Validation.Protocol.v2.0

Backtest vs
Forward Test

Historical results and live results form one validation stack. Robust workflows compare both layers and explain every material deviation.

Updated: 17 May 2026

Key takeaways

A backtest carries every flaw in its data: missing ticks, invented spreads, fills that never happen.
A report worth reading names four things first: data source, spread, commission, assumed slippage.
A backtest proves the logic holds. Only a forward test shows what the market costs you.
Freeze the rules and run through calm and wild sessions, thick markets and thin.
Track four gaps live: requested against filled price, fill time, real cost against assumed, manual overrides.

Backtest vs forward test framework: data quality checks, model assumptions, and live execution verification steps.

// Validation_01_Data

How Good Must
Backtest Data Be

The reliability layer behind every simulation output.

A backtest carries every flaw in the data you fed it. Missing ticks, made-up spreads and fills that would never happen all push the result up. Verification against reliable platforms like TradingView ensures your historical charts match real-world market prices.

A report worth reading names four things before it shows you a single number. Where the data came from, what spread it used, what commission, and how much slippage it assumed.

Check the data before you tune anything. Clean timestamps, honest costs and markets split by type are what stop a smooth curve from fooling you.

Simulation Data Checklist

Clean timestamps, with one time zone throughout.

Spread and commission set to the broker you will actually use.

Slippage and delay written down, with a version you can point at.

The data used to tune kept apart from the data used to test.

A written rule for what happens around news.

A note showing you checked for survivorship and look-ahead bias.

Integrity Questions

Would it still work if the spread doubled at the moment you enter?

Does it still hold once you throw out the best 5 percent of trades?

Does it hold on data it never saw, in more than one kind of market?

A good backtest tells you the logic holds. It cannot tell you how the market will fill your orders.

// Validation_02_Live

Live Execution
Reality

Forward tests verify whether backtest assumptions survive real conditions.

A forward test shows you what the market costs you in real time. The point is to check the model, not to make money this month.

Freeze the rules and watch it trade through different sessions, calm days and wild ones, thick markets and thin. Holding steady across all of them is what counts.

Forward-Test Metrics

Fill Quality

Compare the price you wanted with the price you got, session by session and by order type.

Time to Fill

Time the typical fill and the slowest ones, per pair, per session, in calm and wild markets.

Cost Drift

Put the real spread and commission next to what you assumed, and track the gap each week.

Rule Adherence

Keep what the model did apart from what you overrode by hand, and from the steps you simply missed.

Forward-Test Protocol

Do not touch the rules until the test is over.

Log every trade with the same clock.

Review the gaps weekly. Act only on the ones that keep coming back.

Run it long enough to cover several sessions, calm markets and wild ones. Only then change your mind about the model.

Frequently Asked Questions

What is the difference between backtesting and forward testing?

Backtesting evaluates rules on historical data, while forward testing evaluates the same rules in live market conditions. Each stage answers a different validation question.

Why can a strong backtest fail in live trading?

Live trading adds spread changes, slippage, latency, partial fills, and behavioral overrides. These factors are usually simplified or absent in historical simulations.

How long should forward testing run?

Forward testing should span multiple volatility regimes and sessions so results include different market states. Fixed sample size rules improve statistical confidence.