Bar Replay for Gold: Practice Sessions Like Live

Bar replay is how you practice a gold session without paying the session. You hide every candle that has not happened yet, you mark levels with only the past on the screen, and you take or skip the trade under the same rules you claim you use live. If you can see the future, you are not practicing. You are narrating a chart that already finished, which feels like skill and teaches the wrong reflex: certainty.

Replay is the drill. The standard for not lying about the result sits in backtesting your gold edge: rules first, then evidence. This page is the drill structure for XAUUSD, including a 20-session plan you can run this month. It does not include a performance number, because a number you did not measure is decoration.

Replay versus a static backtest

A static review lets you see the whole swing. You will "take" the entry that worked and forget the two that looked the same and failed. Your memory is a filter with a long bias. Replay removes the filter by removing the right-hand side of the chart.

What replay is good at: discretionary rules that still need a person. "Buy the retest if the daily zone is active and the five-minute candle closes back up" cannot be honestly tested by staring at a completed day. It can be tested by deciding, bar by bar, whether the daily zone was active before the close printed. The Asia range breakout is a clean example because the rules are binary: window, skip, close, retest, stop. You either followed them in the replay or you peeked and improved the entry.

What replay is bad at: proving a tiny edge quickly, and modeling the exact spread and slip you will get on a Tuesday release. You can include a spread assumption. You should label it as an assumption. You cannot replay your way to a tick-perfect fill. If the strategy only works when the fill is the best price on the bar, it does not work. Decide the fill rule before you start. Example: you are filled at the close of the signal bar, plus the spread in the unfavorable direction. Boring, and harder to fake.

Automated tests are a different tool. They are faster and they only test rules a machine can read. If your edge is "it has to look right," you do not have an automated test yet. You have a replay. Do not dress the replay up as a 10-year statistical study after twenty clicks.

Setting up a gold replay session

Pick one setup and one session. A replay of "whatever looks good on gold" produces a diary, not a sample. Write the setup name at the top of the page. If it is the Asia breakout, the window and the skip rules are copied from the strategy page, not restated from memory.

Then set the clock. Gold's session hours move with daylight saving and with your broker's server time. The replay feed has to use the same clock you will use live, or you will practice a London open that is an hour away from the one you trade. Mark Asia, London, and New York on the replay chart before you press play. TradYX Replay Lab is built for that XAUUSD session practice, next to the journal rather than in a separate hobby project.

Choose the start date without looking at how the day ended. A practical method: pick the date from a list of session opens, not from a day you remember as a trend. If you choose days because they "had a clean move," you have already peeked. Randomize from a calendar, or take the next twenty weekdays in a row and accept the dull ones. Dull days are where live traders force trades. If your replay has no dull days, it is not practice.

Set the timeframe you execute on, and the higher timeframe you use for bias, and do not add a third mid-trade. If you trade from the five-minute and the daily, those two are on the screen. A new 30-second chart introduced after a loss is a rule break. Tag it as one.

Write the fill rule, the stop rule, and the daily stop for the replay account. The replay account can be a notional $10,000 or the size of your real account. Use the real risk percent if the point of the drill is sizing. A replay that risks fantasy dollars teaches fantasy size. The position size formula still applies. A replay stop that you would not take live does not count as a strategy trade.

Hiding the future

The discipline rules are the product. The software is only the player.

• You do not scroll past the last revealed bar. Not "just to see if the idea was right." That glance updates your next decision. It is the leak.
• You do not rewind to take a better entry on the same signal. If you missed it, you missed it. Log the miss as a miss. Rewinding is how every replay becomes a highlight reel.
• You reveal bars at a human pace during the decision window. Speeding through Asia is fine if your plan does not trade Asia. Speeding through the London open, where the plan does trade, is how you skip the discomfort that live trading will not skip.
• You place the stop when you place the entry. A stop added after you have seen the next three bars is not a stop. It is a caption.
• News days stay in the sample unless your live rules exclude them. If you skip CPI in real life, skip it in replay and log the skip. If you trade it live and skip it in replay, the replay flatters you.

When you break one of these, the trade can stay on the chart and must leave the strategy average. Put it in a column called "peek" or "rewind." At the end of the block, the strategy column and the peek column are not allowed to merge. Merging is the lie the other backtesting note is about.

If the tool makes peeking easy, add friction. Cover the right side of the screen. Turn off the crosshair's ability to show later prices if the platform has that habit. Friction is a feature. Talent is not the scarce input here. The refusal to look is.

Log replay trades like live trades

A replay trade that is not logged did not happen. Use the same fields as the gold journal: session, setup name, planned R, actual R, spread assumption, level name, process score, and the revenge tag if you re-entered after a loss inside the replay. Yes, you can revenge-trade a historical chart. The loop does not need real money. It needs a red number and a play button.

Log before you reveal the outcome past your exit rule. Write "entered, stop here, target here" and then let the bars run until the stop or the target or the time stop. Editing the target after you see the high of the day is the same sin as scrolling ahead, done with a keyboard.

Separate three outcomes in the log:

• Strategy trade, rules followed.
• Rule break, including peeks, rewinds, and size that ignored the formula.
• Skip, with the filter that fired.

Skips are how you know the filter is alive. A 20-session replay with zero skips on a strategy that has skip rules means you did not use the rules. You used the entries.

At the end of each session, one sentence: what you would do the same live. Not a paragraph. The point is to make the replay transfer. A drill that stays in the drill folder does not change Tuesday.

A 20-session practice plan

Twenty sessions, one setup, one block. Here is a plan you can start without adding a new indicator.

Before the block. Freeze the rules on a single page. Freeze the clock. Freeze the fill assumption. Choose twenty dates without opening those charts. Decide the daily loss limit inside the replay, in R.

Sessions 1–5. The only goal is mechanical honesty. You expect to peek and to catch yourself. Every peek is tagged. You do not "fix" the strategy in this block. If the rule is ambiguous, write the clarification and restart the five. Do not apply the clarification only forward on the days it helps.

Sessions 6–15. The goal is a clean sample. Process score is the metric you look at daily. Average R is not discussed until the fifteen are done, and even then you only compute it on untagged strategy trades. If you have fewer than ten clean trades because the setup is rare, that is a result. Do not lower the standard to fill the quota.

Sessions 16–20. Trade them as if the journal will be graded. No new exceptions. This is the transfer block. If your behavior decays here, it will decay live. The decay is the finding.

After session 20. One page of review. How many clean trades, how many peeks, how many skips, average R on clean trades only, and the one rule you will not change for the next block. Then stop for a day before you touch a live ticket. The itch to trade live because the replay felt good is itself a tell. Live size starts at the small end of your formula, not at the size your replay confidence suggests.

Do a second block of twenty if the first was mostly peeks. You practiced honesty, not the setup. That is still progress, and it is not an edge. Only a block with a high process score is allowed to inform the question "does this idea deserve money?" Even then, twenty is a pilot. The backtesting note's warning stands: a small sample can flatter you. Replay's job is to stop you from flattering yourself with hindsight. It is not a license to bet the account on a month of drills.

When the drill and the live journal use the same tags, you can see the gap between them. If replay process scores are high and live scores collapse, the problem is not the strategy. It is the revenge loop and the missing lockout. If both scores are high and the R is poor, the strategy is the suspect. Replay is how you tell those two stories apart without paying full tuition for the lesson.