Stock Trading curriculum

Like a mutual fund, an exchange-traded fund pools money into a portfolio, and each share is a slice of it. The difference is how you get in and out: you buy and sell existing shares from other traders during the day, at whatever price the market sets, instead of redeeming with the fund at its end-of-day value.

Net asset value (NAV) is the fund's assets minus its liabilities, divided by the shares outstanding, and it is struck once a day at the close. The market price is whatever buyers and sellers agree on minute to minute. The two usually sit close together, but your trade always fills at the market price, never at NAV.

Authorized participants are usually large broker-dealers that deal with the fund in big blocks called creation units. To create, one delivers a basket of securities and cash and receives new shares; to redeem, it hands shares back and takes the basket. As an individual trader you never deal with the fund this way.

If shares trade above the value of the basket, a participant can create shares and sell them; if below, it can buy shares and redeem them for the more valuable basket. Both trades push the price back toward value. The anchor holds best when the holdings themselves trade freely and are easy to price.

Buying at a 1% premium means paying $101 for $100 of holdings. Gaps tend to open when the underlying securities are difficult to trade or to value, exactly when the arbitrage link that holds price near value is weakest. A fund's prospectus or website shows its history of premiums and discounts, which tells you how tightly it has tracked its value.

Quote services publish this estimate roughly every 15 seconds from the estimated value of the fund's holdings. Comparing it with the quoted price shows whether you are about to pay close to what the fund owns. It is an estimate built from the latest holding prices, so it is only as current as they are.

An index ETF tries to match a published index after costs, so you judge it by how closely it tracks. An active ETF's adviser buys and sells with no index to conform to, so you judge it by its holdings and strategy. Two funds with similar names can still hold very different things.

In a capitalization-weighted index the biggest companies make up most of the weight; an equal-weighted version gives every member the same slice. Both can carry "S&P 500" in their names. Before comparing two funds, check the weighting method, because it decides which companies drive the result.

When an authorized participant redeems, the fund can hand over securities instead of selling them for cash, so it realizes fewer capital gains that would otherwise be passed to shareholders. The tax rules for owning the two kinds of fund are broadly similar; the difference comes from how redemptions are met.

A 0.50% expense ratio costs about $50 a year on $10,000, deducted inside the fund, so it shows up only as a return below the index. For an active trader the larger cost is often the bid-ask spread and slippage paid on each round trip, which the expense ratio does not include.

Because authorized participants can create shares on demand, a fund with low daily volume can still absorb a large order near its value if its holdings trade easily. The quoted spread and the liquidity of the basket tell you more than volume alone. Low volume bites hardest when the holdings themselves are thinly traded.

An index fund's return trails its index by roughly its costs, nudged up or down by how it samples the index, cash it holds and its trading. You see the gap by comparing the fund's return with the index over the same period. A persistent gap wider than the fee is worth investigating before you rely on the fund as a stand-in for the index.

SPY, QQQ and IWM are fund tickers, not indexes. Each fund's prospectus states its benchmark, its structure and its costs. When someone says "the market fell 2%," ask which index, because a large-company index, a technology-heavy index and a small-company index can move quite differently on the same day.

Three sponsors run funds on the same index, so their holdings are nearly identical. They differ in legal structure, costs and trading volume: SPY, the oldest, is a unit investment trust, while VOO is built as an ETF share class attached to a larger index mutual fund. For funds on the same index, compare costs and spreads rather than past returns.

A total-market index adds mid-sized and small companies to the large ones. Because it is weighted by market value, the largest companies still dominate it: from 2015 to 2019 its daily returns correlated 0.996 with SPY's. Choosing between the two changes very little about day-to-day risk.

The Nasdaq-100 is drawn from companies listed on the Nasdaq exchange, and its sponsor describes it as rich in leading technology stocks. That tilt makes it swing more than the S&P 500: over 2015 to 2019 its daily returns moved about 1.2 times as much as SPY's. It is a narrower bet than its large-company label suggests.

The Russell 2000 holds companies far smaller than the S&P 500's members. From 2015 to 2019 IWM's daily returns still correlated 0.87 with SPY's, so owning both diversifies less than the different index names suggest. The useful question is how much it adds beyond the large-company exposure you already hold.

A fund like this adds companies the US indexes do not hold, along with exposure to their currencies. Its daily returns correlated 0.83 with SPY's over 2015 to 2019: lower than two US funds, but far from independent.

VWO's benchmark moved from an MSCI index to a series of FTSE indexes after 2013, each with its own rules for which countries and companies qualify. A fund's history can therefore span several different indexes. When you compare two emerging-markets funds or study one fund's past, read the benchmark and its history, not the label.

It tracks an FTSE index of companies of all sizes outside the US, so it combines what EFA-style and emerging-markets funds hold separately. Pairing it with a total US fund gives a whole-world stock exposure built from two pieces. The weights follow market values, not a judgment about which country is more attractive.

Each Select Sector SPDR fund tracks one sector index built from S&P 500 members. A sector fund is far less diversified than the whole index, and the sponsor warns of sector and non-diversification risk. Sectors also differ in sensitivity: from 2015 to 2019 utilities moved about a third as much as SPY, technology about 1.2 times as much.

