Swing Desk
Manual

Read this before anything else

Three corrections up front, because they change what the rest of this document is for.

1. What you described is not swing trading

Swing trading, as the term is used by practitioners and by every broker's education page, means holding for two days to about three weeks — capturing one "swing" between a pullback low and the next high. A two-to-six-month hold is intermediate-term position trading. The chart is the same; almost nothing else is.

This matters because it changes your instruments. At a 2–6 month horizon:

• The daily and weekly chart are your working timeframes. The 15-minute chart is noise you cannot trade around and should not be looking at.
• Your stop has to be wide enough to survive a normal 8–15% correction in a healthy stock. A 3% stop guarantees you get shaken out of every winner.
• You will place roughly 15–40 trades a year, not 400. Every one of them matters, and you have days — not seconds — to decide.
• Earnings reports will land inside your holds. Two to three of them. That is a structural risk you must plan for, not an event you dodge.

2. The base rate is brutal, and you should know the numbers

Barber and Odean's finding is the one to sit with: the most-active households earned essentially the same gross returns as everyone else — 18.7% — and turned that into 11.4% net. They didn't pick worse stocks. They traded the edge away in costs. The gap between the least-active and most-active quintiles was about 7 percentage points a year, and it was almost entirely friction.

Chague's study found no evidence of learning: traders who persisted for years did not get better. That is the single most uncomfortable result in the literature, and the honest reading is that persistence alone is not a strategy. Deliberate practice with a written system and a reviewed journal might be. Screen time alone is not.

3. You do not have an edge yet, and that is the whole problem

Everything in Parts 03–06 is machinery for expressing an edge. None of it creates one. Your entire job for the next 90 days is to establish, with evidence, whether the specific rules you write down produce a positive expectancy after costs — and to risk almost nothing while you find out.

What the evidence actually says

Technical analysis is not one thing. Parts of it are among the best-documented effects in empirical finance. Other parts have never survived a careful test. Knowing which is which is most of your edge over the average retail trader.

The strong evidence: cross-sectional momentum

Jegadeesh and Titman (1993) showed that buying the stocks with the best returns over the past 3–12 months and holding them for 3–12 months produced significant abnormal returns. It has been the most attacked, most replicated result in the field for thirty years, and it has largely held up out of sample.

The 2022 review marking the paper's 30th anniversary reports standard cross-sectional momentum at roughly 0.67% per month, with the "12-2" formation (rank on the past 12 months, skip the most recent month, then hold) the best-performing standard variant. Intermediate momentum — measuring returns from 12 to 7 months ago — reaches about 1.20% per month. Note what the formation and holding periods are: they are your horizon.

The critical caveat: momentum crashes

Momentum's return profile is short-option-like. It grinds and then it detonates. Daniel and Moskowitz documented the two worst episodes:

The conditions are consistent and — critically for you — identifiable in advance. Fourteen of the fifteen worst momentum months occurred when the trailing two-year market return was negative, market volatility was elevated, and the market rose sharply in that same month. In plain language: momentum dies when a beaten-down market rips off the bottom.

Daniel and Moskowitz showed that scaling exposure down when volatility is high and the market is in a bear regime lifts the Sharpe ratio from about 0.60–0.71 to about 1.19–1.20 — roughly a doubling, and it holds out of sample. This is why Part 03 has a market-regime filter and Part 04 has a volatility-scaled position size. Those two rules are not decoration; they are where most of the risk-adjusted return in this entire approach comes from.

The 52-week high effect

George and Hwang (2004) found that a stock's proximity to its 52-week high predicts future returns better than its raw past return — and that a 52-week-high strategy is less prone to the crash behavior above. This is the academic backing for the piece of chart-reading folklore that says to buy stocks near highs and avoid "cheap" broken stocks. It is one of the few pieces of trader lore that survived a careful test.

Trend following on the index

Faber's simple rule — hold the index when it closes above its 10-month (≈200-day) moving average, otherwise hold cash — returned 10.18% annualized versus 9.32% buy-and-hold from 1901–2012, with maximum drawdown of 42% vs 84%. It generates fewer than one round trip per year per asset. The return edge is small and arguably noise; the drawdown reduction is the real result, and it is exactly the tool you need to stay out of momentum crashes.

