Why One Indicator Isn't Enough (and Why Three Tuned Indicators Aren't Better)

Article #1 made the case that the SMA200 is a useful trend filter, not a magic bullet. Reasonable people read it and asked the obvious follow-up: if one indicator isn't enough, which one should I add?

This is the wrong question, but you can't tell people that until you've answered the right one.

What a single indicator actually does

The SMA200 answers exactly one thing: is the underlying in a long-term uptrend or downtrend relative to its 200-day moving average?

That's it. Above SMA200 means the trend regime is up. Below means it's down. Everything else (entry timing, exit timing, position size, sector exposure, when to take profits) is outside its scope.

This is a feature, not a bug. The SMA200 is the most-cited trend filter in markets because it's hard to argue with at 200 days. Shorter averages whipsaw on minor corrections. Longer averages miss real reversals. 200 days is roughly the modal answer across hedge fund desks, academic papers, and retail trading communities. It works precisely because it's simple and slow.

But "the regime is up" is only half the story. It doesn't tell you:

• Whether the trend is overextended and about to mean-revert
• Whether momentum is accelerating or decelerating
• Whether volume confirms the move
• Whether the stock is cheap or expensive at its current level
• What's happening in correlated sectors

This is why people add more indicators. The intent is correct: more dimensions equal more information. The execution is where it usually breaks.

Why adding indicators usually fails

Three failure modes that show up constantly in retail trading content.

1. Correlation collapse

Most popular indicators are derivatives of price. RSI is a price-momentum derivative. MACD is two SMAs subtracted. Stochastics, Williams %R, Money Flow Index, Bollinger Bands. All functions of price alone, with slightly different mathematical lensing.

Adding three of them gives you one piece of information about price, three different ways. You think you're getting confluence. You're getting echo.

The fix: pick indicators that measure genuinely different dimensions.

• Price-based (one): SMA200 for trend regime. Pick one and move on.
• Momentum-based (one): MACD or RSI for direction acceleration.
• Volume-based (one): OBV or volume relative to its 20-day average for conviction.
• Volatility-based (one): ATR or Bollinger Width for regime mode.

One from each bucket, not three from the price bucket.

2. Parameter tuning theater

This is the bigger trap. Pick the right indicators and the urge to optimize their parameters is overwhelming. What's the right RSI period? 14, 21, 30, 50? What's the right threshold? 70, 65, 80?

I ran a walk-forward test on this exact question. About 60 parameter combinations on each of TQQQ, SOXL, and UPRO. Trained on 2000-2015, evaluated on 2016-2026. The question: do the best parameters from one decade predict the best parameters in the next?

Train-to-test Sharpe correlation:

Ticker	Correlation
TQQQ	+0.19
SOXL	-0.09
UPRO	+0.09

Real predictive correlation would be 0.4 to 0.7. Near zero means the parameter optimization has essentially no signal about future performance. Top-trained parameters degraded 0.3 to 0.5 Sharpe on out-of-sample test. The "winners" from the in-sample period were not the winners going forward.

Translation: when you read a Reddit post saying "SMA(177) + RSI(23) at threshold 67 + MACD(11,28,9) gives Sharpe 1.8," you are reading the output of a parameter sweep on past data. That exact combination is essentially random for the next decade. The author isn't lying. They are just measuring something that won't repeat.

3. Confirmation seeking

This is the hardest one to see in yourself. Most people who add indicators do so because they want their existing thesis confirmed.

You are long QQQ. You want to stay long. You find an oscillator setting that says "still bullish." You ignore the one that says "overbought." You think you're doing confluence analysis. You're cherry-picking.

The discipline: pick your indicator stack BEFORE you have a position. Write down the rule. Follow it even when it disagrees with what you want to believe.

The honest case for confluence

After all the above pessimism, there IS a real case for using more than one indicator. It just isn't what most retail content sells.

The case is statistical. If two genuinely independent indicators each have a 60% true-positive rate on entry signals (generous), the joint probability that both fire as false positives drops sharply. A trade that requires both to agree gives up some opportunities (you miss entries one signal would have caught) but cuts false-positive rate disproportionately.

This is why "confluence" gets cited so much. The math is real. The trap is in the execution. Most retail traders who run "confluence strategies" are not picking independent indicators with stable parameters and waiting for them to agree. They are tuning a multi-knob system on historical data and reporting the in-sample fit.