In a cap-weighted index each company's weight is its market value divided by the total, so a few giants can make up a large share of the fund. A broad index fund can therefore behave much like a bet on its top holdings. Check the fund's top-ten weight before treating it as evenly spread.

SPY and QQQ share many of their largest holdings, and their daily returns correlated 0.93 over 2015 to 2019. Adding QQQ to SPY mostly raises the weight of the same big technology names. Compare holdings, or at least correlations, before counting two funds as two separate bets.

If foreign stocks rise 5% in their own currency while that currency falls 3% against the dollar, a US holder earns about 1.85%: 1.05 × 0.97 − 1. The currency leg can add to or erase the stock return. A fund that hedges currency removes most of that leg and pays a cost to do it.

An existing bond pays a fixed coupon. When new bonds offer more, the old one is worth less, so its price drops until its yield matches the market; when rates fall, the reverse happens. A bond ETF holds many such bonds, so its price moves opposite to rates in the same way.

The further away a bond's final payment, the longer a rate change keeps affecting its value, so long bonds swing more for the same rate move. Investors usually demand a higher coupon to hold that risk. A fund of long bonds is a very different trade from a fund of short ones, even when both hold the same issuer.

These funds hold Treasuries, government-agency mortgage securities and investment-grade corporate bonds across many maturities, so their rate sensitivity sits in the middle. The two follow closely related versions of the same Bloomberg aggregate index. They are a common stand-in for "the US bond market" in correlation studies.

Both hold only US government debt, so credit risk is minimal, yet they behave like different assets. As rates climbed, TLT fell 48.4% on dividend-adjusted prices from August 2020 to October 2023, while SHY fell about 3%. "Safe" describes the issuer's credit, not the price path.

If you hold one bond to maturity, rate swings along the way do not change what you are repaid. A bond fund sells aging bonds and buys new ones to keep its target maturity, so there is no date when its price is guaranteed to return to par. Waiting does not undo a fund's rate loss the way it can for a single bond.

Rate risk is the price effect of changing market rates; credit risk is the chance the issuer cannot pay. Treasury funds carry mostly rate risk, while funds of lower-rated corporate bonds carry much more credit risk. When a fund is labeled "bonds," ask which of the two risks it mainly holds.

A 2x fund seeks twice the index's return for one day; a −1x inverse fund seeks the opposite of it; a −2x fund seeks double the opposite. The target is reset every day. Nothing in the fund's objective promises twice the index's return over a week, a month or a year.

Suppose an index gains 10% one day and loses 10% the next, ending down 1%. A 2x fund gains 20% then loses 20%, ending down 4%: four times the index's loss, not two. Each day the fund did exactly what it promised; compounding those daily results is what breaks the simple multiple.

Over ten days alternating between +5% and the move that brings it back, an index finishes flat, but a 2x fund finishes down about 2.4%. The more an index swings without going anywhere, the more a leveraged fund loses. This decay comes from the arithmetic of compounding, not from fees.

With the index up 10% and then down 10%, the index ends down 1%. A −1x fund falls 10% and then rises 10%, also ending down 1%. A trader who bought the inverse fund expecting to profit from a falling index lost on both legs of the bet.

These contracts add costs and risks beyond those of the index, and the fund may still miss its daily target. Reading the prospectus shows the techniques used and their risks. A fund that relies on swaps also depends on the institutions on the other side of those contracts.

It removes the diversification an index provides and then amplifies one company's daily moves, so its swings are larger than the stock's own. The daily reset and compounding effects apply here too, often more sharply because single stocks are more volatile than indexes.

The SEC's investor bulletin warns that holding these funds longer than a day can produce significant losses even when the index rises over the same period. They can also cost more and realize short-term gains from daily rebalancing. Holding period is part of what the product is.

Exchange-traded products that invest mainly in commodities, currencies or related futures sit outside the Investment Company Act that governs ordinary ETFs. Limits on leverage and illiquid holdings that protect fund investors do not apply in the same way. The ticker looks the same; the legal wrapper does not.

Owning a share is a claim on part of a vault of gold, not on gold-mining companies, whose shares behave more like stocks. Over 2010 to 2019 its daily returns correlated −0.31 with a broad dollar index. As a commodity trust, it is not a registered investment company.

Take the standard deviation of daily returns and multiply by the square root of about 252 trading days. SPY's figure was 6.7% in 2017 and 33.3% in 2020, the same fund in two very different years. A volatility number belongs to its period, so quote one without its dates and it says little.

Realized volatility describes what already happened. Implied volatility is the level that makes an options pricing model match current option prices, so it reflects what traders are paying for protection. The two often differ, because one measures the past and the other prices the future.

It is computed from a range of S&P 500 call and put option prices, under a method its exchange, the Cboe, publishes. It is often called the investor fear gauge. Like any implied measure, it reflects what traders are paying for options, not a forecast that a move will happen.

Funds and notes linked to volatility track indexes of VIX futures, often blended to a constant maturity such as one month. Futures prices can differ a lot from the spot VIX you see quoted. A product can fall while the VIX headline rises, because it owns futures, not the index.

