To test forex predict candles MT5 tester forecasts honestly, lock each forecast on the first tick of the target candle, save the predicted OHLC and probabilities, then score it only after that candle closes. Use real ticks when spread and intrabar order affect execution, and keep the UTC anchor in every log row.
The buyer question is procedural: can a new candle-prediction indicator produce a timestamp-safe record that another trader can reproduce? The Forex Next Candle Forecast 3D MT5 listing provides the useful pattern—first-tick locking, CSV output, and post-close evaluation.
What must a forex predict candles MT5 tester measure?
A valid forex predict candles MT5 tester measures one forecast event against one completed candle. You need a fixed UTC anchor, immutable forecast values, and an actual OHLC record captured after the close. That separates model error from execution results.
The Forex Next Candle Forecast 3D MT5 listing describes a forecast that locks on the first tick, writes the forecast to CSV, and evaluates only after the candle closes. Treat those as separate events. If later high, low, or close values influence the forecast, the measurement is contaminated.
See also: Forecast accuracy backtest forex
Which MT5 mode suits a real ticks forex forecast?
For a real ticks forex forecast, choose Every tick based on real ticks when spread and intrabar order matter. Use faster modes for screening or diagnostics, not as the primary execution-sensitive result.
| Tester mode | What it provides | Use when |
|---|---|---|
| Open prices only | Bar-open events with minimal intrabar detail | You need a rough timing or data sanity check |
| 1 Minute OHLC | Open, high, low, and close values for each minute | You need faster screening and intrabar order is not material |
| Every tick | Generated ticks based on minute history | Real broker ticks are unavailable or you need a sensitivity run |
| Every tick based on real ticks | Broker-provided ticks with changing intraminute conditions | Spread and the order of intrabar prices affect the result |
The MetaTrader 5 Built-in Trading Strategy Tester distinguishes Open prices only, 1 Minute OHLC, and Every tick as different speed and quality choices. Real and Generated Ticks — MetaTrader 5 Help explains why real ticks are preferred for execution-sensitive work: spread can change within one minute, and trades execute at Bid or Ask rather than at an abstract candle value.
How do you run a no look-ahead candle test in MT5?
A no look-ahead candle test locks the forecast before the target candle develops and scores it only after the close. Use the following sequence for an M15 test with a 15-minute forecast horizon.
- Set the sample: choose a symbol such as EURUSD, timeframe M15, a fixed test period, and UTC as the reporting timezone. Failure mode — changing the timezone or candle boundary between lock and evaluation joins the forecast to the wrong candle.
- Select Every tick based on real ticks in the MT5 Strategy Tester. Failure mode — using Open prices only, 1 Minute OHLC, or generated Every tick data can hide spread changes and alter intrabar order.
- At the first tick of each target M15 candle, lock the 15-minute forecast. Capture forecast open, high, low, close, direction probability, and confidence. Failure mode — sampling after the candle has traded gives the model information it could not have had.
- Write the locked values to CSV immediately. Include the symbol, timeframe, and UTC anchor in the same row. Failure mode — relying on a chart label or manual copy can create duplicate anchors and missing forecasts.
- Wait until the target candle closes before recording actual open, high, low, and close. Failure mode — filling actual OHLC from an unfinished candle creates a partial result.
- Record spread in pips and the UTC evaluation timestamp after the close. Failure mode — storing server time without conversion makes results difficult to reconcile across brokers.
- Score direction and OHLC error separately, then audit missing rows, duplicate anchors, and incomplete candles. Failure mode — dropping failed or unavailable rows makes the final sample look cleaner than the test actually was.
What should your MT5 candle forecast backtest log?
An MT5 candle forecast backtest is auditable only when every forecast maps to one target candle. Use one CSV row per forecast and require the fields below.
| Required field | Record | When captured |
|---|---|---|
| Symbol | Exact MT5 symbol, such as EURUSD | At forecast lock |
| Timeframe | M15 | At forecast lock |
| UTC anchor | First-tick timestamp of the target candle | At forecast lock |
| Forecast OHLC | Predicted open, high, low, and close | At forecast lock |
| Direction probability | For example, 62% probability of an up close | At forecast lock |
| Confidence | For example, 71% model confidence | At forecast lock |
| Actual OHLC | Observed open, high, low, and close | After candle close |
| Spread in pips | Observed spread associated with the test event | At lock and close when execution matters |
| Evaluation timestamp | UTC time when the completed candle was scored | After candle close |
Keep direction probability distinct from confidence. A 62% direction probability estimates the likelihood of the directional class. A 71% confidence value describes the model’s certainty score. Neither number is a profit probability or a position-sizing instruction.
If your objective is OHLC quality rather than a single close direction, use the Forex Prediction Range Forecast: Point vs OHLC article to define whether your primary score is a point, range, or complete OHLC forecast.
See also: News impact on forex forecast
How should you read PRISM’s 48-hour forecast context?
For an M15 forecast, PRISM’s 48-hour context contains 192 M15 candles, 48 H1 candles, 12 H4 candles, and 2 daily candles. Keep those context windows fixed when comparing forecasts so each lock uses the same information structure.
PRISM publishes 48-hour OHLC candle forecasts across 11 markets and 4 timeframes, an hourly AI briefing that reads market news against each forecast, an MT5 indicator, and a REST API. The published M15 range MAE is 0.079%, with 54% error reduction versus its no-skill baseline. Those are model-error measurements, not trading profit, hit rate, drawdown, or return.
FAQ: forex predict candles MT5 tester
Which MT5 mode is the honest default?
Every tick based on real ticks is the default when spread and intrabar order can change the result. Generated Every tick is a fallback for sensitivity testing.
When should a forecast be scored?
Score it after the target candle closes. Before then, actual high, low, and close values are incomplete.
Does a 0.079% MAE mean profitable trading?
No. The 0.079% figure is PRISM’s published M15 range model error. It does not include spread, execution, sizing, drawdown, or trading profit.
Build the test before you trust the number
A first-tick lock, real-tick data, UTC timestamps, and post-close OHLC scoring give your test a defensible measurement boundary. Forecasts are model output, not financial advice. Review the MT5 candle forecast indicator when you are ready to implement this workflow in your own tester →
