Gold Tempo TechnoTrader
- エキスパート
- バージョン: 1.1
- アップデート済み: 21 8月 2026
- アクティベーション: 10
This is not a holy grail and we will not sell it as one. It is one measured edge, executed with discipline, on real historical data.
Backtest, 3000 USD account, January 2022 to August 2026, defaults: +1442.4% in total, 82.3% a year, profit factor 1.56, 23.08% equity drawdown, 1444 trades.
Same test, the part most listings omit: worst month -14.3%, 16 losing months out of 55, longest losing run 3 months, longest wait between equity peaks 258 days.
The live account is days old, not years. Every figure on this page is a Strategy Tester result, and the live record is published as a signal under the same seller name.
Six gold instruments, one chart, one EA
Gold Tempo runs one strategy family: previous-day level breaks, M5 and M15 entries.
XAUUSD leads. XAUGBP, XAUEUR, XAUCHF, XAUAUD and XAUJPY are each restricted to the role it measured best at, not the same rules.
A hard six-hour backstop closes anything open. No weekend exposure, no swap accumulation.
Gold only: this EA wants volatility, and gold supplies it more reliably than most of what you can trade. Nowhere else.
Everything TechnoTrader ships has been rules-based for years: the live signals, and the two other expert advisors already on this Market, Swing Meridian and Index Cadence. No discretion, no overrides, nothing closed by hand.
Rules-based means checkable numbers. Every figure here is measured, not claimed, flattering or not.
What you actually get
- Six gold pairs from one XAUUSD M15 chart. One EA, one licence, not six copies.
- Each pair has its own measured session and confirmation.
- Nothing held overnight. No weekend gaps, no swap charges.
- Every order carries a verified stop. If the broker drops it, the EA restores it or closes the position.
- One setting to pick, the risk package. The rest ships as measured.
- The free demo runs in your Strategy Tester, on your broker's data, at these defaults.
What makes a trade happen
Four entry mechanisms. Each breaks a level the market drew itself: yesterday's high or low, or a recent range extreme.
| Mechanism | Chart | When it runs | Used by |
|---|---|---|---|
| Break of yesterday's high or low | M15 | European session | XAUUSD, XAUGBP |
| Break of a recent range extreme, quiet conditions only | M15 | all day | XAUUSD |
| Break of yesterday's high or low, quiet conditions only | M5 | all day | XAUUSD, XAUEUR, XAUCHF, XAUAUD, XAUJPY |
| Break of a recent range extreme | M5 | Asian session | XAUJPY |
Two gates sit above them. Each that applies must open.
The volatility gate refuses a stretched market, not a quiet one. A break into a move that already happened fails.
It guards only the two mechanisms marked "quiet conditions only". One fixed multiple of recent range, same for all six instruments, compiled in, not a setting.
The lead-confirmation gate blocks four of the five crosses unless XAUUSD holds a position in the same direction. XAUAUD is the fifth and trades free. Five measured restrictions on it all hurt the book.
Exits: a volatility-scaled stop-loss, a wider trailing stop that moves only in your favour, a time backstop per module.
Total open risk is capped across all instruments as one cluster.
Six gold pairs are one gold bet, not diversification. The EA treats them that way and so should you.
Without volatility nothing happens. It needs follow-through: a level gives way and price keeps going.
The weakest stretch in the test is the one where gold did the least.
Not trend-following. It needs range in the day, not a direction that holds for months. Active is not the same as chaotic.
Which pair does what
Each pair keeps the session and confirmation rule that measured best. The four most restricted are the four best per unit of volume traded.
XAUUSD leads, three mechanisms all day. XAUAUD trades free.
Crosses XAUEUR, XAUCHF and XAUJPY need XAUUSD already holding a position in the same direction. XAUJPY adds a second, Asian-session mechanism.
XAUGBP is confirmed too. Its two hours are not a setting, just where mechanism and instrument rule overlap, and a trade needs both, narrowest of the six.
Full role table in the screenshots.
Hours are broker server time, not GMT. Nothing to set.
An hour either way is inside the measurement, because the tested window holds nine daylight-saving shifts, moving the sessions an hour twice a year.
What it does not do
No martingale. No grid. No averaging into a loser. No news-spike scalping, no sub-minute trades, no hidden recovery logic, nothing held overnight.
Stops are verified after every entry, repaired or closed if missing.
Daily, weekly and monthly loss brakes block new entries, never open-position management.
On a netting account: one position per symbol (a second would merge), lead-confirmed module off entirely at start-up, journal note.