In contango the futures curve slopes upward; in backwardation, near futures trade above later ones. Studies of the VIX futures curve find contango is the usual state, consistent with buyers paying a premium for volatility protection.

To stay near a one-month maturity, a product keeps selling the cheaper expiring future and buying the dearer later one. In contango that roll costs money every month, so a long-volatility fund tends to bleed between spikes. It can still jump in a crash, which makes it a short-term instrument rather than a hold.

Over 2010 to 2024, SPY's daily returns and the VIX's daily point changes correlated −0.80, a strong negative relationship. The link is mechanical as much as emotional: falling markets raise demand for put options, which lifts implied volatility. Strong as it is, a −0.80 correlation still leaves plenty of days that break the pattern.

A common setting is a 20-period average with bands two standard deviations away. For five closes of 10, 11, 12, 11 and 13 the average is 11.4 and the population standard deviation about 1.02, giving bands near 9.36 and 13.44. The bands widen when prices swing more and narrow when they calm.

The band only measures distance from the average in units of recent volatility. In a strong trend prices can ride along the outer band for weeks, and in calm markets a small move can poke outside. Read a band touch as a description of the last move; any trading rule built on it needs testing like any other.

The open is the first trade of the period, the close the last, and the high and low the extremes in between. A daily bar compresses a whole session into four numbers, so it hides the path: two days with identical bars can have traveled very differently.

The real body is filled or colored by whether the close was above or below the open, and the thin lines above and below, called shadows or wicks, mark the high and low. Colors are a display setting, so check the legend before reading one.

A move that is a sharp spike on a five-minute chart can be invisible on a weekly one. Each interval shows a different slice of noise and trend, and an indicator set to 14 periods means 14 minutes on one chart and 14 weeks on another. State the interval whenever you state an indicator reading.

A rise from 10 to 20 and a rise from 50 to 100 are both +100%. On a linear scale the second looks five times as tall; on a log scale they look the same. For a long history or a fast-growing stock, a linear chart exaggerates recent moves and flattens old ones.

A 2-for-1 split halves the share price overnight without anyone losing money, and a dividend drops the price by roughly the payout while handing it to holders. Raw closes show these as false losses. Use adjusted closes for returns, drawdowns and correlations, and raw prices for what actually traded.

Earnings released after the close can send the next open far from the prior close, with no trades in between. A stop order sitting inside that gap cannot fill at its stop price. Overnight and weekend gaps are a risk that no intraday chart pattern shows in advance.

Traders mark prior lows as support and prior highs as resistance, and some indicators build on those levels. They describe where price turned before; whether it turns there again is an empirical question. A zone drawn after the fact always fits the past better than it will fit the future.

Two analysts can draw different lines on the same chart, and a pattern such as a head-and-shoulders is often recognized only once it has finished. That flexibility makes patterns easy to find in hindsight and hard to test. A pattern claim becomes testable only when it is written as an exact rule.

Catalogs list patterns such as dojis, engulfing patterns and shooting stars, each a rule about the relative size of bodies and shadows. Researchers have tested them, including with machine-learning models. Treat any pattern the same way as an indicator: define it exactly, then test it out of sample and after costs.

Volume is usually drawn as bars beneath the price chart. High volume on a move means many shares traded, not that the move was right or will continue. Compare a bar with the stock's own typical volume, since a busy day for one stock is a quiet day for another.

For closes of 10, 11, 12, 11 and 13, the five-day simple moving average is 11.4. Each new day drops the oldest close and adds the newest, so the line moves smoothly and every close in the window counts equally.

An average is built from past closes, so it turns only after price has turned. A simple average's data have an average age of (N − 1) ÷ 2 periods: 9.5 days for a 20-day average. Smoothness and speed trade off directly, and no window gives both.

A classic study called this the most popular moving-average rule: a one-day "average", which is just the price, against a 200-day average. In code it is a one-line condition, which is why it is so widely charted. What it computes is whether the recent price sits above or below its own long-run mean, nothing more.

The labels describe two lines intersecting, for example a 5-day average crossing a 20-day one. Studies treat the golden cross as an entry and the death cross, sometimes called a dead cross, as an exit. The names sound like forecasts, but the event is just two averages of the same past prices changing order.

Practitioners favor round numbers tied to the calendar, like 5, 20 and 60 days for a week, month and quarter, or 200 days. Nothing about markets makes these numbers special. Because they are widely watched, results that depend on one exact window and collapse at its neighbors deserve suspicion.

When prices chop within a range, short and long averages keep crossing, so the rule buys after small rises and sells after small dips. Each round trip pays the spread and commissions. Crossover rules depend on trends long enough to cover all the false starts.

A table of results for every combination of short and long windows will always contain a top cell, even in random data. Choosing it and reporting its return is a backtest of the selection, not of the rule. Test the chosen pair on data that played no part in choosing it.

Each day the EMA moves a fixed fraction of the way toward the new close. With N = 5 the factor is 1/3, so from an average of 11.4 a close of 14 lifts it to about 12.27. A simple average over the same five closes would read 12.2. The EMA reacts faster because old data fade instead of dropping out.

When recent prices run above the longer trend, the fast EMA pulls ahead and MACD turns positive; when the short-term pace slows, the gap narrows. It is a measure of the gap between two trends, in price units, so its size is not comparable across stocks at different prices.

