Step 3: Choose a Strategy Based on Your Style — and Understand WHY They Work

Now we'll translate that style into specific strategies. And most importantly: we'll explain why different families of strategies have historically worked — because without understanding the principle, you're just copying a formula that will stop working when you least expect it.

This is not a "buy this" list. It's a map of schools of thought — from Graham to O'Neil to Jim Simons' quant funds — and a guide on how to distinguish a genuine strategy from repackaged marketing.

---

What is a Strategy and What is EDGE

A strategy is a repeatable, documentable way to select stocks, when to enter them, and when to exit them. The key word is repeatable — if you can't write your rules on paper so that someone else can understand them tomorrow, you don't have a strategy, you just have feelings.

Edge is the reason why your strategy should, on average, earn more than it risks. Edge is not "I have a good intuition." Edge is a statistically demonstrable imbalance — a situation where the market behaves more predictably than it should by chance.

Examples of real edges confirmed by science and practice:

• Momentum: stocks that have risen in the last 6–12 months tend to continue rising (described by Jegadeesh & Titman, 1993).
• Value: cheap stocks (low P/E, P/B) have historically outperformed expensive ones (Fama & French).
• Quality: companies with high returns on capital and stable margins survive longer.

★ Insight: Edge is always probabilistic, not certain. A good strategy doesn't win every trade — it wins on average over many trades. If you judge it by the last three trades, you're judging noise, not signal. This is the most common beginner's mistake: "this strategy doesn't work" after four losses, even though it has a long-term 58% success rate.

Why Every Strategy Needs a Backtest — and Where the Trap Lies

A backtest means: I take a strategy, run it on historical data (say, 5–20 years back), and measure how it would have performed. Without a backtest, a strategy is just a story. With a backtest, it's a measurable hypothesis.

But a backtest can easily be falsified to oneself — not intentionally, but out of naivety. Two deadly traps:

1. Over-optimization (curve-fitting / overfitting). When you mine the strategy rules until they fit exactly to past data, you create a beautiful chart that fails immediately in reality. Example: "buy when RSI falls below 27.3 and it's Tuesday and the month ends on an even number." Such finely tuned rules describe past noise, not future law. The more parameters and the more precise the numbers, the more suspicion.

2. Look-ahead bias. When a backtest accidentally uses information that you wouldn't have had on that day. Classic: "sort stocks by their annual return and buy the best" — but you only know the annual return after a year. On the day of entry, you didn't know it. The backtest then shows fantastic numbers that you can never replicate.

★ Insight: The defense against both traps is an out-of-sample test — you tune the strategy on one part of the data (e.g., 2010–2018) and test it on another, untouched part (2019–2024). If it works even where you didn't tune it, there's a chance it has a real edge. If it only works on the tuning data, it's a curve-fit. At QMA, we therefore distinguish live results vs. backtest — live data is the hardest out-of-sample test that exists because it cannot be rewritten.

---

Seven Main Families of Strategies

None of them is "the best." Each works in a different market phase and suits a different type of person. We'll go through the logic of when a given family shines and when it suffers, and for whom it is.

1. MOMENTUM / breakout — "buys strength"

Logic: Strength begets strength. A stock that breaks to a new high from a good consolidation often attracts more buyers (institutional funds, index flows), and the trend continues. You buy what is already rising, not what is "cheap."

Legends and concepts unknown to retail:

• William O'Neil — CANSLIM: 7 criteria (Current earnings, Annual earnings, New product/management, Supply & demand, Leader, Institutional sponsorship, Market direction). O'Neil empirically found that the biggest winners had a specific fundamental profile before their meteoric rise — accelerating earnings and institutional buying.
• Mark Minervini — SEPA & VCP: "Volatility Contraction Pattern" — the price gradually calms down before a breakout (candles narrow, volume dries up), signaling that sellers are exhausted. Minervini waits for this contraction.
• Nicolas Darvas — Darvas Box: a dancer who made a fortune in the 50s by tracking price "boxes" and buying at the breakout of the upper edge.

When it works: in clear bull trends and rotations where leaders have room to run.
When it suffers: in choppy, sideways markets — breakouts fail (false breakouts) and the strategy bleeds small losses.
For whom: an active trader who can handle quick exits, stick to stops, and doesn't mind buying "expensive."

2. VALUE — "buys cheap"

Logic: The market sometimes overshoots downward and values a good company below its true worth. When you buy with a margin of safety, time works for you.

