Statistical Parameters of 21 Forex Indicators: Whipsaw and Lag Ranked the Same on 4 Charts at 0.94 or Better, and None of Them Predicted Profit

Most indicator reviews describe behaviour in words: “fast but noisy”, “smooth but lagging”, “rarely gives false signals”. Those words are measurable. If an indicator really is noisy, you can count how often it reverses itself. If it lags, you can measure how far price had already gone when it spoke.

I took the 21 rules from the free MQL5 indicator test and the earlier single-indicator tests, measured each one’s statistical parameters, and then asked the only question that matters to a trader: does any of these numbers tell you in advance which indicator will make money?

What did I measure, and on what data?

Data: EURUSD M5 from June 3 to August 26, 2026, 16,926 bars; gold M5 from June 16 to August 26, 13,736 bars; M15 built from the same bars. ATR scaling lets EURUSD and gold sit in one table. The profit column comes from the same 21 rules traded with one exit for all: stop 1.5 ATR, target 3 ATR, 48 bars at most, 1 pip of cost on EURUSD and $0.30 on gold. As a yardstick for the 12-bar move I used random bars with a random direction, 200 runs of the same size on each chart.

How often does each indicator reverse itself?

Heikin Ashi colour flips came back the other way within 6 bars 81.7% to 86.5% of the time, on every chart. The RSI 14 exit from 30/70 almost never did: 0% to 0.8%. That is not a market effect. It is how the rule is built: a Heikin Ashi candle changes colour on a one-bar wobble, while an RSI exit needs price to travel from one extreme back across a level. The colour flip in detail is in the Heikin Ashi strategy test.

The breakouts sat at the bottom with the RSI exit: Donchian 20 at 2.2% to 3.1%, Keltner at 0.5% to 4.2%. A breakout rarely flips back within 6 bars because the opposite breakout needs price to cross the whole channel.

How late does each indicator speak?

The Donchian 20 breakout fired after price had already moved 4.69 to 6.23 ATR off its 24-bar extreme. Bollinger re-entry fired after 0.97 to 1.06 ATR. Breakouts are late by design, overbought and oversold exits are early by design, and every trend and momentum rule sat between 2.1 and 4.9 ATR. For scale, a random bar sits 2.2 to 2.7 ATR from its own 24-bar extreme on these charts, so the overbought/oversold exits fired earlier than chance and the breakouts about twice as late.

Are these numbers the same on every chart?

This is the main finding. Rank the 21 rules by whipsaw on EURUSD M5, then on gold M15, and the two rankings agree at 0.94. By lateness they agree at 0.99. The scale changes, gold M15 gives fewer signals per day than EURUSD M5, but the order of the rules does not. The same held inside one chart: ranked by bars between signals, the first and second half of the period matched at 0.91 to 0.99.

The next two columns are the opposite. Ranked by the 12-bar move, EURUSD M5 and gold M5 came out at −0.66: the rules that did well after the signal on one chart did badly on the other. Even the two halves of one chart only reached −0.05 to 0.47.

What did price do after the signal?

On EURUSD M5 the picture was one-sided. 15 of 21 rules had the whole 90% bootstrap band of the 12-bar move below zero. Twelve of them were below the random band too. Supertrend 10/3 averaged −0.40 ATR, the EMA 9/21 cross −0.35. Nothing was above the random band.

On gold M5, Keltner and Donchian breakouts were the only rules with the whole band above zero, at +0.23 and +0.20 ATR, and both still lost money with the 1.5/3 ATR exit, −$0.84 and −$0.90 a trade. A positive average move 12 bars later does not survive a stop that gets hit first. That gap between “price went my way” and “my trade made money” is the same one found in bar close vs live signal.

Does any parameter predict which indicator makes money?

Not the ones you can see before the trade. Signals per day, bars between signals, whipsaw rate and lateness all changed sign from chart to chart. Lateness had the strongest single link, −0.52 on EURUSD M5, and +0.29 on gold M15. A trader who learned “early indicators win” on EURUSD M5 would pick the wrong rules on gold.

The two parameters with the same sign on all four charts were the best move after the signal and the share of signals on the right side 12 bars later. Both are measured after the signal, so they are part of the result, not a forecast of it.

Which signals per day did each rule give?

Heikin Ashi gave 80.9 signals a day on EURUSD M5 and 21.9 on gold M15. The DI cross with ADX above 25 gave 3.4 and 1.4. Frequency moved with the chart’s bar count, and the order of the rules stayed put.

What should a trader take from this?

• Use the parameters to describe an indicator, not to choose one. Whipsaw and lateness tell you what the chart will look like. They repeated at 0.94 to 1.00 across four charts.
• Do not expect the follow-through to repeat. The 12-bar move ranked at −0.66 between EURUSD M5 and gold M5.
• Low whipsaw is not low risk. Donchian had one of the lowest whipsaw rates and price went against it on EURUSD M5 by 0.31 ATR on average.
• Check the market first. On EURUSD M5 almost every trend and breakout signal was followed by a move against it. That is a range, and no indicator parameter fixes a range. The range filters are tested in Choppiness Index vs ADX.

Rules that repaint would break every number here, because the signal you see later is not the one you had. All 21 were computed on closed bars only; how to check that yourself is in non-repaint indicators for MT5.

Frequently asked questions

What are the statistical parameters of an indicator?

Numbers that describe how it behaves: how many signals it gives, how often it reverses itself, how far price has moved when it signals, and what price does afterwards.

Which indicator had the fewest false signals?

Counting a reversal within 6 bars as false, the RSI 14 exit from 30/70 at 0% to 0.8%, then the breakouts at 0.5% to 5.2%.

Which indicator lagged the most?

The Donchian 20 and Keltner breakouts, 4.7 to 6.2 ATR after the 24-bar extreme. Bollinger re-entry lagged least at about 1 ATR.

Can I pick an indicator by its lag or whipsaw rate?

Not on this data. None of the four had the same link to profit on all four charts.

Why use ATR instead of pips?

So that EURUSD and gold, M5 and M15 can be compared in one table. One ATR is the typical bar range of that chart.