Home › Learn › Guides

How to Forward-Test a Strategy Before Risking Real Money

Updated 28 July 2026 · 8 min read · PipTax education

If you want to forward-test a strategy properly, you need more than a demo account and good intentions — you need a fixed process, a defined sample size, and honest record-keeping before a single pound of real money is at risk. Forward testing is the bridge between a backtest that looks good on paper and a live account that has to survive real spreads, real slippage, and real emotion. Skip it, or do it sloppily, and you're gambling with extra steps.

What Forward Testing Actually Means

Forward testing is running your strategy in real time — on demo or with minimal live size — going forward from today, rather than on historical data. It answers questions a backtest can't:

• Does the strategy work when you don't already know the outcome?
• Can you actually follow your own rules under time pressure?
• Do execution costs (spread, commission, slippage) eat the edge?
• Does the strategy hold up across different market conditions as they unfold?

Backtesting and forward testing are not substitutes for each other. A backtest is fast and cheap but prone to hindsight bias and curve-fitting. Forward testing is slower but far more honest. The two work best together: backtest to find and refine an idea, forward-test to validate it before committing capital. If you haven't backtested at all, forward testing is where the first real cracks usually show up — get familiar with the basics in the /school/index.html section before you start.

Setting Up a Proper Forward Test

A forward test only means something if it's structured. Loose "let's see how it goes" testing wastes weeks and teaches you nothing reliable. Before you place a single trade:

1. Write the rules down — entry, exit, stop-loss, take-profit, position sizing, and the exact conditions that invalidate a setup. If it's not written, it's not a rule. 2. Pick your market and timeframe and stick to them for the full test period. 3. Choose a sample size in advance — a minimum of 30-50 trades, or a fixed period (commonly 4-8 weeks), whichever comes first. Fewer trades than that and you're reacting to noise, not results. 4. Use a demo account that mirrors your intended live broker's conditions as closely as possible — same instrument, same account type, same typical spread environment. 5. Decide your pass/fail criteria before you start, not after you see the equity curve.

This last point matters more than people think. It's very easy to retroactively decide a losing streak was "just bad luck" once you're emotionally invested in the strategy working.

Demo vs Small Live: Choosing Your Test Environment

Demo accounts are free and low-stress, but they miss two things: realistic fills during volatile news, and your own psychology when money is genuinely on the line. Many traders run a two-stage forward test:

| Stage | Purpose | Typical length | |---|---|---| | Demo | Confirm the rules work mechanically, no execution surprises | 4-6 weeks | | Small live | Confirm you can follow the rules with real money at stake | 4-6 weeks |

For the small-live stage, size positions so a full losing streak (say, 10 trades) wouldn't bother you emotionally or financially — often 0.1-0.25% risk per trade rather than your intended full size. The goal isn't profit yet; it's proving the process holds under real conditions.

Whichever broker you use for this stage, check how its demo environment compares to live execution. Pepperstone and IG, for example, both publish details on their platforms and order execution — but demo spread behaviour can differ from live, especially around news. Always confirm current live spreads and commissions on /brokers/index.html rather than assuming demo pricing carries over exactly.

Metrics to Track During the Test

A spreadsheet or trading journal is non-negotiable. At minimum, log per trade:

• Date, instrument, direction, entry, stop, target
• Result in pips and in account currency
• Reason for entry (which rule triggered it)
• Whether you followed your own plan exactly — yes/no, no exceptions

From that raw data, calculate at the end of each week:

• Win rate — percentage of winning trades
• Average win vs average loss (risk-reward realised, not planned)
• Maximum drawdown — largest peak-to-trough equity dip
• Expectancy — (win rate × average win) minus (loss rate × average loss)
• Rule adherence rate — how often you actually stuck to the plan

Expectancy and adherence matter more than raw profit over a short sample. A strategy with slightly negative expectancy but 100% rule adherence tells you something useful and fixable. A strategy with positive results but 60% adherence tells you the numbers aren't really testing the strategy at all — they're testing your improvisation.

Accounting for Real Trading Costs

This is the step most home-tested strategies fail on. A strategy that looks profitable on a backtest with zero costs, or on a demo with unusually tight spreads, can turn negative once real spreads, commissions, and swaps are applied — particularly for high-frequency or scalping approaches where costs are a bigger share of each trade's expected value.

Before and during your forward test:

• Model your expected cost per trade using realistic spread and commission figures for your instrument and account type.
• Re-run your expectancy calculation with those costs included, not excluded.
• Check overnight swap costs if you hold trades past rollover — these compound quickly on longer-held positions.
• Compare account types — a raw-spread-plus-commission account can cost less overall than a wider "commission-free" spread account, depending on your trade frequency.

Don't guess these numbers or rely on old figures — they change. Run your actual instrument and expected trade volume through /audit.html to see total cost impact, and use /cost-impact.html to see how spread and commission differences translate into pounds over a realistic number of trades. This single step catches more broken strategies than any amount of extra backtesting.

Reviewing Results and Deciding Next Steps

Once you hit your predetermined sample size or time period, review honestly against the criteria you set at the start:

• Pass: expectancy positive after realistic costs, adherence high, drawdown within your tolerance → move to the next stage (larger live size, or a longer test).
• Marginal: mixed results, low sample size, or adherence issues → extend the test, don't just call it early because you're impatient.
• Fail: negative expectancy after costs, or you couldn't follow the rules → rework the strategy or abandon it. This is a good outcome, not a wasted one — you found out on demo money, not real money.

Keep the full record. Even a failed forward test is useful data for the next idea, and a pattern of "always fails at the costs step" usually points to a strategy that trades too frequently for its edge size.

Bringing It All Together

To properly forward-test a strategy, you need written rules, a fixed sample size decided in advance, honest journaling, and — critically — realistic cost modelling before you scale into live trading. Skipping any one of these steps is how traders end up "confirming" strategies that were never actually profitable once real-world costs were applied. Treat forward testing as a real experiment with a real pass/fail bar, not a formality on the way to going live, and use tools like /audit.html and /methodology.html to keep your cost assumptions grounded in reality rather than hope.

Trading involves substantial risk of loss, and no amount of testing removes that risk — forward testing simply reduces the chance you're carrying it unknowingly.

Keep going: Audit Cost Impact Index Index