Market risk researchCalm before the run? The market when a gold grid basket starts
Huyen X. Vu · Quantex Research, AlgoTrade Pte. Ltd., Singapore · ORCID 0000-0002-2209-842X · Download PDF
Abstract
Does the market at the moment a grid basket opens tell us which baskets will grow to 20 orders or more? Using only one-minute prices before the first order, we measure the last hour's range, the day's range, the four-hour trend and the place of price in the day's range, over three periods from 2018 to October 2026. In every period, baskets opened in the calmest fifth of the last hour or of the day reached 20 orders about two to three times as often as those opened in the busiest fifth (2025–26: 1.56% against 0.62% for the last hour). None of the simple screens we tried removed the deepest basket of each period. The main test, calm against busy last-hour starts, was set on 23 October 2026 and is checked on the live account.
Keywords: grid trading; gold; XAUUSD; volatility; market state; live test
JEL: G11; G17; G32
1. Introduction
A few grid baskets grow very large and carry most of the risk. If the market at the moment a basket opens already hinted at that, a model could step aside. We measured four simple market states at the first order of every replayed basket since 2018, and found one consistent pattern and one clear limit.
Where this project stands
| Stage | What it means here | Status |
|---|---|---|
| 01 Review | Question and data | Done |
| 02 Backtest | Replay of the model on one-minute prices, January 2018 to 9 October 2026 | Done |
| 03 Fine-tune | Not used: the model itself is not changed in this study | Skipped |
| 04 Live test | Live watch of the running account, with tests set in advance. Watch only: the account trades as before. | Now, from 23 Oct 2026 |
| 05 Handed over | Only if a finding is confirmed and a later study shows it is worth using | Not started |
Research question
Using only prices before the first order, do the market's recent range, the day's range, the four-hour trend or the place of price in the day's range tell us which baskets will reach 20 orders?
2. Data and method
The model we watch
Altair, a gold (XAUUSD) grid model shown on the AlgoTrade live board and studied in 26.07.ALW. When price moves against it, it adds orders on the same side, each larger than the one before, and closes them together. One such group is a basket. As with all our models, we do not publish its rules.
Measures, taken from the bars before the first order
- Last-hour range: high minus low of the 60 one-minute bars before the first order, divided by the median of all such 60-minute ranges over the previous 7,200 one-minute bars (about five trading days).
- Day's range: high minus low so far today (broker day), divided by the average high-to-low range of the previous 20 full days.
- Four-hour trend: close minus the close 240 bars earlier, divided by the high-to-low range of those 240 bars, and signed so that a positive value means price has been moving against the basket's side (down for a buy, up for a sell).
- Place in the day's range: where price sits between today's low (0) and high (1), turned so that 1 means the end that is bad for the basket (the day's low for a buy, the day's high for a sell).
A fifth is one of five equal groups of baskets ranked by a measure. "Calmest" means the lowest fifth; "busiest" the highest.
Outcome: the basket reaches 20 orders. In the backtest, fifths are set within each period. Live, the cut-offs are fixed in advance at the 2025–26 fifth boundaries (below), so live groups need not be exact fifths.
Data and replay
The backtest replays the model's order rules, as seen in its live orders, on one-minute gold prices: January 2018 to December 2021 from one price source (fixed spread) and May 2022 to 9 October 2026 from a broker's own one-minute history with its recorded spread (January to mid-May 2022 is not covered). Orders over the broker's 500-lot limit are capped at 500 lots. The replay was checked in two steps. First against a tick-by-tick platform run of the same rules on the same broker: for 2025 both had the same deepest basket (105 orders) and a net result within about 7% of each other. Then the platform run against the live account for January to 9 October 2026: 53,020 baskets against 47,072, and a deepest basket of 92 orders against 84. Most of that gap is from before 12 August 2026, when the model's order-size setting changed: before that date the live account had 87% of the platform run's baskets, after it 96%. So most of the gap comes from the replay using today's setting; the rest from price feed and fill differences. Swap and the live account's exact fills are not modelled. The model's order-size setting changed on 12 August 2026; the replay uses the current setting for the whole period. Price extremes inside a minute come from one-minute highs and lows.
Periods: results are shown for 2018–2021, 2022–2024 and 2025–October 2026 separately. A pattern counts only if it holds in all three.
Live counting. The live report counts only baskets whose first order is on or after 23 October 2026. A basket still open counts as soon as it has passed a test's limit; other open baskets wait until they close. The final read covers baskets that start in the first 15 weeks and is taken one week later, so that almost all of them have closed; any still open are listed. Days and times follow the broker's server clock (UTC+2 in winter, UTC+3 in summer).
How the live tests are read. The main 95% ranges treat baskets as independent. Baskets bunch together in busy weeks, so the final report also gives each range from resampling whole weeks (Monday to Sunday, broker time; 1,000 resamples with replacement; the middle 95% of the results), and a test counts as confirmed only if it passes on both. A basket counts as closed if it closed before the cut-off time of the final read; a position is one order when it opens, however it is filled. The test plan on this page was fixed on 23 October 2026, before any live basket, and a dated copy is kept in our research register. Results will be published whether they confirm, fail or stay unclear.
