ProOrder live entries differ from backtest entries…..

LivJoJade · 2026-10-17T18:03:17+00:00

Hi everyone, First of all for complete transparency I inform the forum that the script has been written with the help and collaboration of AI. I’m looking for some technical help understanding why my ProOrder autotrading entries do not always correspond with the entries produced by the backtest, despite using the same strategy code and parameter settings. Strategy setup Platform: ProRealTime v12 Instrument: US Tech 100 (DFB) Execution timeframe: 10 seconds Directional filters: Internally calculated, confirmed 5-minute and 1-minute swing directions must agree. Entry logic: A 10-second entry score, EMA-based calculations, spread and slope filters, and a minimum entry-score threshold. Parameters: Min1MinEntryScore = 5, MinSpread = 0, MinSlope = 1. Risk management: 12-point stop loss and 10-point profit target. Trading session tested: 17:00–19:30 on 16 October 2026. The strategy calculates its own swing directions internally; it does not rely on the chart’s external swing indicator to determine directional alignment. What I am observing Some live entries correspond very closely with the backtest, both in timing and price. However, other entries differ substantially in both respects. For example: BacktestLive Entry time 17:40:00 Entry price 29,442.6 ProOrderShort Entry time 17:41:30 Entry price 29,422.8 That is a difference of 90 seconds and 19.8 points. Another example: BacktestLive entry time18:12:20 Entry price 29,308.3 ProOrderShort entry time18:13:10 Entry price 29,293.6 That is a difference of 50 seconds and 14.7 points. Other entries match closely, with price differences of less than a point. This is why I am trying to understand what causes the inconsistent behaviour rather than assuming every live entry is different. What I have checked The same strategy code and parameter settings are being used. The backtest and live trading cover the same session. ProOrder is running the strategy on the 10-second timeframe. The 5-minute and 1-minute directional charts show established downward movement around the examples above. The ProOrder order list confirms that the live orders were executed, but the order-status information I have accessed does not identify when the strategy’s entry conditions first became true. My questions Can ProOrder evaluate a multi-timeframe strategy differently from the backtest, even when the same code, settings and execution timeframe are used? Could the interaction between TIMEFRAME(..., UPDATEONCLOSE), the internally calculated swing states and the 10-second execution timeframe cause the live strategy to recognise an entry later than the backtest? Are there differences in historical initialisation, bar synchronisation or strategy evaluation that could explain why some entries match closely while others occur one or more bars later? I can provide the complete strategy code if required… My objective is to identify the technical reason for the discrepancy so that live ProOrder can reproduce the intended backtest behaviour as faithfully as possible. I would prefer to establish the cause before changing the strategy logic or parameters. Thanks in advance for any guidance.

Viewing 31 post (of 31 total)

← 1 2 3

Author

Posts
10/25/2026 at 9:14 PM #265815 quote

LivJoJade
Participant

Junior

I would like to thank those who have taken the time to contribute to this discussion. I genuinely appreciate the knowledge, patience and suggestions offered by the experts here.
Having compared the live trading and ProBacktest results and explored the possible causes, I think I have to reluctantly accept that identical entry levels and trade sequences may not be achievable in practice. Together we have not found a definitive solution, and I feel it is time to stop chasing exact replication.
That doesn’t mean the investigation has been wasted; it has helped me understand the limitations and focus on what matters most.
I think the next step is to concentrate on the stop-loss and target levels, and how best to manage these when live execution prices differ from those recorded in ProBacktest. The aim is to make Precision 10 as consistent and robust as possible in real trading.
As a practical next step, I propose testing the suggested approach of using absolute stop-loss and target prices, anchored to a consistent reference price rather than allowing the different live entry fill to shift those levels.
I would do this in a separate copy of the strategy, retaining our existing 12-point stop-loss and 10-point target, and compare the results to establish whether this reduces the divergence between live trading and ProBacktest.
If anyone has further advice on the best way to implement this in ProOrder, I would be very grateful. Otherwise, I think this gives us a sensible direction to pursue.
Thanks again to everyone who has contributed their time and expertise. It has been greatly appreciated.

GraHal thanked this post
Author

Posts