The usual signal line is a 9-day EMA of MACD. The histogram plots MACD minus the signal line, so it shrinks before a crossover and flips sign at one. All three lines come from the same closing prices, smoothed three times, so they add views of one series rather than new information.

In a study of five Southeast Asian indexes from 2000 to 2013, this rule was among the few that still earned net returns in some markets. Its winning trades were usually fewer than half; it profited because average gains were larger than average losses, not because it called direction well.

With a high of 52, a low of 49 and a previous close of 54, the three candidates are 3, 2 and 5, so the true range is 5. Using the previous close captures overnight gaps that the day's own high and low would miss.

The usual window is 14 periods, smoothed with Wilder's method. ATR is in price units, not percent, and it says nothing about direction. Traders use it to scale stop distances to how much a stock normally moves, so that a volatile stock gets a wider stop and a smaller position.

Plus and minus directional indicators divide smoothed upward and downward movement by ATR. ADX smooths their normalized difference, so it rises in any strong trend, up or down. A high ADX with prices falling still reads as a strong trend, which is the commonest misreading.

These cutoffs are practitioner conventions, and ADX lags, since a trend must be well under way before it rises. Rules built on it, like other trend filters, aim to stay out of sideways markets and so give up the early part of each new trend. The thresholds are inputs to test, not facts about markets.

Trend tools assume a move will continue; these oscillators assume an extended move is due to fade. The two families can give opposite readings on the same chart at the same moment. Knowing which idea an indicator encodes tells you which market conditions will make it look clever and which will make it look broken.

RSI = 100 × average gain ÷ (average gain + average loss). For five daily changes of +1, −0.5, +1.5, +1 and −0.5, the average gain is 0.7 and the average loss 0.2, so RSI is about 77.8. The standard window is 14 periods.

Above 70 means recent gains have far outweighed losses; it does not mean the gains must stop. In a strong uptrend RSI can stay above 70 for a long time, and a rule that sells at 70 keeps stepping in front of the trend. The levels describe the past window only.

Some use an exponential average, some Wilder's smoothing, some a simple average. Fed the same prices, they give different numbers and cross 70 on different days. When comparing readings or reproducing a study, match the averaging method and the window, not only the name.

With a close of 48 in a range whose low is 40 and high is 50, %K is (48 − 40) ÷ (50 − 40) × 100 = 80. Readings above 80 are conventionally called overbought and below 20 oversold. The standard setting in the study looks back only five days.

%D smooths %K, typically over three periods. Instead of fixed levels, the crossing rule goes long when %K crosses above %D and short when it crosses below. In the Southeast Asian study, this crossing version and MACD were the rules the authors singled out as still possibly earning net excess returns.

The label invites a reading that the price is too high and must fall. All the indicator measures is the balance of recent gains and losses, or the close's position in a range. Whether extremes tend to reverse in a given market is a question for evidence, and the answer changes across markets and periods.

Default windows such as 14 for RSI were proposed decades ago and became conventions. The study's authors note nothing guarantees they work at other times or places. Optimizing them on past data has its own trap, covered in the evidence chapter.

With closes of 50, 51, 50.5 and 52 and volumes of 1,000, 1,200, 800 and 1,500, OBV runs 0, 1,200, 400, 1,900. A study rule went long when OBV crossed above its own short EMA. The line's level is arbitrary; only its direction and turns are read.

A day that closes up one cent adds the whole day's volume, exactly as a day up 5% would. That makes OBV sensitive to noise near unchanged closes. Two stocks with very different moves can show the same OBV path.

The story is that informed traders accumulate before prices move, lifting OBV first. If that were reliably true, a rule built on it would earn money after costs. Treat the story as a hypothesis, and test an OBV rule against costs like any price rule.

Divide the total value traded by the total shares traded. If 100 shares trade at $10, 300 at $10.50 and 100 at $11, VWAP is $5,250 ÷ 500 = $10.50. It resets each session and is often used as a yardstick for whether a fill was cheap or dear relative to the day's trading.

A buy filled below VWAP got a better price than the session's average, but the stock can still fall. Large traders split orders through the day partly so they do not push the price away from VWAP. Judge execution and judgment separately.

High volume shows strong participation; it does not show which side was right. Heavy volume on a drop can mark panic selling near a low or the start of a longer decline. Any rule that conditions on volume needs the same out-of-sample test as any price rule.

If prices already reflect everything in their own history, rules built only on that history cannot beat the market after accounting for risk. Whether any rule does is therefore an empirical test of how efficient a particular market is, not a matter of opinion about charts.

A widely cited review of 95 studies published from 1988 to 2004 counted 56 finding profits, 20 finding none and 19 mixed, with profits common at least until the early 1990s. Later work points to declining profitability over time. A rule that worked in older data has to show it still works now.

In a study of five Southeast Asian stock indexes from 2000 to 2013, several rules produced statistically significant gross returns, but after estimated round-trip costs of 0.5% to 1.5%, most did not earn net returns. A backtest without costs measures the rule's signal, not a trader's result.

