Gold and silver: monthly MACD and RSI above 55

Idea in brief

The author of the Financial Wisdom channel shows a long-term trend system for gold and silver on the monthly chart. MACD answers whether a trend has started. RSI answers whether the trend is strong enough. A buy happens only when both conditions are met in the same month.

The exit is not an opposite cross. When MACD crosses below its signal line, the author marks the low of that month. The position is closed only when price moves below that low.

The system is out of the market most of the time. According to the author, it was in gold about 28% of the 50 years and in silver about 23%, and the end result is close to simply holding the metal.

Why it might work

The author's explanation is short: the system buys strength. A MACD cross on monthly bars marks the start of a new move. RSI above 55 filters out crosses with no strength behind them. With the 1982 example, the author shows that such crosses without strength are skipped, even if RSI rises above 55 a couple of months later.

The exit through the month's low adds confirmation. A downward cross by itself does not close the position, only a break of the level does. This protects against exiting on a temporary weakening inside a long trend.

The author's argument for the system is about time in the market. The result is close to buy and hold, while capital is free two thirds of the time and can work in other assets.

A caveat about the figures. The 28% a year is calculated only over 14 years in the market, while the 7.3% for buy and hold covers all 50 years. Spreading $313,000 over the full period gives about 7.1% a year (our recalculation from the author's figures), slightly less than buy and hold. Silver shows the same picture. $174,800-178,400 over 11.5 years in the market (23% of 50 years) gives about 28% a year rather than 30%, and about 5.9% a year over the full period versus 6% for buy and hold (our recalculation from the author's figures). The system's real advantage is the smaller drawdown and the free capital, if that capital earns something in the meantime.

Rules

Indicators

// monthly bars, spot gold or silver in dollars per ounce
MACDLine = EMA(Close, 12) - EMA(Close, 26)     // author: MACD 12 26 9, the indicator label on the chart in the video
Signal   = EMA(MACDLine, 9)
RSIm     = RSI(Close, 14)                      // Finetiq: the RSI period is not named and cannot be read on the chart, we use the standard 14

Entry

// checked at the month's close
BullCross = MACDLine[1] <= Signal[1] AND MACDLine > Signal

IF MarketPosition = flat AND BullCross AND RSIm > 55     // author: both conditions in the same month
    BUY AT NEXT BAR OPEN                                 // author: at the open of the next month

// author: entry "at a nominal price above the previous month's close"
// Finetiq: the main variant is at market on the open; for testing, BUY STOP at Close + 1 tick for the whole next month
// author: if RSI reached 55 later than the month of the cross, the trade is not taken and we wait for a new cross

Exit

BearCross = MACDLine[1] >= Signal[1] AND MACDLine < Signal

IF MarketPosition = long AND BearCross
    ExitLevel = Low                           // author: the low of the month in which MACD crossed down

IF ExitLevel defined
    SELL STOP at ExitLevel                    // author: exit when price moves below that low
// Finetiq: the order is active from the open of the next month until filled
// Finetiq: if MACD crossed up again and then down again before the break, ExitLevel = low of the new cross month
// Finetiq: if MACD crossed up and there was no break, the level stays in force (the author does not cover this case)
// the author has no protective stop at entry

Position size (author)

Allocation = 10–20% of the portfolio          // author: share allocated to the strategy
Leverage   = none                             // author: the risk of ruin is high even at 2x leverage

Parameters

What to test

1. A fair comparison with buy and hold. Calculate the annual return of both curves over all 50 years. Add the return on cash while out of the market, for example from Treasury bills. Without that return, the system does not beat buy and hold by the author's own figures.
2. RSI threshold. Compare 50, 55, 60 and a variant without RSI. With seven trades, one added or missed entry changes the result many times over. If 55 is noticeably better than its neighbors, it was chosen from history.
3. Period neighborhood. RSI 9, 14 and 21, MACD 10/22/7, 12/26/9 and 14/30/11. The result should change smoothly, not jump.
4. More markets. Seven trades give no statistics. Run the same rules unchanged on platinum, copper, crude oil and indices on monthly bars. Look at the median across markets, not the best one.
5. Weekly timeframe. The same rules on weekly bars. There will be many times more trades, and you can judge whether there is an effect beyond a few long trends.
6. Exit. Compare the stop at the low of the cross month with an exit right at the downward cross and with an exit on a monthly close below the level. For each variant, calculate how much profit is given back from the trade's peak to the exit.

Platform notes

TradingView (Pine Script)

• The author built the chart on TradingView, symbol "CFDs on Gold (US$/OZ)" from TVC, monthly bars. The history of this series on the chart in the video starts before 1976, so the test period can be reproduced.
• ta.macd(close, 12, 26, 9) and ta.rsi(close, 14) match the indicators visible on the author's chart.
• Exit stop: strategy.exit(..., stop = exitLevel) stays active until filled. If the month opens below the level, the fill will be at the open price.
• The current monthly bar is not closed until the end of the month. Set alerts on bar close, otherwise a cross that disappears by the end of the month will have time to trigger a signal.

MultiCharts and TradeStation (EasyLanguage)

• Futures history for gold and silver is stitched together from contracts. MACD and RSI are not distorted by a subtraction-adjusted series, since they are based on price differences. The level of the month's low and the percentage comparison with buy and hold are distorted.
• MACD(Close, 12, 26), signal line XAverage(MACD(Close, 12, 26), 9), RSI(Close, 14) with Wilder smoothing.
• Sell next bar at ExitLevel stop lives for one bar, which on a monthly chart is a month. The order has to be sent on every bar while the position is open.
• A futures daily bar in TradeStation closes at settlement. A month built from such bars also closes at the settlement of the last day, not at the last trade.

MetaTrader 5 (MQL5)

• In iMACD, the signal line is a simple average, not an EMA. Crosses will shift relative to the author's chart. Take the main line (buffer 0) and calculate the 9 EMA of it yourself.
• Brokers' monthly bar history for XAUUSD and XAGUSD is usually much shorter than 50 years. An EA on such history will not see most of the author's trades, so testing has to be done on external data.
• The position is held for years. CFDs on metals charge a swap fee for every night, which adds up to a noticeable amount over four years and is absent from the author's calculation.
• Calculate the signal on the closed monthly bar (index 1) and check for a new bar.

Where the idea can break

• Seven trades in 50 years. Such a sample cannot tell a rule from luck: one extra or missed entry changes the result many times over.
• The rules are shown on the same two markets and the same period on which the result was calculated. The video has no out-of-sample test.
• The 28% and 30% a year are calculated only on time in the market. Over the full period, the gold strategy gave about 7.1% a year versus 7.3% for buy and hold, and the silver strategy about 5.9% versus 6% (our recalculation). The claimed 30% for silver does not hold even on time in the market: the author's figures give about 28%. The video first presents this as a four- to fivefold advantage, then as almost threefold.
• The silver figures differ within the video: $174,800 at the start and $178,400 at the end.
• There is no protective stop, and drawdowns are large: 48% on gold and 78% on silver according to the author. The exit through the month's low can give back a significant part of the move.
• The video leads to a paid scanner and community. The equity curves are shown as a picture, with no list of trades and dates.
• The return on cash while out of the market is not included in the calculation. The author mentions "other assets" but does not show what happened with them.