The weak evidence: chart patterns and indicator rules

This is where you should be skeptical, because this is where most retail education lives.

• Park and Irwin's survey (2007) of 95 modern studies found 56 reported positive results, 20 negative, 19 mixed — but flagged that most positive results are compromised by data snooping, ex-post rule selection, and difficulties with statistical testing, and that profitability declined sharply after roughly 1990 as markets became more efficient and costs fell.
• Sullivan, Timmermann and White tested thousands of trading rules on the Dow across a century. Once you correct for the fact that somebody testing 8,000 rules will find winners by chance, the earlier positive findings weaken considerably, and rule performance deteriorated in the out-of-sample period after the mid-1980s.
• Lo, Mamaysky and Wang (2000) automated pattern recognition (head-and-shoulders, double bottoms, etc.) over US stocks 1962–1996 and concluded that several technical indicators "do provide incremental information and may have some practical value" — a genuinely positive result, but a modest one about information, not about a profitable trading system after costs.

The regime you are starting in

Late August 2026. This section will go stale — treat it as a worked example of how to read a regime, and re-run the same four questions every month.

Four facts define the tape you are walking into:

1. The trend is up and intact. The index is near highs. Faber's filter is green: you are permitted to be long.
2. Breadth is dangerously narrow. Roughly one stock in six is beating the index. Leadership is concentrated in AI infrastructure and energy. Aggregate 2026 S&P earnings growth of ~25% is disproportionately driven by a handful of high-momentum names; the median company is growing around 13%.
3. The bond market is the pressure valve. Ten-year yields above 4.6% with a thin equity risk premium means stocks and bonds are moving inversely — there is no diversification cushion inside a broad long book.
4. Positioning is stretched. Risk appetite sits near the 99th percentile, and the Fed under its new chair, Kevin Warsh, is a fresh source of communication volatility.

The four questions to re-answer on the first weekend of every month

1. Is the S&P 500 above its 200-day moving average, and is that average rising? Yes → full size permitted. No → new positions at half size or none.
2. Is breadth confirming or diverging? Percentage of stocks above their own 200-day MA, and the advance/decline line. Index at highs with breadth falling = late-stage.
3. What is volatility doing? Rising VIX with a rising index is a warning; the crash literature says elevated volatility is a necessary condition for momentum reversals.
4. What is actually leading? Rank the eleven S&P sectors by 6-month relative strength. Your trades should come from the top three, and you should notice the day that list changes.

The system: four decisions, written down

A trading system is four decisions made in advance. If you cannot state all four in a sentence each, you do not have a system — you have opinions with a brokerage account attached.

The universe filter

Start with a hard liquidity gate, then a trend gate. Everything that fails is invisible to you.

The trend gate: Minervini's Trend Template

This is the most useful published screen for the 2–6 month horizon, because it operationalizes "the stock is in a sustained institutional uptrend" as eight boolean tests. It is a filter, not a signal — passing it does not mean buy. Failing it means don't look.

Notice that criteria 7 and 8 are the two effects with the strongest academic support, expressed as screen conditions. That is not a coincidence, and it is the reason to prefer this template over the dozens of others floating around.

Weinstein's four stages: the mental model behind it

Stan Weinstein's framing — built around the 30-week moving average on a weekly chart — is the cleanest way to hold all of this in your head. Every stock, at every moment, is in one of four stages:

Almost every catastrophic retail loss is a Stage 4 purchase justified by a story about value. The single most valuable habit you can build is the reflex of pulling up the weekly chart and asking "what stage is this?" before any other thought.

Risk: the only part that is actually mathematics

Everything else in this document is judgment. This part is arithmetic, and getting it wrong is what ends accounts. Read it twice.

Think in R, not dollars

Define R as the dollar amount you lose if a trade hits its stop. Every trade risks exactly 1R. A winner that makes three times your risk is +3R. Now every trade is comparable regardless of position size, share price, or account size, and your whole track record collapses to a single stream of R-multiples you can actually analyze.