The figures on this page are from a hedging account. Use one.
Almost every gold EA prints that list. Starting point, not differentiator. What it decides is how an account dies.
Grid and martingale die on one move that never comes back, because the position grows while wrong.
Here every position is sized once at entry from your risk setting, broker-side stop on before the order is sent.
Six instruments share one open-risk cap. The whole book is bounded, not each trade.
That is not safety. Losses come as a grind of losing months, counted in the losing-months table, not one trade that ends the account.
Results
Strategy Tester, not account statements. Default settings, raw-spread data, 200 ms execution delay, M1 OHLC model, 1:100 leverage. No adjectives added.
Equity drawdown is MetaTrader's relative figure: larger and more conservative than the maximal one.
| Deposit | Jan 2022 - Aug 2026 | Per year | Equity DD | Profit factor | Trades |
|---|---|---|---|---|---|
| 3 000 (reference) | +1442.4% | 82.3% | 23.08% | 1.56 | 1444 |
| 1 000 | +1301.3% | 78.5% | 21.44% | 1.58 | 1468 |
| 500 | +1241.5% | 76.8% | 19.86% | 1.64 | 1470 |
Quote the 3 000 row. Both columns rise, as proportional sizing should, so the small rows measure the strategy, not the broker's lot floor.
That floor caps forced risk on a small account. Number in the risks section.
About 2.3 trades per active trading day, 617 trading days, longest hold six hours.
Gold: roughly flat 2022, about 13% in 2023, strong trend 2024. A mix lifted by the trend year, not the recent regime.
Same binary and settings, 2022 to 2024 alone: +215.1% over three years, 46.9% a year, profit factor 1.30, equity drawdown 21.46%, 713 trades.
That drawdown is under the full-window figure, so the deepest decline came in strong 2025-2026, not the weak years.
Not regime-proof, and we do not claim it is. The lead symbol's 2020 and 2021 show a bad regime for this mechanism.
The risks, and the numbers they come from
The window is a high-volatility stretch for gold, not a calm sample.
Every figure here is a measurement, not a promise, including the unflattering ones.
Listings publish depth. Almost none publish duration, the harder half.
A decline that recovers in a month is a different product from one that takes a year. Both are below, on the 3000 reference account over the full window.
| What you would have lived through | Measured |
|---|---|
| Worst single month | -14.3% |
| Losing months | 16 of 55 |
| Longest run of losing months | 3 months, twice |
| Longest run of losing trades | 14 trades (typical run: 3) |
| Deepest decline on closed trades | 21.0%, 136 days to recover |
| Longest wait to a new equity peak | 258 days |
The two three-month runs were April to June 2022 and June to August 2023. No fourth consecutive losing month appears, and nothing in the design prevents one.
The 258-day line is the one that ends subscriptions. May 2023 to February 2024, never more than 15% down, eight and a half months back to the old peak.
If you would switch the EA off mid-way through that, this is not your product. Better to learn it here than with money on it.
Losing trades are the mechanism, not a malfunction. About 44% of trades win.
Average win 190.16 against average loss 93.51. Fewer, larger wins pay for many small losses.
Expect far more losers than winners while the account does what it should.
The trades are fast, the product is not. Positions last about two hours, but a system with three-month losing runs is judged in quarters and years, not weeks, and never trade by trade.
Leave it alone. It works best when nobody closes its trades by hand.
Every position carries its stop. The six instruments are one gold bet under a shared risk cap that counts what is actually open.
Close one leg yourself and the book is no longer the one that was measured. If a position feels too large, change the risk package or the deposit, not the open trade.
The rest of what can go wrong:
- Regime product. Every figure comes from the window where all six instruments exist on a broker feed. Nothing older is shown, though the lead symbol alone has years more history. One instrument at portfolio risk over a period this package was never built or measured for is a different test under the same name, not evidence about the package. It does show one thing: losing years happen with this kind of mechanism. No year here is negative, and that is not a floor, because the window starts in 2022.
- A small account changes who sets position size. Below roughly 3000 the broker minimum lot decides instead of your risk package, and it decides upward: one forced trade can carry up to 4.0% of balance on the lead and 2.0% on a cross. Settings explains that ceiling and how to lower it. The deposit ladder is well behaved, return and drawdown both rising with it, but the figures here describe the 3000 row.
- The gold crosses have wider spreads than XAUUSD. The EA demands proportionally more room from them and skips entries when the live spread is too wide. Cross trades are fewer by design.
