Rule of four: a breakout of the fourth candle after the news

Idea in brief

On major news, the first minutes are messy: price lurches both ways and takes out stops. Tom Hougaard does not trade the first reaction. Tom counts candles from the one in which the news came out: one, two, three, four. After the fourth one closes, a buy goes above its high and a sell below its low. The stop goes at the other end of the fourth candle.

The timeframe depends on the event. On FOMC, whose decision comes out during the US trading session, the candles are 10-minute, and the fourth one closes 40 minutes after the decision. On pre-open news (NFP, CPI, PPI), the candles are 5-minute. The mechanics are the same as in School Run, only the count starts from the news time.

Tom created the rule for FOMC and speaks most highly of it there. On NFP and PPI, Tom's own examples include many failures.

Why it might work

Tom's explanation is empirical. Tom noticed that FOMC trading was not making money, downloaded every FOMC evening that could be found, and saw that by the fourth 10-minute candle the market usually picks a direction. The press conference starts 30 minutes after the decision, and the fourth candle closes 10 minutes after it begins. By that point the first reaction to the decision itself has already passed.

The streams have no verified statistics, only chart examples and claims. Tom's own streams show the weak spots: on PPI the rule "brought disappointment", the FTSE on NFP often comes back after the first move, and in a selection of 18 NFP days the rule worked poorly in many cases. So the explanation should be treated as a hypothesis, and the event type as a separate test parameter.

Rules

Time and candles (author)

// Finetiq: time per the US calendar. The FOMC decision is at 14:00 ET, the press conference at 14:30 ET.
// NFP, CPI and PPI come out at 08:30 ET, before the cash session opens

TF = 10 min  IF the news comes out during the session (FOMC)
TF = 5 min   IF the news comes out before the open (NFP, CPI, PPI)

Bar1 = the TF candle in which the news came out   // FOMC: from the decision, not from the press conference (V111)
Bar4 = the third candle after Bar1

// FOMC: Bar4 = 14:30-14:40 ET, orders after 14:40 ET
// NFP:  Bar4 = 08:45-08:50 ET

Tom's instruments: the Dow, NASDAQ 100, S&P 500, and optionally the DAX and FTSE. European indices have a wider spread on FOMC evenings (V111, V143).

Entry (author)

// after Bar4 closes
BuyLevel  = Bar4.High + Offset
SellLevel = Bar4.Low  - Offset
BUY STOP at BuyLevel
SELL SHORT STOP at SellLevel
// after a fill, Tom deletes the second order (V143)

// offset: differs between streams
// V67: levels are named at the edges of the candle, not a word about an offset
// V7:  NASDAQ about 5 points, Dow about 10 points
// V13: "a small offset";  V102: "a little margin"
// Finetiq: no offset is named for the DAX and FTSE. Starting value 5 points

Stop and trade management (author)

Long:  StopLoss = Bar4.Low        // the "other end" of the candle (V67, V111, V7)
Short: StopLoss = Bar4.High

// a big Bar4 gives a big stop (V111): before the candle closes about 80 points on NASDAQ and 120+ on the Dow,
// by the candle 100 on NASDAQ and 200 on the Dow, Tom set 250
IF the stop is too large THEN reduce size OR skip OR look for a stop on a lower TF   // author
// Finetiq: for the test, size from 1% risk, skip trades with a stop larger than 2 * ATR(14) of the same TF

IF OpenProfit >= 30 points THEN StopLoss = EntryPrice    // V67 "optional", V143 "a habit"

// Tom does not name targets and closes depending on the situation (V102, V143)
EXIT at 15:55 ET                  // Finetiq: for all indices, on futures or CFDs

Momentum filter (author, not always applied)

// V67: take the index with "real" momentum and skip the "piano" (overlapping candles).
//      There is no point shorting the DAX if the Dow or NASDAQ have not broken their levels.
// V143: candles 1-4 almost without bodies, no skew. Tom reduced size and did not trade the DAX and FTSE.
// V102, V111: traded all indices that broke their levels

// Finetiq: formalization for the test
Piano = each of the candles Bar2, Bar3, Bar4 overlaps the previous one by more than 50% of its range
IF Piano THEN skip the index
IF index = DAX OR index = FTSE
   AND neither the Dow nor NASDAQ has yet broken its level in the same direction
    THEN do not enter

Event filters (author)

IF event = PPI THEN questionable, test separately          // V13: conclusion from the last five PPI releases
IF event = NFP AND index = FTSE THEN skip                  // V44: the FTSE comes back after the reaction
// V44: on NFP Tom does not place the bracket in advance, being unsure about execution

Variants (author)

// A. Flip the switch: after a stop, enter in the opposite direction
IF the long is stopped out at Bar4.Low THEN SELL SHORT   // V111: "even if the reversal comes later, I still take the short"
// V143: the short was not taken right away, but after a close below the lows of the last candles
// Finetiq: no more than one reversal per event