Divide the rule's excess return over buy-and-hold by the number of signals. If that is 0.3% per trade and a round trip costs 0.5%, the rule loses money in practice. Frequent-signal rules need very cheap execution, so the cost a trader actually pays decides whether a published edge exists for them.

In the same Southeast Asian study, rules performed strongly in Thailand but earned no significant returns in Singapore, the most mature market tested. Results from one market do not transfer to another, and the large, heavily traded US markets are the hardest place to expect leftovers.

Searching for the best window makes a rule fit its sample. In the study, optimized settings made already-profitable rules more profitable but did not rescue unprofitable ones, and the best settings differed by market. Only performance on data held back from the tuning tells you what the rule might do next.

Try enough rules on one dataset and some will look profitable by chance. Bootstrap tests developed for trading-rule studies, such as White's Reality Check, adjust for how many rules were searched. Ask how many variants were tried before the reported one.

Look-ahead bias uses information that was not available at the time, such as a closing price to decide a trade placed at the open. Survivorship bias tests only stocks that still exist, silently dropping the failures. Both push results up, and both hide easily in code.

The authors measured trade efficiency, the share of each available price move a trade captured, and found it generally low. Even the strategies that beat buy-and-hold earned their profits by letting winning trades run and cutting losers, not by calling turning points.

The disposition effect is the habit of selling winners too soon and holding losers too long. The study's authors suggest rules may help individual traders by countering it, even where markets are close to efficient. That is a claim about behavior, and it is weaker than a claim that the rule predicts prices.

+1 means they always move in the same direction in proportion, −1 always opposite, and 0 no linear relationship. It says nothing about the size of the moves. For x = 1, 2, 3, 4, 5 and y = 2, 1, 4, 3, 5 the correlation is 0.8: closely related, yet the second pair moves against the first.

Two series that both drift upward will show a high correlation of levels even if their day-to-day moves are unrelated. From 2016 to 2019 the price levels of SPY and GLD correlated +0.53, while their daily returns correlated −0.21. The level figure measured shared drift, not co-movement.

The FEDS Note measures stock-bond correlation over a rolling one-year window of 252 trading days. Even inside 2019, a rolling 63-day SPY–TLT correlation ranged from −0.60 to −0.27, and a 20-day window ranged from −0.79 to +0.27. Short windows are noisy; always state the window with the number.

Bond prices fall when yields rise. So when stocks fall alongside falling yields, the stock–yield correlation is positive and the stock–bond-price correlation negative, and both say bonds cushioned stocks. In 2011, SPY's correlation with 10-year yield changes was +0.62 and with TLT returns −0.71. Check which series a chart uses before reading it.

The Federal Reserve note groups stock–bond correlations into a 1990s period, a 2000 to 2021 low-inflation period and a post-2021 period, with a different sign in different eras. A correlation estimated in one regime can mislead badly in the next. Quote it as "in that period, under those conditions."

Two assets can move together because a third factor, such as interest-rate expectations, drives both. Oil and energy stocks are linked by a real business connection; many other pairings are coincidence or shared exposure. Name a mechanism before treating a correlation as a lever.

Beta equals the correlation times the asset's volatility divided by the market's. From 2015 to 2019, utilities (XLU) had a beta near 0.35 to SPY and technology (XLK) near 1.20. Two assets can share the same correlation with the market and still have very different betas.

The FEDS Note argues that bonds hedge stocks when inflation is near target and central banks can cut rates in a sell-off, and fail to when inflation pressures force policy to stay tight. A hedge sized from one era's correlation can turn into a second source of loss in the next.

In SPY and TLT daily returns the correlation was −0.46 in 2019 and +0.08 in 2022. In 2022 both fell together; SPY lost 24.5% from its January peak and long Treasuries fell sharply too. A portfolio built on the earlier negative correlation had less protection than its history implied.

With inflation anchored, bad news for stocks lets central banks ease, so bond yields fall and bonds rally. When inflation runs high, weak data may arrive alongside tighter policy, so stocks and bonds can fall together. A 2026 FEDS Note found US yields reacting about three times as much to inflation surprises after 2021 as before.

SPY's daily returns and the VIX's daily changes correlated −0.80 over 2010 to 2024. That is why VIX products are sold as crash hedges. Their structural cost from rolling futures, covered in the volatility chapter, is what makes holding one as a permanent hedge expensive.

GLD's daily returns correlated −0.31 with a broad dollar index over 2010 to 2019 and −0.36 over 2020 to 2024. Gold is priced in dollars, so a weaker dollar makes it cheaper for other buyers. A moderate negative correlation still leaves many days when both rise or both fall.

The study compared the dollar index with the VIX and gold as hedges and found the dollar's hedge ratio rose during the crisis as its correlation with the S&P 500 increased. The finding is tied to one crisis; a hedge that worked in a flight to dollars may not work in a dollar-driven shock.

XLE's daily returns correlated +0.59 with crude oil prices over 2010 to 2019 and +0.61 over 2021 to 2024. A study of Japanese stocks found the influence of oil on stock returns fell by about 89.5% during COVID-19. Energy companies also respond to costs, debt and the broad market, not oil alone.

