you're losing $1,000+ per month in missed trades on a retail account, it pays for itself in 5 months.
Option D: Build for a different platform. Crypto exchanges like Binance allow 1,200–2,400 requests per minute for authenticated users. No "retail vs pro" infrastructure divide. Higher liquidity. Same EA logic, different venue.
Why This Matters For Custom EA Development
This is the hidden reason why off-the-shelf EAs usually fail on retail accounts. They're built for pro infrastructure, then sold to retail traders. The backtests are real. The logic is sound. But the broker's rate limits kill it in production.
When you hire a developer to build a custom EA, what you're really paying for is knowledge of your specific broker's limits. Knowledge about which request patterns work. Knowledge about how to batch operations or cache data to stay under the limit while still getting enough signals.
That's not something you learn from a tutorial. Alorny builds custom EAs starting from $100 for simple strategies. What separates a custom build from an off-the-shelf EA isn't the code quality—it's that we know the rate limits for every major broker and build inside those constraints from day one. Your EA is tested against your actual broker's throttling behavior before you deploy it live.
Key Takeaways
- Retail accounts are rate-limited to 100–200 requests per minute by design. Your broker isn't bug—it's intentional.
- If your EA fails after a few hours of live trading, it's hitting the throttle wall, not a code problem.
- Professional traders upgrade account tiers or use infrastructure workarounds. Retail traders usually don't know these options exist.
- You cannot optimize your way past a broker's rate limit. You can reduce overhead, but you'll still hit the ceiling.
- Upgrading to a pro account ($5,000+/month) or hiring a developer who knows the limits is how you escape the death spiral.