Expectancy is your average R per trade:

A system that wins 40% of the time with average winners of 2R and average losers of 1R has an expectancy of 0.40 × 2 − 0.60 × 1 = +0.20R. Over 40 trades a year, that is +8R. At 1% risk per trade, roughly +8% before costs and taxes. That is a genuinely good outcome, and notice how unglamorous it is.

Position sizing: the stop determines the size, never the reverse

This inverts how most people trade. You do not decide to "buy $5,000 of it." You decide where the trade is proven wrong, and the distance to that price tells you how many shares you may own.

Why 1% and not 5%

Here is the same system — 40% win rate, 2R winners, 200 trades — simulated 20,000 times at four different risk levels. The bar shows the typical worst drawdown you would live through; the lighter extension shows the bad-but-not-unusual case that happens one run in ten.

The asymmetry that makes deep drawdowns unrecoverable

Losing streaks are guaranteed, so plan for them

At a 40% win rate, a run of seven consecutive losers is not bad luck — it is the expected experience. Over 100 trades:

If a seven-trade losing streak would make you abandon the system, change the system now — or change the size — because you will get one. This is the number to write on a sticky note.

The rules that follow from all of this

Entries: exactly three setups

Three, not thirty. Every additional pattern you allow yourself is another way to rationalize a trade you wanted to take anyway. Master these, and add a fourth only after 100 logged trades prove you need one.

All three require the stock to already pass the Part 03 universe filter. All three assume the S&P 500 is above its rising 200-day moving average. If it isn't, none of them are valid.

The pre-trade checklist

Run this before every entry, without exception. Its purpose is not to find trades — it is to stop you taking the ones you already know are marginal.

Your ticks are saved in this browser only. Any unticked box is a veto, not a discount.

Exits: where the money is actually made or lost

Entries are taught endlessly because they are easy to teach. Exits determine your average win, which is half of your expectancy — and they are the thing you will get wrong for the first two years.

Three exits, all defined before you enter

Why a 10-week trailing MA, specifically

Your horizon is 2–6 months. A trailing stop tight enough for a two-week swing will remove you from every position in week three. The 10-week MA on a weekly close is roughly the tightest trail that survives normal Stage 2 volatility — healthy leaders routinely pull back 8–15% mid-advance without breaking anything. Checking it weekly rather than daily is a feature: it prevents intraday noise from making the decision for you.

Earnings inside a hold — the structural problem

Over a 2–6 month hold you will sit through two or three earnings reports. Each is a coin flip that can gap through your stop overnight, meaning your realized loss exceeds 1R. Three honest options:

1. Hold through, at reduced size. Cut the position by a third or half before the print. You keep exposure to the trend and cap the gap risk. Most position traders do this.
2. Exit before, re-enter after. Cleanest risk control, but you pay two round trips and will miss some of the biggest single-day gains — which in momentum names often are the earnings gaps.
3. Hold full size and accept it. Only defensible if the position is small and you have modelled a −2R outcome as acceptable.

What is not an option is having no policy. Decide which of the three you use, write it in your plan, and apply it to every trade the same way.

The mistakes that quietly destroy expectancy

• Moving a stop down. This is the single most destructive habit in trading. It converts a −1R into a −3R and it is always rationalized the same way ("it's just noise").
• Taking profits at +0.5R because it feels good. A system with 2R average winners becomes a losing system with 0.5R winners. Your win rate goes up and your account goes down.
• Holding a loser because selling makes it "real." The disposition effect, and it is well documented in exactly the retail data cited in Part 00.
• Exiting a winner early because you're up on the month. Your P&L for the month is not information about the trade.

Mechanics: rules, costs, taxes, tools

The regulatory picture just changed

Also worth knowing: US equities settle T+1. In a cash account, proceeds are available the next business day, so you are no longer meaningfully constrained by settlement at this horizon.

Costs: small per trade, large per year

Commissions on US stocks are effectively zero at every major broker. Your real costs are the bid-ask spread and slippage, and they scale with how often you trade — which is the mechanism behind the Barber–Odean result.

Read the bottom row again. Trading illiquid stocks with market orders can cost more per year than the entire expected return of the strategy. The $25M liquidity gate in Part 03 exists for this reason, and it is why you use limit orders for everything except a stop that must fill.

