Minutes 0–3: choose a market and check the data
Begin in Market Overview. Select a pair you understand rather than scanning for the largest green percentage. The live quote, embedded chart and model may come from different providers and update at different times. A chart moving down does not, by itself, mean a new sell has passed the model. Check the quote age and whether the market is open. If a scan is unavailable, treat its output as unknown.
The terminal orders Trader Radar, Market Overview, Live Chart, Model Output, Economic Calendar and Current Model Wins/Losses in that sequence. Quick view hides the longer research sections below them. It does not change the engine or turn a watch-only bias into a trade.
Minutes 3–6: read the signal state
- New paper signal: a newly published model entry with stop and target, recorded in the forward ledger.
- Open paper position: an earlier publication remains unresolved. Its original entry stays fixed. It is not an instruction to enter at today's quote.
- EARLY paper: technical alignment exists, but historical evidence or timing validation is incomplete. No live probability or Kelly allocation is assigned.
- Developing: a direction is visible, but there is no published entry. Keep it on a watchlist if useful.
- WAIT: a scan completed without a qualified new setup.
- Unavailable: the scan or data could not be obtained. This is not evidence for buying, selling or waiting.
One open signal can prevent another correlated signal from being published. For example, two different currency pairs can express the same USD direction. The portfolio filter treats this as overlapping exposure. It can reduce frequency even when several charts appear aligned.
Minutes 6–9: write a plan before considering execution
Use Your Trade Workspace under Model Output. Add the market to your watchlist, then write a short note with the reason you are monitoring it, the condition that would invalidate the idea and the event you need to check. Notes and watchlists are stored only in the current browser. They do not follow you to another device. Export a copy if you need a durable record.
Download Trade Plan includes the original entry, stop, target, tier, publication time, macro status and your note. It is enabled only for a published candidate that is still fresh or an existing open paper position. A developing bias cannot export invented trade levels. An open-position export is explicitly labelled as an existing position, so it should not be confused with a new entry.
The position-size worksheet requires account equity, your chosen risk percentage and your broker's tick size, tick value per lot and lot step. Tick value must be expressed in your account currency. Different brokers and contract types can use different specifications, particularly for gold and crypto. If you cannot verify them, use the broker's own calculator rather than guessing.
A hypothetical sizing example
Assume equity of 1,000 account-currency units and a chosen risk budget of 0.5%, or 5 units. Suppose a stop is 0.0020 price units away, the broker's tick size is 0.0001 and its tick value is 10 account-currency units per lot. The stop spans 20 ticks, so one lot has a nominal stop loss of 200 units. The unconstrained size is 5 ÷ 200 = 0.025 lots. With a 0.01 lot step, rounding down gives 0.02 lots and a nominal stop loss of 4 units.
This is arithmetic, not a risk recommendation. Commission, spread, slippage and gaps can increase loss. The worksheet does not check minimum lots, maximum volume or available margin. It assigns no size to EARLY paper research.
Minutes 9–12: review the event calendar
Check events relevant to both currencies. A EUR/USD position can be affected by euro-area and US releases. Calendar times are shown in your device's timezone; verify your device settings before relying on them. An event marked low importance is not a promise of a quiet market. Unscheduled announcements do not have a predictable calendar slot.
The engine checks high-impact event windows before a new publication. An existing paper position is still displayed when new entries are paused. That distinction matters: the event filter is a publication rule, not an automatic broker order or a guarantee that an open position is protected.
Minutes 12–15: inspect evidence, not just the win rate
Open the Track Record and read the model version, tier, sample count, unresolved positions, ambiguous bars and net R. A new model has a new current-model sample. Earlier outcomes remain in the archive; they are not discarded or silently treated as proof of the revised rules.
Win rate alone misses payoff size and costs. In a hypothetical series with six +0.5R wins and four −1R losses, the win rate is 60% but gross performance is −1R. If each of ten trades costs 0.05R, net performance becomes −1.5R. At 0.10R cost per trade, it becomes −2R. These are illustrative arithmetic scenarios, not GodzillaBTC backtest results.
Record your observation before the outcome, then revisit it. Do not adjust the original stop or target in your notes merely to make a past prediction look right. A paper result is not a broker execution record: actual fills and costs can differ.
Plain-language glossary
What is R?
R measures an outcome relative to the original entry-to-stop distance. A +1.25R target is 1.25 times that distance. Net R includes the engine's assumed costs; a gap can cause a loss larger than 1R.
What are ATR and regime?
ATR describes recent price movement in price units. A regime is the model's classification of current conditions, such as trend, range or unusually disorderly movement. These classifications can be wrong and change as candles close.
What is a holdout or OOS sample?
A later segment of historical data reserved for an out-of-sample check. The engine removes overlapping outcomes and separates training from holdout labels. Small samples remain uncertain; repeated rule changes based on those results can still overfit them.
What does Bayesian edge probability mean?
It is a model-based score derived from a historical sample and assumptions. It is not the probability that the next trade wins. EARLY paper discovery has no assigned live probability.
What is Kelly sizing?
A sizing calculation based on estimated payoff and success probability. If those estimates are unreliable, its output is unreliable too. GodzillaBTC uses reductions and caps; EARLY research gets zero allocation.
Your repeatable checklist
- Verify the market, quote freshness and signal state.
- Read original levels and distinguish an open position from a new candidate.
- Check events and your broker's contract specifications.
- Write your thesis and invalidation; export the paper plan.
- Review results including losses, costs and unresolved outcomes.
Continue with the methodology, position sizing, backtesting and the current validation status.