Legends:

• Benjamin Graham (Buffett's teacher): founder of value investing. "Margin of safety" — buy with such a cushion that even if you're wrong, you won't lose. Graham looked for stocks below liquidation value.
• Warren Buffett: shifted Graham from "cheap" to "great company at a reasonable price" — quality + price.
• Joel Greenblatt — Magic Formula: a brilliantly simple combination of two numbers: high return on capital (quality company) + low valuation (cheap). Greenblatt demonstrated long-term outperformance with a backtest in The Little Book That Beats the Market.
• Joel Tillinghast (Fidelity, Low-Priced Stock fund): one of the longest successful managers ever; a master of patience and avoiding mistakes in cheap stocks.

When it works: after crises, in sector panic, long-term.
When it suffers: in euphoric growth bubbles, when "cheap" stays cheap for years (value trap) and everyone chases expensive tech.
For whom: a patient, long-term investor who isn't shaken by being right for two years while the market doesn't know it yet.

3. GROWTH / quality growth

Logic: Some companies grow revenues and profits at a pace that justifies a higher price. You pay dearly today for the company to be much larger in five years.

When it works: in periods of innovation and cheap money (low rates favor future profits).
When it suffers: when rates rise — expensive growth stocks fall the most because their value lies far in the future.
For whom: an investor who can distinguish quality growth (real revenues, expanding margins) from story growth (just promises and cash burn).

4. DIVIDEND — "let the company pay you"

Logic: Companies that have increased their dividend continuously for decades (dividend aristocrats — 25+ years of growth) are typically stable, profitable, and disciplined. The strategy builds on yield on cost and the compound effect of reinvestments.

When it works: long-term, in uncertain and sideways markets as a stabilizer; great for passive income.
When it suffers: in relative comparison, it lags in wild growth rallies; beware of the "dividend trap" — extremely high yield often means the market expects a dividend cut.
For whom: a conservative investor, a builder of passive income, someone who appreciates peace of mind.

5. MEAN-REVERSION — "returns to the mean"

Logic: Price deviates too far from its average in the short term (panic, oversold) and tends to return. You buy weakness in hopes of a bounce — the exact opposite of momentum.

Concepts:

• TD sequence (DeMark): counts candles within a move and identifies trend exhaustion — the point where sellers "run out."
• Oversold indicators (RSI < 30, distance from moving average): classic oversold signals.

When it works: in sideways, range-bound markets where price bounces between boundaries.
When it suffers: in strong trends — "oversold" can keep falling further (catching a falling knife).
For whom: a disciplined trader with firm stops who understands that this strategy is the exact opposite of momentum and cannot mix them without rules.

6. TREND-FOLLOWING — "ride the trend while it lasts"

Logic: Don't guess tops and bottoms. Jump on an established trend (up or down), let winners run, and cut losses quickly. The asymmetry of "small frequent losses + rare huge gains" creates the edge.

Legends:

• Ed Seykota: pioneer of mechanical trend-following; famous for discipline and the rule "cut losses, let profits run."
• Richard Dennis & Turtle Traders: legendary experiment — Dennis took a group of novices ("turtles"), taught them a single mechanical trend-following system, and they made millions. Proof that edge is in the system and discipline, not in the trader's genius.
• Bill Dunn: decades of trend-following with huge but sustained swings.

When it works: in long, persistent trends (commodities, currencies, indices).
When it suffers: in choppy markets — a series of false signals eats away at capital (death by a thousand cuts).
For whom: a systematic, emotionally resilient trader who can handle 60–70% losing trades in exchange for a few huge winners.

7. SMART-MONEY / following institutions

Logic: Big players — insiders, funds, congressmen — sometimes know (or suspect) more. Tracking their disclosed footprints (insider buys, 13F reports, dark-pool activity) can reveal where "smart money" is accumulating.

Data footprints:

• Insider transactions: when a CEO/CFO buys their own stock with their own money, it's a strong signal of confidence (factual SEC reporting).
• 13F reports: quarterly disclosure of large fund positions.
• Dark pool / off-exchange: where large blocks are traded off the public exchange.

When it works: as a confirming layer over fundamentals/technique — confluence (overlap) of multiple signals.
When it suffers: data is delayed (13F up to 45 days), so you're never riding in real-time with insiders.
For whom: an investor who wants to add institutional context to their own analysis. This family is the backbone of QMA Smart Money.