Tax: your horizon has an expensive edge case

A 2–6 month hold is always a short-term capital gain, taxed as ordinary income at your marginal rate — the seven-bracket 10%–37% structure, where in 2026 the 24% bracket runs $105,701–$201,775 for single filers and the top 37% rate begins at $640,601 (single) or $768,701 (married filing jointly). Long-term treatment (0/15/20%) only begins after a hold of more than one year.

The wash sale rule

If you sell at a loss and buy the same or a "substantially identical" security within 30 days before or 30 days after the sale — a 61-day window — the loss is disallowed for the current year. It is not lost: it is added to the cost basis of the replacement position, and the holding period carries over. This matters for you in two specific ways:

• Re-entering a failed breakout. Stopped out of a name, then it sets up again three weeks later — a very common sequence — and you have created a wash sale.
• Year-end. A December loss-harvest is only clean if you stay out of the name through late January. Buying on 1 December puts you at risk through 30 January, spanning two tax years.

Inside an IRA the basis adjustment does not apply (Rev. Rul. 2008-5), so a wash sale triggered by an IRA purchase forfeits the loss permanently. Talk to a CPA if you're trading the same names in both a taxable account and an IRA.

The tool stack

Do not buy a course. The material in Parts 01–06 is the content of most $2,000 courses, and the reading list at the end is the content of the good ones.

The information stack

The instinct is to add sources until you feel informed. The evidence says that is backwards — and that a wide, always-on feed is closer to a liability than an edge.

Why your feed is the problem, not the solution

Barber and Odean's All That Glitters (2008) documents what they call attention-driven buying: individual investors, facing thousands of choices, restrict their search to whatever recently caught their attention. Measured across brokerage records, retail investors were net buyers of stocks in the news — a 9.35% buy-sell imbalance on news days versus 2.70% without. On stocks with the highest abnormal volume they bought nearly twice as much as they sold, and they piled in after both extreme positive returns (24% imbalance) and extreme negative ones (29%).

Professional managers do the opposite. Value managers in the same data were aggressive net buyers on days of low abnormal volume, because they run screens instead of reacting to headlines. The paper's title is the conclusion: attention-grabbing stocks are not, on average, the good ones.

Tier 1 — free, and genuinely sufficient

You could run this entire system on nothing but the list below, indefinitely. Start here and do not spend a dollar until you have 100 logged trades.

Tier 2 — worth money, in this order

Each of these buys you time or coverage, not edge. Add one at a time, and only when you can name the specific decision it improves.

Tier 3 — primary sources, but only for stocks you own

Once a position is open, these are what tell you whether the thesis is intact. Set them up per-position, not as a general feed.

• 8-K filings. Material events — guidance changes, executive departures, major contracts. An EDGAR alert on your holdings gives you these before any news outlet's summary.
• The quarterly transcript. Read the Q&A section and skip the prepared remarks entirely. The prepared remarks are marketing; the analyst questions are where the pressure points show.
• Estimate revisions. The single most useful fundamental stream at your horizon. A Stage 2 advance is usually sustained by forward earnings estimates being revised upward; when revisions flatten or turn down while the price keeps rising, you are in the late stage of the move. This is worth checking monthly per holding.
• Insider Form 4 filings. Cluster buying by multiple insiders is mildly informative; routine scheduled selling under a 10b5-1 plan is not. Most insider "signals" are the latter.
• Short interest (FINRA, published twice monthly). Context for why a name moves violently, not a reason to enter or exit.
• 13F institutional ownership. Filed 45 days after quarter-end, so the snapshot is always stale. The direction of change across quarters carries some information; the holdings list itself is history, not a trade idea.

The routine: when you look at what

This is the part that matters more than the source list. An unscheduled information diet is how a 2–6 month plan turns into day trading without you deciding to.

The ignore list

A useful test for any new source: name the decision it changes. If you can't finish the sentence "I read this so that I can decide whether to ___", it isn't in the stack.

The first 90 days

The goal of this quarter is not profit. It is to produce a written, tested system and 20 logged trades at a size where mistakes are cheap. Judge yourself on process compliance, not P&L.

The journal and the weekly review

The journal is not a diary. It is the dataset you use to find out whether your edge is real and where it comes from. Design it as a table you can filter and pivot.

