Do simple signals work on gold? Big candles, streaks and the 200 SMA, tested
Plenty of chart rules sound convincing. Here are three of them, tested on the same ten years of gold prices the game uses, with the number of cases next to every result so you can judge how much to trust it.
Start with the baseline
Before testing any signal, you need to know what "no signal" looks like. Over the decade, 50.2% of 15-minute candles, 50.7% of 1-hour candles and 51.3% of 4-hour candles closed higher than they opened. That small tilt comes from gold's long rise. A signal is only interesting if it clearly beats these numbers.
| Situation | Chart | Cases | What followed | Baseline |
|---|---|---|---|---|
| Any candle closes higher than it opened | 15m | 231,331 | 50.2% | 50.0% |
| Any candle closes higher than it opened | 1H | 57,890 | 50.7% | 50.0% |
| Any candle closes higher than it opened | 4H | 15,317 | 51.3% | 50.0% |
| After a big up candle: next candle is up | 1H | 701 | 50.4% | 50.7% |
| After a big up candle: higher 10 candles later | 1H | 701 | 52.9% | 52.2% |
| After a big down candle: next candle is down | 1H | 690 | 48.6% | 49.0% |
| After a big down candle: lower 10 candles later | 1H | 690 | 42.8% | 47.8% |
| After a big up candle: higher 10 candles later | 4H | 148 | 58.8% | 53.8% |
| After a big down candle: lower 10 candles later | 4H | 156 | 51.3% | 46.2% |
| After 3 up candles in a row: next is up | 15m | 26,983 | 47.2% | 50.2% |
| After 3 up candles in a row: next is up | 1H | 6,990 | 48.0% | 50.7% |
| After 3 down candles in a row: next is down | 1H | 6,328 | 46.6% | 49.0% |
| After 5 up candles in a row: next is up | 1H | 1,598 | 46.8% | 50.7% |
| Close above SMA 200: higher 10 candles later | 1H | 32,824 | 52.3% | 52.2% |
| Close below SMA 200: higher 10 candles later | 1H | 25,066 | 52.1% | 52.2% |
| Close above SMA 200: higher 10 candles later | 4H | 8,715 | 53.6% | 53.8% |
| Close below SMA 200: higher 10 candles later | 4H | 6,602 | 54.1% | 53.8% |
| A big candle has a body of at least 2 × ATR(14). "What followed" is the share of cases where the move went the stated way; "Baseline" is the same share over all candles on that chart. Candles with no change count as not continuing. | ||||
Big candles: momentum or exhaustion?
We called a candle "big" when its body was at least twice the ATR(14) of the candles before it.
- After a big up candle on the 1-hour chart, the next candle was up only 50.4% of the time (701 cases), the same as any other candle. Ten candles later price was higher 52.9% of the time, against a baseline of 52.2%. There is no real edge either way.
- After a big down candle on the 1-hour chart, price was lower ten candles later only 42.8% of the time (690 cases), against a baseline of about 47.8%. In other words, sharp hourly drops were bought back more often than they continued. With 690 cases, a gap this size is unlikely to be pure chance. Remember, though, that the decade was a strong uptrend.
- On the 4-hour chart, big up candles were followed by higher prices ten candles later 58.8% of the time. There are only 148 cases, so the uncertainty is roughly ±8 percentage points. Treat it as a hint, not a rule.
Streaks: do three in a row mean a fourth?
- After three up candles in a row on the 1-hour chart, the next candle was up only 48.0% of the time (6,990 cases). After three down candles, the next was down 46.6% of the time.
- On the 15-minute chart the figures are 47.2% and 45.9%.
- Longer streaks on the 1-hour chart behaved the same way. After five up candles, the next was up 46.8% of the time.
On the shorter charts, a streak was slightly more likely to pause than to continue, by 2 to 4 points. That is not enough to trade on alone, especially after spreads. It does contradict the instinct that "it has gone up four times, so it will go up again."
The 200 SMA: a trend filter that does not time entries
Price closed above its 200-period simple moving average on 56.7% of 1-hour candles. Being above or below it made almost no difference to what came next:
- 1-hour: higher ten candles later in 52.3% of cases when above the SMA 200, and 52.1% when below.
- 4-hour: 53.6% above against 54.1% below.
The 200 SMA describes the trend well. On its own, it did not predict the next ten candles.
What to take from this
- Single conditions land within a few points of 50%. Edges, where they exist, come from combining context, such as the session, the higher-timeframe trend and nearby levels, and from passing when nothing lines up.
- Always check the number of cases. With 150 cases, ±8 points is noise. With 7,000, even 2 points is real but small.
- Test your own ideas. In the game, mark your reason with a drawing, make the call, and review the last 30 results to see whether the idea holds up.
- These are historical frequencies from one market and one period. They describe the past and do not predict the future.