What confluence with discipline actually looks like:

1. Pick 3 indicators from 3 different buckets (trend, momentum, volume)
2. Use their conventional parameters (SMA200, RSI(14), MACD(12,26,9)). Do not tune.
3. Define a binary rule per indicator
4. Wait for all 3 to agree before entering
5. Wait for any 2 to disagree before exiting

The Sharpe improvement from this kind of stack over a single-indicator strategy is real but modest. In our portfolio testing, adding SMA200 filtering to a leveraged-equity sleeve (the simplest possible "trend confluence" you can build, equity trend filter on top of equity position) improved Sharpe by up to +0.238 across a range of portfolio compositions, and cut max drawdowns by 15 to 50 percentage points across a 25-year stress window. That's the kind of improvement that's durable. The lift scales with how much equity-vol-decay sits in the portfolio: a 75/25 leveraged-equity/gold split gets the biggest benefit, a 25/75 split gets almost none. The filter is best understood as a portfolio construction tool, not a single-asset rescue. Full breakdown in The Hidden Cost Every Leveraged-ETF Backtest Ignores, which also documents the borrow-cost correction that revised these Sharpe numbers down from the earlier overstated values.

Adding RSI and MACD parameters tuned to in-sample data on top of that? The walk-forward data says the additional gain doesn't survive contact with the next decade.

What this means in practice

If you're using the SMA200 status on a tool like this one as the entry to your trading process, the rational next steps depend on what kind of trader you are.

For most retail traders, the simplest meaningful additions to a single SMA200 check are:

• Days held: how long has the current regime held? A flip 3 days ago is fragile. A flip 47 days ago is durable. This is just a derivative of the SMA200 signal but it tells you something different.
• Sector check: is this stock above its 200-day SMA in a sector that's also above its 200-day? Single-stock signals are noisier than sector signals.
• Volume confirmation: was the most recent SMA200 cross on heavier-than-average volume, or did it sneak through on a quiet day?

Any one of these is a meaningful second dimension. None requires parameter tuning. None of them is more sophisticated than the SMA200 itself.

For active options or LEAPS traders specifically, the confluence case is stronger because your trade duration matches the indicator timeframe. SMA200 is a multi-month regime check. LEAPS are multi-month directional bets. Aligning indicator timeframe to position timeframe is the highest-leverage thing you can do.

A broker built for that workflow makes a real difference. tastytrade is the one I'd consider for a LEAPS-centric stack because the per-leg commission cap doesn't punish multi-leg structures the way some brokers do. See the full broker shortlist for the honest tradeoffs.

For long-term portfolio holders, confluence between a trend filter (SMA200 on the underlying) and macro regime (real yields, gold weighting, defensive sleeves) is more important than confluence within technical indicators. Adding a 10% gold sleeve to a leveraged equity position has, in our 23-year testing, a more durable Sharpe contribution than any RSI overlay we tested.

The discipline is the strategy

Here's the unsexy truth that doesn't make great Twitter content: picking the right indicator stack is maybe 20% of the problem. The other 80% is being able to follow it when it tells you something you don't want to hear.

The SMA200 below your stock right now isn't telling you the stock is bad. It's telling you the regime is currently risk-off for that name. What you do with that information is the actual trade.

If you're going to use a single binary signal, use the SMA200, use it consistently, and don't override it. If you're going to add a second signal, pick one from a different bucket (volume, sector, macro), use conventional parameters, and don't tune it. If you're going to add a third, do the same.

What you cannot do, no matter what the backtest says, is build an 11-parameter confluence stack on three years of data and expect it to generalize. The math is against you. The data is against you. The next-decade regime is against you.

Caveats worth naming

1. The walk-forward result is specific to leveraged-ETF parameter tuning on 27 years of data. It probably generalizes to most "multi-parameter technical strategies on liquid equity instruments," but the honest claim is bounded to what we tested.

2. Some indicators DO have more predictable behavior across regimes. Volume-based indicators behave more consistently than price-derived ones. ATR (volatility) is structurally regime-aware. Macro signals (yield-curve inversion, dollar strength) operate on slower timeframes that are less prone to parameter-tuning theater. Not every indicator is equally subject to the walk-forward critique.

3. Discretionary traders can absolutely beat mechanical systems if they're disciplined and skilled. Nothing here argues against using your judgment. It argues against fooling yourself that backtest-tuned parameters represent your judgment.

4. The "use conventional parameters" advice is conservative. It is not optimal in any specific window. It IS robust across windows. If your goal is "best results in this exact regime," tune away. If your goal is "results I can stand on for the next decade," don't.

5. Past performance is not predictive. Especially of indicator parameters.

Source research

• LRS Family Replication + Walk-Forward — the full walk-forward methodology, parameter sweep results, and train/test Sharpe correlations
• Portfolio Archetypes Search — the portfolio-construction angle (why adding gold beats adding indicators)

Check current SMA200 status on the tickers tested: TQQQ, SOXL, UPRO.

---

If you've made it this far and want to start thinking about confluence for your own setup, the easiest place to begin is just adding the "how long has this state been held" question to your existing SMA200 lookups. Two checks, both simple, both useful.

If you'd rather skip the analysis and have a multi-signal stack shown to you automatically for every US stock, that's what we're building toward. The free SMA200 check on the homepage is the gateway. The next layer is coming.

Want today's SMA200 status for any US stock? Type a ticker on the homepage →

Want to read the original buy-and-hold-vs-filter analysis this article builds on? Read Article #1 →