QQQ's daily returns correlated +0.31 with 10-year yield changes over 2015 to 2019, when rising yields meant a stronger economy, and −0.16 in 2022, when rising yields meant tighter policy. "Rates up, tech down" held in 2022 but was not a fixed rule.

From 2015 to 2019, against SPY, utilities (XLU) had a beta of about 0.35, consumer staples (XLP) 0.63, health care (XLV) 0.93 and technology (XLK) 1.20. Defensive sectors move less with the market on average, which is not the same as moving opposite to it.

The average pairwise correlation of nine sector funds was 0.54 in 2019 and 0.88 from February 20 to April 30, 2020. In a broad sell-off, positions that looked different in calm markets start moving as one. Size risk on the assumption that correlations jump toward 1 in a crash.

From 2015 to 2019, SPY's daily returns correlated 0.83 with EFA (developed international), 0.87 with IWM (small caps) and 0.93 with QQQ. Owning all four spreads money across names, but most days they rise and fall together. Diversification is measured in correlations, not in tickers.

A market order fills at the best prices available when it arrives, which can differ from the last trade or the quote on your screen. In a fast market, other orders can take the quoted shares first. The quote is a snapshot; your fill is a new event.

If only 300 shares are offered at $12.00, an order for 800 takes those and then the next best offers, perhaps at $12.02 and $12.05. The bigger the order relative to the shares available, the further it moves through the price levels. Thinly traded stocks are where this costs most.

It comes from the spread, from prices moving while the order travels, and from walking through price levels. Small per trade, it adds up for anyone who trades often. Backtests that assume fills at the last close leave it out, which is one reason live results trail tested ones.

The stop price is not a guaranteed exit price. In a fast decline the triggered market order fills wherever buyers are, which can be well below the stop. A stop limits how long you hold a losing position more reliably than it limits the price you get.

Holding 100 shares bought at $50 with a sell stop at $48 plans a $200 loss. If bad news sends the next open to $45, the stop triggers there and the loss is about $500. Overnight and event risk is not capped by a stop.

Once triggered it becomes a limit order, so a sell stop-limit with a stop of $25 and a limit of $24 can only sell at $24 or better. If the price drops straight through $24, nothing sells and the position keeps falling. You trade the risk of a bad price for the risk of no exit.

Some trigger a stop when a trade prints at the stop price, others when the bid or offer reaches it. The same order can therefore trigger at one firm and not at another. Find out the rule your broker uses before depending on a stop.

Set $2 below the market, it follows a stock from $40 to a peak of $46 with the stop at $44, then stays at $44 as the price falls. It still becomes a market order when hit, so the fill can be below $44. A short-term dip can trigger it during a longer advance.

A sudden dip in the first minutes can fill a stop near the low, after which the stock recovers to close near where it opened. The SEC's bulletin warns that such fills can be substantially worse than the closing price. Stop placement is a trade-off between noise and protection.

A day order expires at the close; a good-'til-canceled order stays until filled or canceled, within the broker's limit. Immediate-or-cancel fills what it can at once and cancels the rest. Fill-or-kill needs the whole order at once or nothing, and all-or-none needs the whole order but may wait.

A market-on-close order fills at the closing price, and any unfilled balance is canceled. They remove intraday timing decisions but accept whatever the opening or closing price turns out to be.

If each round trip costs 0.1% in spread and slippage, 250 round trips a year costs about 25 percentage points of the capital traded. A strategy needs an edge per trade larger than its cost per trade. That is why high-frequency rules that look strong before costs often vanish after.

Shares = (account × risk fraction) ÷ (entry − stop). With $20,000, 1% risk, entry $50 and stop $48, that is $200 ÷ $2 = 100 shares. A wider stop means fewer shares for the same planned loss, so the stop decides the size, not the other way around.

If a trade risked $200 and made $600, it returned 3R; a full stop-out is −1R. Expressing results in R separates the quality of the setup from how much money was put on it, and makes slippage visible as losses worse than −1R.

Expectancy = win rate × average win − loss rate × average loss. Winning 40% of the time with an average of 3R and losing 60% at 1R gives 0.4 × 3 − 0.6 × 1 = +0.6R per trade, before costs. Neither the win rate nor the payoff alone tells you whether a method makes money.

Winning 70% of trades at +1R and losing 30% at −3R gives 0.7 − 0.9 = −0.2R per trade. Methods that take small profits quickly and let losses run often show impressive win rates while losing money. In the Southeast Asian rule study, the profitable rules usually won fewer than half their trades.

With wins twice the size of losses, you break even winning one trade in three, since 1/3 × 2 − 2/3 × 1 = 0. With wins equal to losses you need over half. A planned ratio only counts if exits actually happen near the planned prices.

Recovering from −20% needs +25%, from −50% needs +100%, and from −80% needs +400%. The gain required is 1 ÷ (1 − loss) − 1. This asymmetry is why limiting the depth of losses matters more than the size of any single win.

SPY's dividend-adjusted price fell 33.7% from February 19 to March 23, 2020. A backtest's maximum drawdown is a floor on what to expect, not a ceiling, because the future can produce a worse one. The study scored strategies partly by their largest open drawdown for this reason.