What we test live (set 23 Oct 2026, before any live case)
| Test | Backtest, 2025–26 | Confirmed if |
|---|---|---|
| H1, main test: baskets that open in a calm last hour (ratio 0.736 or below) reach 20 orders more often than those that open in a busy one (1.777 or above) | 1.56% vs 0.62% (2.5×) | ratio at least 1.5 and lower end of its 95% range above 1 |
| H2, secondary: the same for the day's range (0.362 or below vs 1.044 or above) | 1.67% vs 0.77% (2.2×) | ratio at least 1.5 and lower end of its 95% range above 1 |
Sample and timing. About 975 baskets a week closed in 2025–26, so the calm and busy groups should each hold roughly 2,900 baskets after 15 weeks. The live 95% range of the ratio uses the standard log method. Treating baskets as independent, the full rule passes about 94% of the time for H1 and 88% for H2 (200,000 simulations; 83% and 74% with 2,000 per group). Baskets bunch in busy weeks, so we also drew 15 whole weeks at random from the 93 weeks of 2025–26 replay (2,000 draws) and applied both the basket-level and the week-level ranges: H1 then passes in about 85% of draws and H2 in only about 50%. The live test therefore rests on 15 weekly units, not 2,900 independent baskets. H1 is the only main test. H2 is reported as a secondary result and is not used on its own to claim the finding. With shares near 1.5%, each group will hold only about 20 to 45 baskets that reach 20 orders, so the result will stay uncertain. Until the recorder holds five days of prices (last hour) or twenty days (day's range), new baskets are counted as not measured. Final read in early February 2027; all results will be published.
3. Results
What the backtest found
| Measure | 2018–2021 | 2022–2024 | 2025–Oct 2026 |
|---|---|---|---|
| Last-hour range | 9.9% vs 4.1% | 9.0% vs 3.2% | 1.6% vs 0.6% |
| Day's range | 10.4% vs 4.7% | 9.5% vs 4.7% | 1.7% vs 0.8% |
| Four-hour trend (with the basket vs against it) | 8.9% vs 6.1% | 6.8% vs 5.6% | 1.1% vs 1.0% |
| Place in the day's range (good end vs bad end) | 10.1% vs 6.0% | 7.0% vs 5.4% | 1.3% vs 1.2% |
1. Calm starts grow more often. In all three periods, baskets that opened in the calmest fifth of the last hour or of the day reached 20 orders about two to three times as often as those that opened in the busiest fifth. Trend and place in the day's range gave smaller and less steady gaps.
2. But none of the screens we tried removed the worst baskets. We tried simple screens that skip the calmest or busiest 5–30% of starts on each measure, chosen on 2022–2024. None of them made the deepest basket of every period smaller: the deepest baskets (May 2018, October 2023, May 2025) opened in ordinary conditions, and a screen that removed October 2023 left the worst baskets of the other periods in place. This is a result about these screens only, and the deepest basket is a single, noisy observation; it does not show that no screen could help.
The four measures and the screens were exploratory. Only the two tests below are set in advance.
Live report
Rebuilt from the running account's orders, recorded to the second. Tables refresh every 5 minutes.
Loading the live report…
Interim reports
Week 1, data to 23 Oct 2026, 20:56 UTC. One trading day (23 Oct), 223 closed baskets. This is an early read, not a result: the first look is on 18 Nov 2026 and the final read around early February 2027. The live tables above keep updating; the numbers below are frozen at the cut-off.
No basket could be measured yet (223 of 223 without a measure, for both the 60-minute range and the day's range). Each measure needs history before the basket starts: the 60-minute range is compared with about 5 trading days of one-minute bars, and the day's range with the 20 days before. So week 1 is a warm-up, not a result. Baskets in the warm-up period stay without a measure, which makes the live sample smaller.
4. Limitations
Scope: what this study cannot tell
- Association, not cause. A calm start is not a reason a basket grows.
- It does not show that skipping calm starts would help an account: those baskets also close in profit most of the time.
- Four measures only. News times, sessions and longer trends are left for a later study.
- The replay uses today's order-size setting for all years. Under the same setting, from 12 August 2026, it makes about 4% more baskets than the live account. The live test, not the replay, is the check.
Declarations
- Author: Huyen X. Vu, Quantex Research, AlgoTrade Pte. Ltd., Singapore (ORCID 0000-0002-2209-842X).
- Data availability: Live results are published in aggregate on this page and updated every five minutes. Order-level data can be shared with reviewers under a confidentiality agreement. Model rules are not published.
- Use of AI tools: Data processing, analysis code and drafting were assisted by AI tools (Claude, Anthropic). The author checked the results and is responsible for the content.
- Conflict of interest: The author is the founder of AlgoTrade Pte. Ltd., which develops and runs systematic trading models, including the model studied here.
- Funding: No external funding.
- Peer review: Not yet peer reviewed. Comments are welcome at info@quantex.com.sg.
- How to cite: Vu, H. X. (2026). Calm before the run? The market when a gold grid basket starts. Quantex Research, AlgoTrade Pte. Ltd. https://quantex.com.sg/research/altair-start-state/