- The live record is weeks old at best, the single biggest limitation here. It runs as Techno Gold Tempo, a real account on these defaults, nothing changed, so you can watch it age. No backtest shows slippage on a real fill, a broker widening spreads in a fast market, or a requote. Honest status: measured carefully, not yet proven forward. The free demo runs in your own tester on your own broker's data, and that result, not this page, is what your account would have seen.
Past performance does not guarantee future results, meant literally and not as small print. Everything above measures what already happened on one broker's data, and none of it is a forecast.
Risk packages, one input
Most buyers touch only this setting. The three differ in risk per trade, same signals, sessions and exits. Custom takes your own number.
| Package | Risk per trade | Total | Per year | Equity DD | Trades |
|---|---|---|---|---|---|
| Conservative | 1.00% | +541.1% | 50.4% | 15.75% | 1466 |
| Balanced (default) | 1.50% | +1442.4% | 82.3% | 23.08% | 1444 |
| Aggressive | 2.00% | +3749.0% | 122.8% | 28.26% | 1419 |
Return roughly doubles, drawdown about half that: return compounds, drawdown does not. Trade count barely moves. Size changes, signals do not. Too much drawdown? Turn the package down. It costs return, not trades.
The same dropdown's fifth option is fixed lot: risk sizing off, one field you set, in two shapes:
| What you type | What it does |
|---|---|
| 0.02 | Sets the lead. The five crosses derive from it on the same weights the risk mode uses, so the shape of the basket stays as it was measured. |
| 0.02,0.01,0.01,0.01,0.01,0.01 | Sets every pair yourself, in symbol-field order. No weighting is applied, because you were explicit. |
Anything else and the EA refuses to start, naming what it found, rather than quietly trade an unintended size.
The total open-risk cap still applies, computed from lot and stop distance, not risk percentage. The results on this page are risk mode, not fixed lot, where outcome depends entirely on your lot and balance.
Year by year
A four and a half year average hides where the money came from. Same test by year, continuous account:
| Year (3 000 deposit) | Start | End | Return |
|---|---|---|---|
| 2022 | 3 000 | 4 104 | +36.8% |
| 2023 | 4 104 | 5 706 | +39.0% |
| 2024 | 5 706 | 9 454 | +65.7% |
| 2025 | 9 454 | 21 692 | +129.4% |
| 2026 (7 months) | 21 692 | 46 272 | +113.3% |
No losing year in the window. Worst +36.8%, best +129.4%, a ratio of about one to three and a half.
16 of 55 months lost, worst -14.3%. Recent years carry most of the total.
The same EA on eight brokers
These columns measure execution, not instrument count. Every number is XAUUSD alone, the one instrument all eight feeds carry, so the signals are identical and only the fill differs. Hence below the 82.3% on this page: that is six instruments, these are one.
An EA result is partly strategy, partly your broker's fill. Coverage is the other broker question, and it is the one nobody warns you about. Check it before you buy.
Most brokers do not carry all six gold pairs. Of our eight, two do.
The EA skips a missing cross, notes it in the journal and runs on. With two pairs you run a different EA than this page.
Lead symbol: 35.9% to 68.5% a year, 913 to 967 trades, a 5.9% spread in trade count.
Two rows are a shorter window, marked not comparable in the screenshots.
Of the six comparable feeds, our own broker is the top row. We say it before you do.
Broker and platform names are their owners' trademarks, naming only the feed a test ran on. TechnoTrader is not affiliated with them, and this is no broker ranking.
Why this period, and why we did not make it prettier
Not our longest window. The longest one with real data for all six instruments.
XAUGBP and XAUCHF have broker history only from 24 January 2022. Earlier, a backtest runs four instruments and reports six.
That hit us in development: a four-instrument figure was briefly published as six.
The test starts 1 January 2022 with four, the other two join 27 January, and the EA journals both activations for you to check.
Data availability chose the window. We did not shop for a kinder one.
We did not tune until it passed either.
The first three years originally lost money. Instead of curve-fitting that away we measured why: profit factor just under 1.00, so the edge existed and costs ate exactly it.
Two measured changes fixed it: a wider stop, meaning a smaller position at fixed risk and proportionally less spread, then per-instrument roles.
Neither was found by trying values until the window looked better.
Both came from a regime inside this window, so these figures are partly in-sample, and forward evidence is only as long as the live account.
Trade count is a choice, and the volatility gate is its one lever.
Count moves more than fivefold across that gate's full curve, both ends hold a losing year, and the extra trades are bad ones.
The best setting per unit of drawdown is not the one we ship. It takes the default past a 30% equity drawdown and adds two losing months.