Five trades each risking 1% in the same sector can lose close to 5% together in a sell-off, because correlations rise toward 1 when markets fall. Count risk by exposure, not by the number of tickets. A sector fund and three of its largest members are mostly the same bet.

For even-money bets won 55% of the time, Kelly says stake (1 × 0.55 − 0.45) ÷ 1 = 10% of capital. Real trading edges are estimated, not known, and overestimating the edge leads to overbetting, which lowers growth and deepens drawdowns. The formula is exact only for an edge you do not actually have.

A stop two ATRs away sits farther from the entry on a volatile stock and closer on a calm one, so noise is less likely to trigger it. Paired with fixed-fractional sizing, the volatile stock then gets fewer shares. The method sizes risk; it does not make the entry any better.

The broker lends part of the purchase price and charges interest, and the securities secure the loan. Gains and losses then fall on your smaller stake. Interest accrues whether the trade works or not, which raises the bar a leveraged trade has to clear.

Buy $10,000 of stock with $5,000 of your own and $5,000 borrowed. A 20% fall cuts the position by $2,000, which is 40% of your equity. The same arithmetic doubles a gain, but losses come with interest and the risk of a forced sale.

That limit is the initial margin. Firms may require more, and some securities cannot be bought on margin at all. Regulation T sets the most you can borrow when you buy, not how much equity you must keep afterward.

Equity is the market value of your securities minus the loan. If falling prices push equity below the maintenance requirement, the firm issues a margin call. Firms can raise their requirements at any time without advance notice.

The SEC's bulletin notes you are not entitled to choose which securities are sold or to an extension of time on a call. You can lose more than you deposited and still owe the firm. The worst-case loss on a margin account is not your initial deposit.

Borrow and sell at $40; if the price drops to $30, buy back and return the shares for a $10 gain, before costs. Short sellers also hedge long positions or supply shares when buyers rush in. The position is held in a margin account and is subject to margin rules.

A long position can lose at most what you paid; a short sold at $40 that rises to $50 has lost $10 a share, 25%, and the price can keep climbing. Margin calls may force you to buy back at the worst moment, locking in the loss.

If the borrowed stock pays a dividend, the short seller pays it. Hard-to-borrow stocks can carry high fees. These costs mean a short that goes nowhere still loses money over time.

Each forced purchase adds buying pressure, which pushes the price higher and forces more covering. Stocks with a large share of their float sold short and few shares to borrow are most exposed. The spike can have nothing to do with the company's prospects.

The seller then fails to deliver at settlement. Regulation SHO, in force since 2005, addresses persistent fails and abusive naked shorting. Not every fail is illegal, since market makers in thinly traded stocks can sometimes struggle to borrow, but shorting to manipulate a price is prohibited.

The definition covers options as well as stocks. Holding overnight is not a day trade; closing an opening position the same session is. Counting day trades is what the old pattern-day-trader rule was built on.

The day trades also had to be more than 6% of the account's trades in that period, and a broker could designate a customer it reasonably believed would day trade. The designation brought special margin requirements. This framework is being replaced; see the next concepts.

The $25,000 had to be in the account before day trading and kept there; below it, day trading stopped until the account was restored. Exceeding buying power triggered a day-trading margin call with five business days to meet it. Firms may still apply this during the transition.

The new standards took effect June 4, 2026, and firms may keep the old day-trading rules until October 20, 2027 while they migrate. Which framework applies to an account depends on its firm. Rules here change; confirm the current one with the firm before relying on a number.

Margin requirements are calculated from positions through the session, not only at the close. A trader who repeatedly runs deficits and fails to cover them can have margin trading frozen for 90 days or until the deficit is met. The minimum equity to trade on margin is $2,000, and firms can require more.

Settlement on T+1 means cash from a sale becomes settled funds one business day later. Frequent trading in a cash account therefore depends on tracking settled cash. Cash accounts avoid margin rules but bring their own violations.

It violates the Federal Reserve's Regulation T, and brokers respond with account restrictions. The trap is assuming a sale funds the purchase it closes; in a cash account the purchase must be paid for on its own.

Suppose you sell stock A, use that unsettled cash to buy stock B the same day, and sell B before A's sale settles: B was never paid for with settled money. Repeated violations lead brokers to restrict the account. Keep count of your settled cash instead of trusting the available balance on screen.

The pattern-day-trader definition was a floor, and firms used broader ones; maintenance margin and minimum equity work the same way. A strategy that fits the rulebook can still break a firm's house rules. The account agreement, not the rulebook alone, decides what you can do.

The holding period ends on the day you dispose of the asset. Nearly every trade in a short-term trading strategy produces short-term results. Long-term treatment requires holding past one year.

Long-term gains may get lower rates, but short-term gains are added to ordinary income. A trading strategy and a buy-and-hold strategy with the same pretax return can leave very different after-tax results. Compare strategies after tax as well as after costs.

Excess net capital loss is deductible against other income up to an annual limit, with the rest carried forward to later years. A large trading loss can therefore take years to use. Dollar limits are set in the tax code and can change.

The window covers 61 days around the sale. Frequent traders hit it easily by selling a losing stock and buying it back days later. Purchases by your spouse or by a corporation you control count too.