// B. Scalp from the third candle (V111): Tom switched the FOMC chart to 5 minutes
BUY STOP at Bar3_5m.High           // according to Tom, "with much less risk"
// Bar3_5m = the third 5-minute candle, not the 10-minute Bar3 of the main rule
// Finetiq: counted from the decision candle, as in the main rule

// C. Bracket on candles 2 and 3 (V44, Tom: "just came up with it")
BUY STOP at max(Bar2.High, Bar3.High)
SELL SHORT STOP at min(Bar2.Low, Bar3.Low)

// D. Line in the sand (V44): the close of Bar1 above or below sets the side

// E. Scaling in (V143): half at market on the breakout,
//    the second half if price pulls back into the range of the previous candle

Parameters

What to test

1. By event type. FOMC, NFP, CPI, PPI separately, and each index separately. Tom has doubts about PPI and about the FTSE on NFP. If the rule holds only on FOMC, that is about eight events a year.
2. The fourth candle versus its neighbors. Bar3, Bar4, Bar5 and the bracket on candles 2-3. For FOMC, compare 10-minute candles with 5-minute ones, and for NFP the other way around.
3. Offset and slippage. Offset of 0, 5, 10, 20 points. Separately, add stop order slippage of 5-20 points: Tom avoids placing orders in advance on NFP.
4. Momentum filter. All signals versus signals without the "piano" and versus the DAX only with confirmation from the Dow or NASDAQ.
5. Flip the switch. With and without the reversal. Separately, a reversal right at the stop and a reversal after a close beyond the level, as on V143.
6. Breakeven. +30 points versus 0.5 and 1 range of the fourth candle and versus no breakeven. 30 points on the Dow at about 41,000 (V111) and on the FTSE at about 8,700 (V67) are very different percentages of price.
7. Sample size. On an FOMC evening the indices are highly correlated, so four indices on one night are almost a single bet. Calculate statistics per event, not per trade.

Platform notes

TradingView (Pine Script)

• There is no news calendar. FOMC, NFP, CPI and PPI dates and times are set as an array of timestamps. FOMC dates are irregular, and NFP also does not always come out on the first Friday.
• Intraday history is limited by the subscription plan. A 10-minute chart may hold only a few years, that is, a couple of dozen FOMC meetings.
• News at 08:30 ET is visible only on futures or CFDs. The Dow and NASDAQ cash indices have no candles at that time.
• strategy.entry(..., stop = x) fills at the level in a test. In a spike, the real price jumps further. Set slippage and compare with the result on minute data.

MultiCharts and TradeStation (EasyLanguage)

• Time is the bar's close time. The fourth 10-minute FOMC candle on a chart in ET has Time = 1440, and the fourth 5-minute NFP candle has Time = 850.
• The symbol's session must include the morning hours. On the main session from 09:30, the NFP candle simply does not exist.
• Store event dates in an array and compare them with Date. Buy next bar at X stop lives for one bar, so the order is sent on every bar until it fills or is canceled.
• Reversal is built into EasyLanguage: SellShort next bar at Bar4Low stop with an open long closes it and opens a short in one order. In the variant without a reversal, this order must not be sent after entering the long, and the stop has to be placed via Sell.

MetaTrader 5 (MQL5)

• MQL5 has economic calendar functions, but for a historical test it is more reliable to hardcode the dates in an array.
• Server time and daylight saving time. The March 19, 2025 FOMC (V67) fell in the weeks when the US had already switched to daylight saving time but Europe had not. A fixed offset between ET and the server is off by an hour on these dates, and Bar1 will be the wrong candle.
• The M10 timeframe is standard in MT5, so 10-minute candles do not need to be built manually.
• CFD spreads widen on news, so test in "Every tick based on real ticks" mode. On a netting account, a reversal is done with one order of double volume, while on a hedging account you need to close the position and open a new one.

Where the idea can break

• There are few events: about eight FOMC meetings a year. The claims of a "practically zero failure rate" and a "100% record" are not backed by a list of trades. In the same V111 stream, the NASDAQ position was down 72 points by the end of the stream, and on V143 the NASDAQ trade closed at a loss.
• Tom's own review of NFP and PPI argues against the rule more than for it. It may work only on FOMC.
• Execution on news is worse than in a bar-based test: slippage on stop orders, wide CFD spreads, price jumping over the level in a spike.
• The stops are large. Skipping big candles or reducing size shrinks an already small sample even further.
• Tom picks the momentum filter, the reversal timing and the exit depending on the situation. Our mechanical version may differ noticeably from what Tom does on air.
• The 30-point breakeven is set the same for all indices, even though a Dow point and an FTSE point are different fractions of price.