Price above the average: on, below: off

Idea in brief

The simplest trend system: one moving average and one comparison. Price closed above the average, the position is on. Below it, the position is off. What "off" means depends on the version: cash for Masonson and Drogen, a short for Aspell.

Three sources give three settings. Leslie Masonson, in a book about QQQ (retold by the Financial Wisdom channel), holds the Nasdaq-100 ETF above its 225-day average and moves to cash below it. The book's base version uses the 200-day. Leigh Drogen, CIO of the crypto fund Starkiller Capital, calls bitcoin above its 50-day average the simplest way to capture crypto market beta. Dave Aspell of Mt Lucas runs the MLM Index: 22 futures, an annual average, long above it, short below.

All three share the same goal: to sit out most of the deep declines at the cost of late entries and exits.

Why it might work

The book summary does not explain the mechanism, but the logic shows in the numbers. The buy and hold drawdown of QQQ over the test period is 83%, the system with the average has 28.6%. Our reading: the average lags and does not catch the top, but it takes you out of the market before the bulk of a prolonged decline. If the return missed on late entries is smaller than the loss avoided in declines, the system wins.

Drogen explains the effect through human behavior. In Drogen's view, momentum is the only persistent alpha across all markets, because human behavior does not change, while other effects get arbitraged away over time. In crypto it is stronger: most tokens have almost no intrinsic value, and price is driven by the flow of buyers. Drogen notes that the fund trades something more complex, and the average is the baseline.

Aspell sees trend as a risk premium similar to other betas. The MLM Index appeared in 1988 as a benchmark: an investor asked futures managers what their results should be compared against, and they proposed an annual average, long and short. Aspell prefers a slow trend because it is more useful in a real crisis, and deliberately does not optimize the length. According to Aspell, breakouts, moving average crossovers and exponential schemes catch the same large moves.

The result figures come from the book summary and have not been verified by us. Drogen and Aspell did not give return figures for their rules.

Rules

Core system: QQQ and the 225-day average (author, execution Finetiq)

// daily QQQ bars
MA = SMA(Close, 225)          // Finetiq: the video does not name the average type, we use a simple one

IF position = FLAT AND Close > MA
    BUY AT NEXT BAR OPEN      // author: price crossed above the average
                              // Finetiq: signal at the close, execution at the next open
IF position = LONG AND Close < MA
    EXIT AT NEXT BAR OPEN     // author: below the average, to cash

// the book's base version: the same logic with SMA(Close, 200)
// the summary does not mention a filter for false crossovers

Variant B. Bitcoin and the 50-day average (Leigh Drogen)

// daily BTC bars, trading 7 days a week: 50 bars = 50 calendar days
MA = SMA(Close, 50)           // Finetiq: the author says "50-day average" without a type

IF position = FLAT AND Close > MA
    BUY AT NEXT BAR OPEN      // author: above the average, long
IF position = LONG AND Close < MA
    EXIT AT NEXT BAR OPEN     // author: below the average, out of the market, no short
// Finetiq: the author did not name the execution timing

Variant C. MLM Index: 22 futures, long and short (Dave Aspell)

// each market separately: 6 currencies, 5 bonds (10-year), 11 commodities (author)
MA = SMA(Close, 252)          // author: annual average. Finetiq: 252 trading days

// Finetiq: check once a month, on the last trading day, switch at the next open
IF Close > MA THEN target = LONG     // author
IF Close < MA THEN target = SHORT    // author: no flat

// size: the author only mentions rebalancing and names neither weights nor frequency
// on the rebalancing date (Finetiq):
Weight(i) = (1 / 22) * TargetVol / Vol(i)
// size is not recalculated between rebalances.
// author: position volatility is not targeted, positions are allowed to run