Columns to log for every trade

The Sunday review, 30 minutes

1. Update the regime read. Index vs 200-day MA, breadth, sector rankings. Write one sentence.
2. Review every open position against its exit rules. Trail stops on weekly closes. Note any position past its time stop.
3. Score last week's trades on compliance, not outcome. A losing trade that followed the rules is a good trade. A winning trade that broke them is a warning.
4. Run the screen and build next week's watchlist. Five to ten names, with the pivot and stop pre-calculated for each. Doing this in advance is what stops you improvising on Tuesday morning.

The quarterly audit

Every 3 months, with at least 20 new trades in the book, pivot the journal and answer:

• Expectancy in R, overall and split by setup type. Very often one of your three setups is carrying the whole system and another is a net loser. That is enormously actionable and invisible without the data.
• Expectancy split by market regime. If your edge only exists above the 200-day MA, tighten the filter.
• Average MAE on winners. If winners rarely go more than 0.4R against you, your stops are too wide and you can size up. If losers routinely gap past the stop, size down.
• Compliance rate, and the P&L difference between compliant and non-compliant trades. This number tends to be brutal and clarifying.

The failure modes

Ranked by how often they end accounts, not by how interesting they are.

Where these numbers come from

1. 1993 · 2022 — Jegadeesh & Titman, Returns to Buying Winners and Selling Losers, and Wiest, Momentum: what do we know 30 years after Jegadeesh and Titman's seminal paper? — monthly momentum returns, 12-2 formation, volatility scaling.
2. 2016 — Daniel & Moskowitz, Momentum Crashes, Journal of Financial Economics — the 1932 and 2009 crash magnitudes, crash conditions, dynamic-weighting Sharpe ratios.
3. 2004 — George & Hwang, The 52-Week High and Momentum Investing, Journal of Finance.
4. 2007 — Park & Irwin, What Do We Know About the Profitability of Technical Analysis?, Journal of Economic Surveys — the 95-study survey and the post-1990 decline.
5. 1999 — Sullivan, Timmermann & White, Data-Snooping, Technical Trading Rule Performance, and the Bootstrap.
6. 2000 — Lo, Mamaysky & Wang, Foundations of Technical Analysis, Journal of Finance — automated pattern recognition, 1962–1996.
7. 2000 — Barber & Odean, Trading Is Hazardous to Your Wealth — 66,465 households, 1991–1997; 16.4% net average vs 11.4% for the most active vs 17.9% index.
8. 2008 — Barber & Odean, All That Glitters: The Effect of Attention and News on the Buying Behavior of Individual and Institutional Investors, Review of Financial Studies — the attention-driven buying imbalances quoted in Part 08.
9. 2020 — Chague, De-Losso & Giovannetti, Day Trading for a Living? — 19,646 Brazilian traders; 97% of those persisting past 300 days lost money.
10. 2007 — Faber, A Quantitative Approach to Tactical Asset Allocation — 10-month moving average timing, 1901–2012.
11. 2026 — FINRA Regulatory Notice 26-10 via ACA Group — elimination of the pattern day trader rule, effective 4 June 2026.
12. 2026 — Schwab 2026 mid-year US market outlook (breadth, earnings concentration, yields, positioning) and Trading Economics (index levels as of 27 Aug 2026).
13. 2026 — Fidelity on wash-sale rules; Kiplinger on 2026 capital-gains thresholds; Tax Foundation 2026 bracket tables.
14. 2026 — Tool pricing: TraderHQ on Koyfin vs Finviz Elite, TraderHQ on IBD Digital and MarketSurge, and Liberated Stock Trader on TradingView tiers. Prices change — check before subscribing.
15. Books — Minervini, Trade Like a Stock Market Wizard (trend template, VCP); Weinstein, Secrets for Profiting in Bull and Bear Markets (stage analysis); Van Tharp, Trade Your Way to Financial Freedom (R-multiples, position sizing); Brett Steenbarger, The Daily Trading Coach (process and psychology).

Drawdown, streak and expectancy figures in Part 04 are from Monte Carlo simulation (20,000 runs × 200 trades) run for this document, not from any published source. They describe the arithmetic of position sizing, not a forecast of your results.