---

Table: Seven Families at a Glance

Family	Logic (what it does)	When it works	For whom
Momentum / breakout	Buys strength, rides breakout to highs	Clear bull trends, leader rotations	Active, can handle quick exits, holds stop
Value	Buys cheap with margin of safety	After crises, sector panic	Patient long-term investor
Growth / quality growth	Pays more for future size	Innovation, low rates	Can distinguish quality from story
Dividend	Stable payouts + compounding	Sideways markets, long-term	Conservative, passive income
Mean-reversion	Buys weakness for a bounce	Range-bound markets	Disciplined, firm stops
Trend-following	Rides trend, cuts losses	Long persistent trends	Systematic, emotionally resilient
Smart-money	Follows institutional footprints	As a confluence layer	Wants institutional context

★ Insight: Notice that momentum and mean-reversion are opposites — one buys strength, the other buys weakness. Both work, but in different market regimes. A beginner who unknowingly mixes them ("I buy strength, but this one is falling, so I'll buy more because it's cheap now") has neither edge — just chaos. Choose one family and understand it deeply before adding a second.

---

How to Recognize a VERIFIED Strategy from Marketing

The internet is full of "strategies with 90% success rates." Almost all are either curve-fit, look-ahead bias, or just lies. Here are the hard criteria to sift through anything — including what you see on QMA:

1. The t-statistic — and the number of trials behind it. It measures whether the outperformance is statistically significant or just chance. The textbook threshold of 2 applies to a single pre-registered test. Once hundreds of variants are tested and the best one is picked, the threshold has to rise with the number of trials — otherwise you read the result of a search as a finding. Since 31 July 2026 QMA keeps a permanent ledger of trials and derives the bar from it; it sits well above 2. A strategy with a very low t-statistic and a beautiful chart is chance either way.

2. n ≥ 30 trades. Five winning trades prove nothing. Statistics need a sample. Below ~30 trades, every result is exploratory — an interesting hint, not proof. That's why QMA shows n (number of trades) next to the return on strategy cards.

3. Out-of-sample test. Does the strategy work on data it wasn't tuned on? The hardest out-of-sample is live operation — that's why QMA distinguishes ● Live vs ○ Backtest.

4. Honest labeling. A verified strategy isn't ashamed to admit weaknesses. QMA therefore tier-groups strategies:

• ✅ Verified — t-statistic above a threshold derived from the total number of trials, sufficient sample, holds out-of-sample.
• ⚠️ Experimental — interesting signal, but small sample or low significance. You take it at your own risk.
• ❌ Over-optimized — look great on backtest but fail on live/out-of-sample data. Kept visible as a warning, not a recommendation.

★ Insight: When someone (or a platform) shows you a return without t-stat and number of trades, you ask yourself: "What are they hiding?" Honest numbers include both n and significance. That's why on QMA you'll find these metrics directly on the cards in the /strategy hub — long-term "legendary" strategies + swing methodologies, each with its own backtest, t-stat, and live tracking.

---

Important: No Strategy Works All the Time

This is the most important sentence of the entire step, and that's why it's at the end:

Every family of strategies has its regime when it shines and its regime when it bleeds.

• Momentum loves bull trends and hates choppy markets.
• Value shines after crises and sleeps in bubbles.
• Mean-reversion profits in range-bound markets and dies in strong trends.

If you choose a momentum strategy and run it in a sideways market, it won't be a flaw in the strategy — it will be the wrong regime. And that's exactly why there is Step 4: Understand Market Phases, where you'll learn to recognize what regime we're in (bull / bear / sideways / volatile), and thus know when to run your strategy full throttle and when to pull it back.

Choosing a strategy is half the work. The other half is knowing when to trust it.

---

Takeaways from Step 3

1. Strategy = repeatable rules. Edge = documented statistical advantage. Without edge, you're just flipping a coin.
2. Backtest is mandatory, but beware of curve-fitting and look-ahead bias. Defense = out-of-sample / live test.
3. Seven families, seven logics. Choose one that fits your style from Step 2 and understand it deeply.
4. Verified ≠ marketing. Demand t-stat ≥ 2, n ≥ 30, out-of-sample. On QMA: ✅ / ⚠️ / ❌ tiers + n + t-stat on cards.
5. No strategy works in all phases. Continue with Step 4 — market phases.

Your next step: Open the /strategy hub, find the family that fits your style, and look at its n, t-stat, and live vs. backtest result. Then proceed to Step 4: Understand Market Phases.

---

Historical results and backtests do not guarantee future returns; every strategy carries the risk of loss. Consider your own situation before making investment decisions, or consult a licensed advisor.*