The gate is also the most period-bound part: relative to each instrument's own recent range, but its multiple came from high-volatility gold. Quiet gold is our thinnest evidence.
What we tried and threw away
All were built and measured. Most did not ship.
- Counter-trading a losing streak. Autocorrelation 0.3460, 33.69% wins after a loss against 56.97% after a win. Sorted by close distance it falls to 0.0093 at four hours: six instruments on one gold move, not one outcome predicting the next.
- Fixed take-profits instead of the trailing stop. A split, not a rejection. Trailing wins the full window on trend tails, fixed wins the harder 2022-2024 stretch, where trailing gives back part of every range reversal. Not shipped: shipping either guesses the next regime.
- A trend filter on the 200-day average. Offline it looked like a large improvement. Lost on every window in the engine: the trades it cut were part of the recoveries.
- Break-even stops. Win rate rises to 66 to 76% and looks like a breakthrough. Profit factor flat, return down, drawdown up. Outcomes redistributed, no value created.
- A second search outside price levels: the two London gold fixings, Monday opens, month-end, volatility state and term structure, tick-volume absorption, volume-price divergence, candle wick structure. Sixteen mechanisms, none survived. The fixings did not beat placebo clocks set an hour early and an hour late, and the largest reading in the table was a placebo.
- Volume drought into a volume burst. Not a failure, it does not exist: 14836 quiet stretches over 6.6 years, 667 bursts, one coincidence.
- A mean-reversion engine we hold, profitable on its own. On gold M15 it fires on the level book's bars 51 to 57% of the time and takes the opposite side on over 99% of them. Not diversification, two positions cancelling and paying two spreads.
The previous-day level break is the only mechanism positive on all the gold data we hold. The only reliable way to improve the portfolio was narrowing each instrument's job, not adding a mechanism. Narrow because we looked, not because we did not.
Settings
43 inputs in 11 groups, and the strategy is almost none of them. Session windows, volatility gate, stop and trailing distances and module assignment are compiled in at measured values.
You set risk, protection and infrastructure.
Two exceptions. The six symbol fields are inputs, and the role sits in the field, not the symbol name. Clearing a field turns that instrument off. Move one elsewhere and it takes a role measured for a different instrument, not what was tested. Pyramiding is the second: entry logic, yours to switch on, shipped off because off measured best.
- Risk package, default Balanced. Very low on a small account is not safer: it pushes size under the broker's minimum lot, which then decides for you.
- Cross weight, default 0.5. Five crosses at half the lead's risk. Six gold pairs are one bet, the crosses its correlated half, so raising it buys more of that bet, not diversification.
- Cluster cap, default 3.0. Total open risk across the six instruments, in units of per-trade risk.
Also: 5, 10 and 15% daily, weekly and monthly loss brakes, optional year-end pause, news blackout on MetaTrader's own calendar, prop-firm maximum-drawdown and daily-target stops, entry and lot jitter for multiple accounts, panel appearance.
The Strategy Tester has no economic calendar, so the news filter there uses the monthly US jobs report alone.
Forced-risk ceiling, before you pick a deposit. If your package asks for less than the broker minimum lot, the EA trades that lot anyway and logs the real risk. The ceiling is the lower of your input and your risk times the cluster cap: on Balanced 4.0% of balance for a lead trade, 2.0% for a cross. The cap works on real risk, so the total stays bounded, but one forced trade can use most of it. Set the ceiling lower, or add capital.
What the panel shows
Equity, day and week profit and loss, open cluster risk vs the cap, a row per instrument with spread vs its ceiling, room under all three loss brakes, minimum-lot mode, and a warning if the account is netting, not hedging.
By design, no live row for the volatility gate, session windows or lead confirmation. Their per-gate counts print once at the end of a Strategy Tester run.
Requirements
Raw-spread or ECN account. Hedging, not netting, the netting case is above.
One XAUUSD M15 chart runs all six instruments.
Crosses your broker lacks are skipped automatically and named in the journal. Counts: broker table above.
The lead symbol is not optional. Clearing a symbol field while that instrument holds a position blocks startup with a reason, rather than orphan the trade.
Deposit: 500 floor, 3000 behind this page's figures.
Support: the Comments tab, honest questions get honest answers. Post untested-broker demo results there, it helps the next buyer.
Start on demo, and not for a week: losing runs reach three months. Watch it trade, read the panel, and decide whether you can sit beside it. No number on this page conveys the 258-day wait.
Go live once you know your answer.
