The Lab

Where the idea comes from

• Buying in steps as the price falls is a form of dollar-cost averaging. Its history and maths are on our theory page "Steady buying or all at once: the theory of dollar-cost averaging".
• QFL is a way of timing those buys. It waits for the price to break below a recent low, a "base". The method comes from the trader known as Quickfingers Luc, and is named after that nickname. We couldn't confirm where or when it was first set out.
• The two scripts tested here are the accumulator, a buying and selling ladder that waits for an outside signal, and the indicator that supplies QFL's signals to it. We rebuilt both line by line from their code as published on TradingView.
• Judging a signal against random signals is a Monte Carlo test: drawing many random cases to see what chance alone gives. Nicholas Metropolis and Stanisław Ulam named and described the method in 1949.

How we tested

• Spot, starting with 10,000, with a 0.1% fee each way.
• Three windows, never pooled:
  • 14 coins, April 2024 to September 2026, mostly falling: BTC, ETH, SOL, ZEC, NEAR, XRP, DOGE, GTC, STRK, RSR, PEPE, WIF, BONK, SHIB, the same coins as our grid study.
  • BTC, ETH and SOL, October 2020 to September 2026, with up years.
  • BTC alone, October 2017 to September 2026.
• Each window is split in two halves, each run on its own. Settings are picked on the first half and judged on the second, which they never saw.
• Against chance: each ladder also got the same number of buy and sell signals in each month at random times (200 draws), or evenly spaced, so we can tell the signal's part from the ladder's.
• Prices are Binance spot hourly candles.

The maths, step by step

1. The signal's lines. The latest base is 100 and the depth is 3.5%, the indicator's default. A buy fires when a close is under 100 × (1 − 3.5%) = 96.50. If the latest peak is 120, a sell fires above 120 × (1 + 3.5%) = 124.20.

2. Five buys down the ladder. Buys of 2,000 at 100, 97, 94, 91 and 88, each at least 3% under the last.

• Each 2,000 buys more coins at a lower price. The average paid is 93.81, below the plain average of the five prices, 94.00.
• At 88, all 10,000 is in, and it's worth 9,371 after fees: down 6.3%.

3. Selling in quarters. Each sell takes a quarter of what's held. After five sells, 0.75 × 0.75 × 0.75 × 0.75 × 0.75 = 23.7% of the coins are left. On the default setting, that part is never sold.

4. What luck gives. 216 runs, each judged at a 5% bar against its random signals: 216 × 5% = 10.8. About 11 runs would pass by luck alone.

1. Does QFL's timing help?

We tested 18 settings: 1-hour, 4-hour and daily charts; signal depths of 2, 3.5 and 5%; and "stop after five sells" or "start selling again".

• QFL's timing added nothing. In each window, the setting that looked best in the first half was level with random signals in the second. Across every setting and coin in the second halves, 13 of 216 runs beat random firmly. Chance alone gives about 11 (step 4).
• Evenly spaced signals did better more often than QFL's: in 62% of runs in the falling window, 65% in the up years, and 14 of 18 on BTC alone. QFL waits for sharp drops, and the drops kept going.
• The ladder is what matters.
  • In the falling window it beat simply holding the coin in 196 of 252 runs, because it keeps money out of the market.
  • In the up years, holding beat it in 52 of 54 runs. It sells quarters on the way up and is left holding little (step 3).
• What it feels like. In the falling window the typical run spent two thirds of its time holding at a loss. The longest stretch under water was 456 days, and 582 days on BTC since 2017.

2. The accumulator's ladder against ours

With the signal taken out, both ladders got the same random buy signals, so only the ladder differed. Ours is the one on our page "Buying the dip in steps: three tests of our DCA ladder".

• The accumulator buys on a signal, sells a quarter on a sell signal when in profit, and by default stops selling after the fifth sell.
• Our ladder places a first order on a signal, rests further buy orders below it, and sells everything at the average price plus one step.
• The settings tried: 3, 5 or 8 orders; steps of 2, 3 or 5%; equal sizes, or each one 1.5 times the last.

What we found:

• In falling markets, our ladder did much better.
  • On the 14 coins it beat the accumulator on 11, by a median of 45 points.
  • Its first-half pick (8 orders, 5% apart, growing sizes) made +22% in the second half, while the accumulator's pick lost 33%.
  • It spent less of its time at a loss (50% against 73%).
  • It sells the whole position on a bounce and steps aside. The accumulator sells quarters and keeps holding the rest as the price falls.
• In the up years they came out about level. On BTC since 2017, the accumulator won by 23 points: it keeps a holding through a rise, while our ladder keeps cashing out at +5%.
• The cost of our ladder is money tied up. It used all 10,000 at some point in almost every run. In the up years it was holding at a loss 77–81% of the time, once for 666 days. The accumulator used about a third of the money and was almost never under water.
• "Never sell after the fifth" cost nothing in falling markets (it never got there) and paid in rising ones. On BTC from 2017 to 2022, restarting the sells instead made 35 to 374 points less.

Credits

• BTC External Signal Accumulator (v11), published on TradingView, by the author credited at the top of this page. The script and its rules are theirs; we rebuilt and tested it.
• QFL single TF v1.3, published on TradingView by the same author: their version of the QFL method.
• The QFL method: the trader known as Quickfingers Luc. Where and when it was first set out is uncertain.
• Nicholas Metropolis and Stanisław Ulam, "The Monte Carlo Method", Journal of the American Statistical Association, 1949: drawing random cases to see what chance gives.
• Buying in steps: see our theory page "Steady buying or all at once: the theory of dollar-cost averaging" for its history and credits.
• The tests, the random controls and our own ladder are ours.
• Data: Binance's public spot hourly archive.
• PerpGremlin and Claude, The QFL signal in a TradingView accumulator: does the signal help?, TerminalGremlin, 2026: the tests, the code and the words.