Your Backtest Is Lying To You
Your backtest shows a 90% win rate. You're excited. You go live. Within two weeks, you've blown your account. Here's what happened: your backtest wasn't running on real market data—it was running on a simulation so stripped-down it barely resembles what your broker actually traded.
Tick data accuracy determines the difference between a fantasy win rate and real profitability. Most retail traders don't know this. Most retail backtesting platforms don't want you to know this. Because once you understand tick data, you realize the entire backtest is fiction.
Why Retail Backtesting Platforms Use Garbage Data
Here's the thing: real market data is expensive. A single day of tick-by-tick data for a major forex pair can be hundreds of megabytes. Storing years of it costs money. Updating it in real-time costs money. So retail platforms cut corners.
- Sparse tick density. Platforms record only some of the ticks that actually executed. A gap of 10+ seconds between ticks is common. Real brokers see thousands of ticks per second on volatile pairs. Your backtest? Maybe one tick every few seconds.
- Compressed candlestick data. MetaTrader converts ticks into bars. Those bars are reconstructions, not recordings. A 1-minute bar doesn't show the 50 price movements that happened within that minute.
- Missing spread data. Most retail backtests assume a fixed spread. Real spreads widen during news, high volatility, low liquidity. Your backtest ignores this entirely.
- No slippage modeling. Platforms assume your order fills at the backtest price. Real brokers slippage you during volatile moves, especially on large orders. You're 2-5 pips worse on entry, 2-5 pips worse on exit.
- Broker-specific divergence. Each broker has a different feed. Different liquidity providers, different order routing. The data you backtest on doesn't match the data your broker actually trades.
The result? A backtest that bears almost no resemblance to live market behavior.
How This Kills Your Profits
Let's say your backtest shows your strategy wins 90 times out of 100 trades. In reality:
- Slippage cuts 2-4% off every win. Your +10 pip win becomes +6 pips. Your +20 pip win becomes +16 pips. Over 100 trades, you lose 3-4% of total gains.
- Spread widening kills breakeven trades. Backtests assume tight spreads. Live spreads double or triple on volatile moves. Trades that barely won in backtest now lose.
- Gaps between ticks hide real entries. Your backtest shows the price hitting your entry. But between the ticks, the price moved 5 pips past it. Your order never fills at that price. You miss the trade or chase at worse prices.
- The 90% win rate becomes 30-40% live. Not because your logic is wrong. Because your data was fiction.
This is not theoretical. Retail traders blow accounts every single day because they trusted their backtest. According to a CFTC study, 87% of retail forex traders lose money. The tick data problem is one of the reasons why.
The Divergence Is Massive
Here's a concrete example. Take EURUSD on a volatile news day (Fed announcement, CPI release). During the spike:
- Real broker feed: 5,000+ ticks per minute. Every bid/ask update recorded. Spreads widen to 10-20 pips. Slippage on large orders is 5-15 pips.
- Retail backtest feed: Maybe 100-200 ticks per minute. Spread assumed at 1.5 pips throughout. Slippage set to zero or a flat 1 pip.
Your backtest EA holds the position through the spike. The data feeds it one tick per second. So it thinks the volatility is manageable. In reality? The spread exploded to 15 pips. Your stop loss got hit 8 pips earlier than you modeled. The order to exit took 2 seconds to fill because the broker was slammed. You lost 40 pips. Your backtest predicted a +10 pip win.
This isn't an edge divergence. This is a data divergence. And data divergence is pure luck masquerading as skill.
What Professional Traders Do Instead
Professional traders don't use retail backtest data. Here's why:
- They test on institutional-grade data — high-density tick data from actual broker feeds, not compressed simulations. This costs money, but it's the cost of knowing if your strategy actually works.
- They model realistic slippage and spreads. They measure their own broker's actual spreads and slippage patterns, then apply those to the backtest. Not guesses. Data.
- They verify backtest on a demo account first. They run live (but not real-money) backtests on their broker's actual feed to see if the numbers match. If backtest shows 70% win rate and demo shows 40%, they know the data was the issue.
- They optimize for profit, not win rate. They don't chase 90% win rates. They chase consistent profit. A 30% win rate with 1:3 risk-reward beats a 90% win rate with 1:0.5 risk-reward. Tick data doesn't lie about this—it reveals the truth.
- They hire custom EA developers who know the data game. Platforms like Alorny test EAs on verified broker data, run full backtests with realistic slippage, and deliver a backtest report before the EA goes live. You see exactly where the data came from and what assumptions were made.
The traders who do this stay profitable. The traders who rely on retail backtest data blow accounts.
The Real Cost Of Bad Data
Let's calculate what bad tick data costs you.
Say you decide to live trade a retail backtest that shows 70% win rate with +2 pip average. You fund a $10K account. You risk 1% per trade ($100). Over 100 trades:
- Backtest projection: 70 wins × +2 pips = 140 pips profit = $1,400. Account grows to $11,400.
- Live reality (with slippage and spread widening): 40% win rate, +1 pip average, due to real market data. 40 wins × +1 pip = 40 pips profit = $400. Account grows to $10,400.
- Cost of bad data over those 100 trades: $1,000. That's a 71% profit divergence from what you expected.
Over a year of 1,200 trades? You're looking at a $12,000 loss instead of a $16,800 gain. That's a $28,800 swing. All because you trusted a backtest run on low-fidelity data.
And if your strategy is marginally profitable (which most are), that $1,000 divergence is enough to flip it negative and blow your account entirely.
How To Know If Your Data Is Trash
Before you backtest on any platform, ask these questions:
- Where does the tick data come from? Is it from your broker directly, or from a data vendor? How recent is it?
- What's the tick density? How many ticks per second on average during your testing period? If the answer is vague, the data is probably sparse.
- Are spreads and slippage modeled? Or assumed fixed? Real markets have variable spreads. If your platform assumes fixed spreads, your backtest is fiction.
- Can you export the backtest data and verify it? If the platform won't let you inspect the actual ticks being used, that's a red flag.
- Have you verified the backtest on a live demo account? Did the actual results match the backtest? If not, the data was the culprit.
Most retail platforms will either not answer these questions clearly or will admit the data is low-fidelity. That's when you know: time to get serious about tick data accuracy, or stop backtesting altogether and test on a demo account instead.
The Alternative: Custom EA Development With Verified Data
This is why traders hire custom EA developers for serious work. A professional EA developer:
- Tests on your broker's actual feed or institutional-grade tick data
- Includes realistic slippage and spread modeling based on your broker's actual behavior
- Delivers a full backtest report showing the data source, assumptions, and performance
- Runs the EA on a demo account to verify backtest results match live behavior
- Only marks the EA as production-ready when real and backtest align
It costs more than a retail backtest platform. But you pay once and get truth. With retail backtests, you pay monthly and get fiction.
Key Takeaways
- Retail backtest data is sparse and compressed. It's a simulation of a simulation. Your 90% win rate is probably fiction.
- Tick density, spread modeling, and slippage are the killers. Low-fidelity data ignores all three. Live markets don't.
- The divergence between backtest and live can be 50-70%. Your $10K backtest projection becomes a $5K blowup.
- Professional traders test on institutional-grade data. They verify on demo accounts. They hire developers who know the data game.
- The cost of bad tick data is not a data problem—it's a blowup problem. You can't afford to ignore this.
Next step: Stop trusting retail backtests. Either upgrade to institutional data feeds, test exclusively on live demo accounts, or hire a custom EA developer who uses verified data. One of these three will save you thousands in blowups.