Parameters

What to test

1. Length neighborhood. QQQ: 150, 175, 200, 225, 250 days. The book gave figures for 200 and 225 over almost the same period. If 225 is noticeably better than its neighbors, the length is overfitted. Aspell allows 10 months instead of a year: test both values on your futures.
2. SMA versus EMA, close versus open. The summary names neither the average type nor the execution timing. Four combinations on QQQ will show how much the book's figures depend on the implementation.
3. Whipsaws around the average. Count crossovers per year and the result of trades shorter than 10 days. Compare with a filter: entry only on a close 1–2% above the average or after two consecutive closes.
4. Cash with yield. Cash with no interest versus cash in short-term Treasury bills. The system is out of the market about half the time, and the cash rate noticeably changes the total.
5. Cash versus short. Aspell's rule (short below the average) on QQQ and BTC, Masonson's rule (cash) on the MLM futures. On stock indices a short goes against the long-term drift.
6. Periods separately. QQQ: 2000–2012 and 2013–2025. If most of the benefit comes from the 2000–2002 crash, the system depends on a single episode. Bitcoin: each market cycle separately.
7. Leverage. The same average on QQQ and TQQQ, with the signal taken from QQQ and from TQQQ itself. According to the video, on TQQQ the rule barely reduces risk.

Platform notes

TradingView (Pine Script)

• ta.sma(close, 225) and ta.ema(close, 225) give different crossover dates. For the EMA, discard the warm-up: 3–4 average lengths, that is about 900 bars of history.
• By default a strategy fills orders at the open of the next bar. process_orders_on_close = true moves the fill to the close of the signal bar. The book's figures could have been calculated either way.
• A comparison with buy and hold is fair only on dividend-adjusted prices. Check whether dividend adjustment is enabled on the QQQ chart.
• Variant C: a strategy trades one symbol, so a weighted basket of 22 futures cannot be tested in Pine. A single market with an annual average carries over without problems.

MultiCharts and TradeStation (EasyLanguage)

• Average(Close, 225) is a simple average, XAverage(Close, 225) an exponential one.
• Buy next bar at market fills at the open of the next bar. For a fill at the close of the signal bar: Buy this bar on close. Choose one option and do not change it between testing and trading.
• Variant C on continuous futures back-adjusted by subtraction: the price can go negative. Comparing price with the average does not suffer from this, but a percentage filter above the average gives wrong values on such a series.
• The MLM basket with monthly weight rebalancing is tested in the portfolio module: Portfolio Trader in MultiCharts, Portfolio Maestro in TradeStation.

MetaTrader 5 (MQL5)

• Average: iMA(_Symbol, PERIOD_D1, 225, 0, MODE_SMA, PRICE_CLOSE). Take the value from the closed bar (index 1) and check the signal once, when a new daily bar appears.
• Not every broker offers QQQ as an ETF. A Nasdaq-100 CFD is a different instrument: no dividends and a swap for every night in the position. For a system that is long half the time, the swap noticeably changes the total.
• Index CFDs trade almost around the clock, and the daily bar closes on the broker's server time, not at 16:00 ET with the cash session. The close compared with the average is therefore different.
• Some brokers do not quote crypto CFDs on weekends. A week then has 5 daily bars instead of 7, and 50 bars cover a different stretch of time than the author's.

Where the idea can break

• The QQQ figures come from a book as retold by a channel that promotes it. The average type, execution timing, dividends and interest on cash are not disclosed.
• The total for the 225-day system was unclear in the video, and 1461% comes from our review. The same 28.6% drawdown is given for both the 200-day and the 225-day versions. This may be an error in the retelling.
• The 225 length was chosen on a history where buy and hold QQQ lost 83%. Most of the filter's benefit may come from the single 2000–2002 crash.
• On TQQQ, according to the video itself, the rule barely improves risk: the average lags, and leverage magnifies the loss before the exit signal.
• Drogen gave no figures and says the fund trades something more complex. Aspell did not name the rebalancing frequency or the weights, and the index composition is changed by a committee over time.
• In a range, price crosses the average many times. Every crossover costs spread, slippage and tax on realized gains.