- Entry
- 1.11435
- Stop
- 1.11236
- TP1
- 1.11634
- TP2
- 1.11833
- TP3
- 1.12033
Every signal publishes its entry zone, stop and three targets before the trade happens. We then follow it against live prices until it ends, and nothing is edited afterwards. Losses stay on the page, each with a review.
Practice trades, no real money Tracked against live prices. Not investment advice.
Tracking since 6 Nov 2026. Results are counted after spreads, with stops filled before targets whenever a bar touches both.
Updates live
Every closed trade in the order it closed, adding up its result in R.
| Risk | Closed | Win rate | Total | Average |
|---|---|---|---|---|
| Cautious | 0 | - | 0.00R | - |
| High Risk | 72 | 49% | −12.74R | −0.18R |
Each signal is tagged Cautious or High Risk when it is published, and the tag never changes. This table shows whether High Risk signals really do worse.
| Type | Closed | Win rate | Total | Average |
|---|---|---|---|---|
| Scalping | 38 | 42% | −6.74R | −0.18R |
| Intraday | 12 | 42% | −5.04R | −0.42R |
| Swing | 4 | 50% | −1.22R | −0.31R |
| News trading | 18 | 67% | +0.25R | +0.01R |
| Strategy | Closed | Win rate | Total | Average |
|---|---|---|---|---|
| AI scalping desk | 24 | 42% | −1.62R | −0.07R |
| EMA 9/21 scalper | 12 | 33% | −6.86R | −0.57R |
| Post-news drift | 8 | 63% | −0.40R | −0.05R |
| Bollinger Band fade | 6 | 17% | −4.64R | −0.77R |
| Pre-news breakout | 5 | 80% | +1.04R | +0.21R |
| AI news desk | 5 | 60% | −0.39R | −0.08R |
| SMC fair value gap | 4 | 75% | 0.00R | +0.00R |
| RSI(2) pullback | 4 | 50% | −1.22R | −0.31R |
| Gold hourly breakout | 2 | 100% | +1.74R | +0.87R |
| SMC liquidity sweep | 2 | 50% | −0.40R | −0.20R |
| Instrument | Closed | Win rate | Total | Average |
|---|---|---|---|---|
| Gold (XAU/USD) | 16 | 56% | −2.56R | −0.16R |
| Nasdaq 100 (US100) | 14 | 43% | −4.85R | −0.35R |
| EUR/USD | 10 | 70% | +1.78R | +0.18R |
| USD/JPY | 9 | 22% | −4.41R | −0.49R |
| Bitcoin (BTC/USD) | 8 | 38% | +0.36R | +0.05R |
| GBP/USD | 7 | 57% | −0.61R | −0.09R |
| USD/CAD | 4 | 50% | −1.35R | −0.34R |
| WTI Crude Oil (XTI/USD) | 2 | 100% | +0.88R | +0.44R |
| AUD/USD | 1 | 0% | −1.00R | −1.00R |
| Ethereum (ETH/USD) | 1 | 0% | −0.99R | −0.99R |
| Source | Closed | Win rate | Total | Average |
|---|---|---|---|---|
| Data rules | 43 | 51% | −10.74R | −0.25R |
| AI desk | 29 | 45% | −2.00R | −0.07R |
| Confidence | Closed | Win rate | Average |
|---|---|---|---|
| 30-39 | 16 | needs 4 more closed | |
| 40-49 | 47 | 49% | −0.13R |
| 50-59 | 8 | needs 12 more closed | |
| 60-69 | 1 | needs 19 more closed | |
Each signal gets a confidence score from an AI model. A band shows its real win rate only after 20 signals in it have closed.
Each strategy is written down in plain rules before it trades. The backtests show how the data rules would have done on old prices, which is not a promise about the future. The live record beside each one is what has really happened since tracking began.
