Homeใ€€>ใ€€Knowledgeใ€€>ใ€€What is a Forward Test

Forward TestMT5Verification & Evaluation

What is a Forward Test โ€” Differences from Backtest and How to Read Results

Last updated: 2026-05-18 | Est. reading time: 10 min

To evaluate EA reliability, real-time verification in live markets (forward test) is essential alongside historical data verification (backtest). Only by combining both can you assess whether an EA has a strategy that works both in the past and going forward.

Differences Between Backtest and Forward Test

ItemBacktest (BT)Forward Test (FT)
Data usedHistorical price dataReal-time market prices
SpreadFixed or model-estimatedActual spread (variable)
SlippageMostly not consideredActual execution slippage
Execution speedInstant (10 years in minutes)Real-time (actual time required)
ReliabilityRisk of curve fittingReflects actual market conditions
Minimum required periodA few hours to 1 day3 months to 1+ year
CostFree (possible in demo environment)Free (demo account) or real funds required

Why Forward Testing Matters

1. Backtests are only optimized for the "past"

No matter how good the backtest results are, they only evaluate performance on past data. EAs that are overfitted (curve-fitted) may look excellent in backtests but fail to perform in live markets. Forward testing exposes this.

2. Measures real spreads and slippage

Broker spreads fluctuate with market conditions โ€” they can become 5 to 10 times higher than normal during major economic announcements. Because backtests use fixed spreads, actual friction costs are invisible. Forward tests let you evaluate real execution quality.

3. Confirms broker compatibility

The same EA can produce different results depending on execution speed and slippage across brokers. For scalping EAs in particular, broker compatibility is a decisive factor in performance.

How to Read Forward Test Results

Our "Real-Time Results" page publishes the forward test results for each EA. The following explains how to read the key metrics.

Critical

Profit Factor (PF)

Total profit รท total loss. Above 1.0 means positive expected value. 1.3 or above is the benchmark for stability.

Important

Maximum Drawdown (Max DD)

Maximum drop from peak to trough (%). If FT max DD is within 1.5ร— the BT max DD, it is within the normal range.

Reference

Win Rate

Alone, it means little. Always view alongside the risk/reward ratio. Even a 40% win rate is positive expected value at RR 1:2.

Important

Number of Trades

At least 50โ€“100 trades are needed for statistical reliability in a forward test. If the period is short, treat results as reference only.

Critical

BT Comparison (Deviation Rate)

If FT PF is within 70โ€“130% of BT PF, you can determine that the EA is "performing as expected."

Pass/Fail Criteria for Forward Tests

ResultConditionAction
โœ… GoodFT PF โ‰ฅ 1.2, Max DD < 1.5ร— BT, 50+ tradesContinue running with live funds
โš ๏ธ MonitorFT PF 1.0โ€“1.2 or Max DD = 1.0โ€“1.5ร— BTExtend by 3 months and re-evaluate
โŒ ReviewFT PF < 1.0 or Max DD > 1.5ร— BTReturn to demo and re-examine parameters
๐Ÿšจ Stop RecommendedMax DD > 2ร— BT or consecutive errors in FTStop operation and investigate cause

How to Run a Forward Test

Step 1

Run the EA on a demo account

Set up the EA on an HFM, XMTrading, or Exness demo account. Use the same parameter settings as production (RiskPercent, SL, MagicNumber).

Step 2

Wait for at least 3 months and 50+ trades

EAs with low trade frequency (H4, D1 timeframes) can take 3โ€“6 months. Avoid rushing to conclusions โ€” patience is paramount.

Step 3

Record a monthly report

Export an HTML report from MT5's account history tab. Record PF, Max DD, and trade count on a monthly basis.

Step 4

Compare and evaluate against the backtest

Check whether FT performance is within 70โ€“130% of BT. If there is a large deviation, review the parameters or broker.

Step 5

Decide on going live

If it passes the criteria, move to a live account. Start with the minimum lot size, then gradually increase as performance holds.

๐Ÿ“Š Check Our Site's Forward Test Results

We publish real-time forward test results for all distributed EAs using demo accounts. We also include deviations from backtest results so you can assess EA stability.

View Real-Time Results โ†’

FAQ

Q: Should I run a forward test on a demo account or a live account?

We strongly recommend starting with a demo account for 3 to 6 months. While there are slight differences in execution speed between demo and live, a demo account is sufficient to verify strategy effectiveness. The safe procedure is to pass the demo stage and then move to live with the minimum lot size.

Q: My backtest showed PF 1.5, but forward test PF dropped below 1.0. Why?

The most common cause is overfitting (curve fitting). The BT parameters were excessively optimized to historical data and cannot replicate performance in live markets. The next most common cause is that the spread settings in the BT were lower than the actual average. Re-run the BT using the real average spread values.

Q: How many trades in a forward test are needed for statistical reliability?

A minimum of 50 trades; ideally 100 or more. EAs trading on H4 or D1 timeframes have low trade frequency, and it may take 6 months to 1 year to achieve statistical reliability. Treat results with few trades as reference only.

Q: Which broker's data is used for the real-time results on this site?

We verify using an XMTrading demo account. If you use Exness, slight discrepancies may arise due to differences in spreads and swap rates.

Q: Can I change parameters during the forward test?

If you change parameters, record it as a new FT starting from the date of the change. Changing parameters mid-test makes the entire verification period's data meaningless. The correct procedure is: re-run the BT โ†’ start a new FT.