The loss is postponed rather than lost: it raises the basis of the replacement shares, so it reduces the gain or increases the loss when those are sold. For a trader, though, a string of wash sales can push losses into the next tax year.

Covering 42,211 investors and over 5.3 million trades from 2006 to 2016, the study estimated negative abnormal returns of 4% to 4.4% a year on a gross basis. After transaction costs the shortfall grew. It is one market, so it shows a pattern to check elsewhere, not a universal law.

In the Colombian data, the investors who traded most had worse gross returns than those who traded less, and costs widened the gap. Trading more increases what you pay and adds decisions, without adding information.

Individual investors tended to buy and sell at unfavorable moments, at least until the collapse of the country's largest brokerage in 2012. Bad timing is a real cost even for investors who pick reasonable assets, because when you buy and sell decides what return you actually earn.

The study's last finding complicates the simple story: experience and activity together, once trade counts were accounted for, were associated with better results. That describes those investors relative to others, not a claim that most active traders profit.

The figure covers investors who began day trading on the Brazilian market from 2013 to 2015, and the chance of profit fell the longer people persisted. Persistence did not convert losers into winners. Only a minority of day traders profit over time.

Many never reach consistent profits. The SEC's advice is to risk only money you can afford to lose, never money needed for living expenses, retirement or debt. The losing period is the usual case, not bad luck.

The traders who lost money and quit rarely post their results. A feed full of winners is consistent with most participants losing. Before treating any example as typical, ask for the base rate: what fraction of all the people who tried ended up with that result.

The SEC's day-trading guidance tells traders to know in advance how much they must make just to break even. A strategy that is slightly profitable before costs can lose money after them, and many do.

Both involve rapid decisions, uncertain outcomes and fast feedback in money. Researchers describe problematic trading that resembles gambling disorder, with similar psychiatric patterns and diagnostic features. Trading is not purely chance, but the overlap in how it feels matters for how people behave.

In a case series of excessive traders, a common pattern was small early wins, then chasing losses, poor risk assessment and an illusion of control. Each loss raises the urge to recover it, which raises risk exactly when judgment is worst. A pre-set daily loss limit is one way traders cap it.

Screens full of indicators and fast order buttons make trading feel like skill, even when results over short periods are mostly noise. Early wins can come from luck and still feel earned. Judge skill over many trades with costs counted, not by a good week.

Research describes thrill-seeking and sensation-seeking, similar to craving, as a motive for some traders. When the activity itself is the reward, people keep trading after the numbers say stop. Asking whether you would still do it without the excitement separates a method from a habit.

In a study of 9,245 adults, day traders reported gambling far more than non-traders, and 7.6% of day traders met the bar for problem gambling. That does not make every trader a gambler, but the overlap is large enough to take seriously.

Selling a winner feels like locking in success, and selling a loser feels like admitting a mistake, so people do the first eagerly and put off the second. Trading rules with set exits can counter it, which is one reason researchers think rules help individuals.

An experiment with 193 professional traders found they leaned toward the disposition effect for assets that revert to a mean, which was the profit-maximizing choice there. A simple prompt that made the bias visible changed their decisions, helping on trending assets and hurting on reverting ones. Context decides whether a habit is a bias.

Researchers point to smartphone trading platforms and investing content on social media as drivers of growth in individual trading. Easy access lowers the friction that once slowed impulsive trades. Fewer steps between an urge and an order means rules have to do the slowing.

In the heat of a trade, losses loom larger and hope stretches stops. A plan written in advance, with the stop and size fixed, turns later choices into checks against the plan. It creates no edge of its own; it only guards any edge the method already has.

Trading always carries risk, so a legitimate firm cannot promise you a profit. Fraudsters use sophisticated persuasion techniques to make a pitch seem real. The more a claim sounds too good to be true, the more checking it deserves.

A promoter's lack of registration is a reason to stop and investigate. The SEC's free Check Out Your Investment Professional tool on Investor.gov confirms registration and shows disciplinary history. Unregistered platforms are where many trading scams operate.

Regulators have warned about unregistered platforms promoting AI systems with promises of guaranteed winners. Even registered firms deserve doubt when they claim AI guarantees strong returns. Software can automate a strategy; it cannot remove the risk of the markets it trades.

The hype travels through social media, newsletters, ads and chat rooms. Once the insiders sell and the promotion stops, the price usually collapses and late buyers lose. Microcap stocks are an easy target because so little public information is available on them.

Social media is saturated with financial content, and a promoter may be paid to recommend. A follower count is not a track record. Ask whether the person is registered and whether they are compensated for what they promote.

The SEC warns that day-trading newsletters, seminars and books may not be objective, and that some sources are paid for their stock picks. Ask whether the seller profits when you trade, and whether they have audited results, before paying for an edge.

Scams include deepfake audio of a family member pleading for money to be moved, and imitations of real advisers or regulators. Verify any urgent request through contact details you look up independently, not the ones in the message.

A chatbot can produce fluent analysis built on false or stale news, and conversations with one can push impulsive decisions. Treat AI output as a lead to verify against primary sources, never as a reason to trade.

The SEC takes tips online at sec.gov/tcr, FINRA through its complaint form, and each state has a securities regulator. Reporting helps regulators spot schemes that target many people at once.