The most common scalping setup: on 5-minute bars, buy at market when the 9 EMA crosses above the 21 EMA while price is above the 200 EMA, and sell on the mirror image. London and New York hours only (07:00 to 20:00 UTC, weekdays), at most six a day per instrument. Stop 1.5 ATR (5-minute), targets 1R, 2R and 3R; the rest closes when the EMAs cross back, or after two hours.
The idea, explained: Forex scalping
| Quarter | 2025 Q4 | 2026 Q1 | 2026 Q2 | 2026 Q3 | 2026 Q4 |
|---|---|---|---|---|---|
| R / trade | −0.12R | −0.06R | +0.05R | −0.01R | −0.04R |
1,814 trades a yearprofit factor 0.94
Caveat. One year only (the price feed keeps a year of 5-minute bars): -0.03R per trade over 1,814 trades, gold -0.02R, Nasdaq 100 -0.04R. On EUR/USD, GBP/USD, USD/JPY and Bitcoin the same rule lost 0.17 to 0.20R a trade, because a 5-minute stop is only a few spreads wide, so it is not run there; that choice was made on the same data.
The best of the rules from our October gold scalping study. At the close of each London-hours candle (02:00 to 08:00 New York), when hourly volatility is at or above its 20-day average, a buy stop waits just above the candle's high and a sell stop just below its low for one hour; the first to trigger cancels the other, and each one's stop is the other's entry. Targets 1R, 2R and 3R; trades close after two hours at most.
The idea, explained: Gold scalping
| Quarter | 2025 Q4 | 2026 Q1 | 2026 Q2 | 2026 Q3 | 2026 Q4 |
|---|---|---|---|---|---|
| R / trade | −0.17R | −0.15R | −0.06R | +0.11R | −0.17R |
165 trades a yearprofit factor 0.82
Caveat. Gold only. -0.07R per trade over 165 trades in the last year. The 2023-2026 study on 5-minute gold found this structure close to break-even before costs and slightly negative after them; no hour of the day was reliably positive.
Every hour of the London and New York sessions, ForexR AI reads 5-minute, 15-minute and 1-hour bars with the same facts pack the analysis desk uses and proposes at most one quick trade per instrument (orders wait up to an hour, trades last up to three hours), or none. Code checks every level against the 5-minute ATR before it is published; it stands aside when a high-impact release is due within 30 minutes.
The idea, explained: Forex scalping
An AI desk has no backtest, because its decisions cannot be replayed on old prices. Its record starts with these signals.
The rules of the ForexR Gold London Breakout EA. The Asian session range (18:00 to 02:00 New York time) is bracketed with a buy stop 50 cents above its high and a sell stop 50 cents below its low; the first to trigger cancels the other. It only trades when the range is at least $5 and hourly volatility (ATR 14) is at or above its 480-hour average. Orders wait until 08:00 New York; open trades close at 17:00 New York. Targets are 1R, 2R and 3R.
The idea, explained: The London breakout strategy
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | +0.01R | +0.11R | +0.11R | +0.18R |
58 trades a yearprofit factor 1.35
Caveat. 2023 was about break-even. The volatility filter was found on the same data, so it is not a clean out-of-sample test.
Bollinger Band strategies are the second most starred family in our Pine library (99 scripts, 6,061 stars; for example Bollinger Bands Winner). On 1-hour bars, buy at market when price closes below the lower band (20, 2) with RSI 14 under 30, and sell when it closes above the upper band with RSI over 70. Stop 1.5 ATR, targets 1R, 2R and 3R; the rest closes back at the middle band.
The idea, explained: Bollinger Bands
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | −0.10R | −0.11R | −0.09R | −0.10R |
178 trades a yearprofit factor 0.80
Caveat. Average per trade across the 11 instruments tested (7 FX majors, gold, silver, Bitcoin, Ether), costs included. Lost in every year on average; positive on only 2 of 11 instruments (AUD/USD +0.04R). Not tested on WTI or the Nasdaq 100.
Supertrend is one of the most used trend indicators (75 strategies, 2,162 stars in our Pine library; for example Supertrend + EMA rebound). On 1-hour bars, buy at market when Supertrend (10, 3) flips up and sell when it flips down. The stop is the Supertrend line (at least half an ATR away), targets 1R, 2R and 3R; the rest closes when it flips back.
The idea, explained: The ATR, which Supertrend is built on
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | −0.03R | −0.07R | −0.03R | −0.03R |
98 trades a yearprofit factor 0.87
Caveat. Average per trade across the 11 instruments tested (7 FX majors, gold, silver, Bitcoin, Ether), costs included. Slightly negative every year on average; positive on 3 of 11 instruments (best USD/JPY +0.08R). Not tested on WTI or the Nasdaq 100.
Smart Money Concepts are the most actively maintained family in our libraries (89 of 91 Pine strategies updated in the last year; SMC expert advisors such as SMC Unified EA). On 1-hour bars: when price trades above the previous day's high and a candle closes back below it within 6 bars, sell at market with the stop above the sweep high (plus 0.1 ATR); the mirror for a sweep of the previous day's low. Targets 1R, 2R and 3R; max 48 hours.
The idea, explained: Smart money concepts
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | −0.16R | −0.14R | −0.16R | −0.09R |
178 trades a yearprofit factor 0.77
Caveat. Average per trade across the 11 instruments tested (7 FX majors, gold, silver, Bitcoin, Ether), costs included. Negative on all 11 instruments tested (best USD/JPY -0.04R). Not tested on WTI or the Nasdaq 100.
Fair value gaps are the core of the ICT / Smart Money Concepts style, the most actively maintained family in our libraries (for example the FVG Gold EA and ICT Gold EA, both updated in 2025-26). On 1-hour bars, a bullish gap is a candle whose low sits above the high from two candles earlier, at least 0.3 ATR tall, while the 4-hour 50 EMA is rising: a buy limit waits at the middle of the gap for 6 hours, with the stop just under the gap. Bearish gaps mirror it. Targets 1R, 2R and 3R; max 48 hours.
The idea, explained: Fair value gaps
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | −0.19R | −0.21R | −0.18R | −0.13R |
133 trades a yearprofit factor 0.67
Caveat. Average per trade across the 11 instruments tested (7 FX majors, gold, silver, Bitcoin, Ether), costs included. Negative on all 11 instruments tested. Hourly bars cannot show whether a target was touched before the limit filled, so the test gave no credit for that; the true figure may be a little better. Not tested on WTI or the Nasdaq 100.
The rules of Gold ORB, the most starred gold expert advisor on GitHub (yulz008/GOLD_ORB, 311 stars, 102 forks). The 18:00 New York hourly candle sets an opening range, which grows while new candles make higher highs (or lower lows) and freezes after three candles that do not. A bullish candle that closes with its body above the frozen high buys at market; a bearish one below the low sells; at most one of each per day. Stop at least $4 (half an ATR), targets 1R, 2R and 3R; max 24 hours.
The idea, explained: Breakout trading
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | −0.08R | −0.15R | −0.02R | −0.01R |
274 trades a yearprofit factor 0.87
Caveat. Gold only. The EA's own fixed $12 target and trailing stop were replaced by the 1R/2R/3R thirds used for every signal here; with its own settings it was about -0.10R per trade.
ForexR AI reads the same facts pack the analysis desk uses (chart levels, trend, volatility, the calendar, headlines, positioning) and proposes at most one trade, or none. Code checks every level before it is published.
The idea, explained: AI trading signals and how to judge them
An AI desk has no backtest, because its decisions cannot be replayed on old prices. Its record starts with these signals.
A classic Donchian 20/10 trend system on 4-hour gold bars. When the last 4-hour close sits within one ATR of the 20-bar high (or low), a buy stop (sell stop) waits at that level for the next 4 hours. The stop loss is the 10-bar low (high), at least half an ATR away. Targets are 1R, 2R and 3R; whatever is left closes when a 4-hour close breaks the previous 10 bars' opposite extreme.
The idea, explained: Breakout trading
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | +0.16R | +0.19R | +0.17R | +0.23R |
47 trades a yearprofit factor 1.67
Caveat. Almost all of the edge came from buy trades during a strong gold uptrend; on daily bars from 2010 to 2022 the same rule made money in only 7 of 13 years.
The most common strategy in our Pine library (moving-average crossovers: 251 scripts, 6,083 GitHub stars; for example the Simple EMA20/50 Cross strategy). On 4-hour bars, buy at market when the 20 EMA crosses above the 50 EMA and sell when it crosses below. Stop 1.5 ATR from the entry, targets 1R, 2R and 3R; whatever is left closes when the averages cross back.
The idea, explained: Moving averages and crossovers
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | −0.14R | +0.12R | +0.02R | −0.09R |
31 trades a yearprofit factor 0.96
Caveat. Average per trade across the 11 instruments tested (7 FX majors, gold, silver, Bitcoin, Ether), costs included. Positive on 4 of 11 instruments (best silver +0.15R, worst USD/CAD -0.11R). Not tested on WTI or the Nasdaq 100.
RSI strategies are the third most starred family in our Pine library (179 scripts, 4,691 stars); this is Larry Connors' RSI(2). On 4-hour bars, buy at market when RSI 2 drops under 10 while price is above its 200 SMA, and sell when RSI 2 rises above 90 below the 200 SMA. Stop 1.5 ATR, targets 1R, 2R and 3R; the rest closes when price crosses back over the 5 SMA, or after 10 bars.
The idea, explained: The RSI indicator
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | −0.02R | −0.07R | −0.01R | 0.00R |
88 trades a yearprofit factor 0.96
Caveat. Average per trade across the 11 instruments tested (7 FX majors, gold, silver, Bitcoin, Ether), costs included. Close to break-even before costs but slightly negative after them; positive on 1 of 11 instruments. Not tested on WTI or the Nasdaq 100.
The widely shared "MACD + 200 EMA" strategy (MACD strategies: 117 scripts, 3,803 stars in our Pine library). On 4-hour bars, buy at market when the MACD line (12, 26) crosses above its signal line (9) while both are below zero and price is above the 200 EMA; sell on the mirror image. Stop 1.5 ATR, targets 1R, 2R and 3R; the rest closes on the opposite MACD cross, or after 48 bars.
The idea, explained: The MACD indicator
| Year | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|
| R / trade | −0.12R | −0.10R | +0.06R | 0.00R |
25 trades a yearprofit factor 0.94
Caveat. Average per trade across the 11 instruments tested (7 FX majors, gold, silver, Bitcoin, Ether), costs included. Positive on 3 of 11 instruments (best USD/JPY +0.11R, worst gold -0.25R). Not tested on WTI or the Nasdaq 100.
Once a day before London, ForexR AI reads 4-hour and daily bars with the facts pack (levels, trend, positioning, the week's calendar) and proposes at most one position trade per instrument, held for days to two weeks, or none. Code checks every level against the 4-hour ATR before it is published.
The idea, explained: Swing trading
An AI desk has no backtest, because its decisions cannot be replayed on old prices. Its record starts with these signals.
The classic news straddle. About four minutes before a high-impact data release for the instrument's currencies (US jobs, inflation, rate decisions, PMIs and the like, from our economic calendar; speeches and press conferences are left out), the range of the last 30 minutes is bracketed: a buy stop half an ATR (5-minute) above it and a sell stop the same below it, at least two ATRs apart. The first to trigger cancels the other, and each one's stop is the other's entry. Orders wait 30 minutes after the release; trades close within three hours. Targets 1R, 2R and 3R.
The idea, explained: Trading the news
| Quarter | 2025 Q4 | 2026 Q1 | 2026 Q2 | 2026 Q3 | 2026 Q4 |
|---|---|---|---|---|---|
| R / trade | +0.03R | −0.04R | +0.09R | +0.13R | −0.15R |
537 trades a yearprofit factor 1.08
Caveat. Slightly positive over the last year, +0.04R per trade over 537 trades (gold +0.13R, Nasdaq 100 +0.07R, USD/JPY +0.06R, EUR/USD -0.06R), but one year is too short to tell that from luck. Fills are at the stop price; in a real release spike the fill can be worse (slippage), so the true result is likely lower.
Trade the first reaction. When the 15-minute candle that contains a high-impact data release closes, buy at market if it closed up and sell if it closed down, with the stop at the other end of that candle (at least half an ATR away). Targets 1R, 2R and 3R; trades close after six hours at most.
The idea, explained: Trading the news
| Quarter | 2025 Q4 | 2026 Q1 | 2026 Q2 | 2026 Q3 | 2026 Q4 |
|---|---|---|---|---|---|
| R / trade | −0.29R | −0.04R | 0.00R | −0.10R | −0.15R |
599 trades a yearprofit factor 0.84
Caveat. -0.08R per trade over 599 trades in the last year (gold +0.02R, EUR/USD -0.06R, USD/JPY -0.07R, Nasdaq 100 -0.23R). Our 2023-2026 study found the same rule positive in every year on gold for US jobs and Fed days only (14-18 trades a year), but negative when all release types were included.
Twenty to sixty minutes before each high-impact release, ForexR AI reads the release (forecast and previous figure), 15-minute and 1-hour bars and the facts pack, and plans at most one trade per instrument for after the numbers: a pending order that waits for the release, or no trade. Code checks every level against the 1-hour ATR before it is published.
The idea, explained: Trading the news
An AI desk has no backtest, because its decisions cannot be replayed on old prices. Its record starts with these signals.
The AI desks' passes and rejected ideas are part of the record too: the scalping, intraday, swing and news desks. This is every look they have taken at the market, including the times they chose not to trade or a proposal failed the code checks.
A signal from this desk is still open.
Already 2 open AI signals.
A signal from this desk is still open.
A signal from this desk is still open.
Already 2 open AI signals.
Already 2 open AI signals.
Already 2 open AI signals.
Already 2 open AI signals.
A pending sell order is already open at the 2250 resistance level, which aligns with the previous day's high; we will let this existing setup play out rather than layering additional short exposure in the current mid-range.
There is already a pending sell order at 76250, which perfectly captures a potential bearish retest of the daily EMA200, making a new signal redundant.
H1 trend is mixed and price is mid-range bouncing off intraday lows with no decisive rejection yet at the pivot resistance (27162.1) or clean support test, so there's no clean level-plus-trend alignment for a scalp right now.
The 1-hour trend is up, but we cannot buy due to an open AI swing sell signal on this instrument, and there is no clear intraday level to justify fading the recent short-term momentum.
A signal from this desk is still open.
Already 2 open AI signals.
The current spread of 4.6 pips is over half of the 5-minute ATR (7.7 pips), making trading costs too high for a scalping strategy.
A signal from this desk is still open.
The signal goes up with its entry zone, stop, three targets and the reason, before price gets there. A SHA-256 fingerprint of those levels is saved with it.
A tracker follows the price every minute. Buys fill at the ask, sells at the bid, and if one bar touches both the stop and a target the stop counts first.
One third of the trade closes at each target and the stop moves to the entry after target 1. A losing trade gets an AI review that checks what happened against the reason it was taken.
Two kinds of signal run side by side: data rules computed in code from the price chart, and AI desks that propose a trade from the same facts the analysis desk reads. Both are scored and shown separately above. Every signal also has a type of trading: scalping (minutes to a few hours), intraday, swing (days to weeks) or news trading (around high-impact releases), so you can follow only the style you trade.
Every signal also carries a risk tag, set when it is published and never changed. It is High Risk when its rule lost money in testing (or its live record is losing), when it is a scalping or news trade, when a high-impact release is due while the order waits or early in the trade, or when the spread is more than a tenth of the stop. It is Cautious when none of these applies. Cautious is not risk-free: any trade can lose.
New to signals? Start with how to trade forex signals, then how to copy a signal into MT5, cTrader or TradingView and how to read a signal track record.