Technical analysis in portfolio management
The answer
Inside a professional portfolio, technical analysis is rarely used to choose what to own. It is used for 4 other things: deciding the overall exposure to risk, setting the size of each position, defining the exit before entry, and executing orders without moving the price. Those 4 uses have far better support than security selection does, and they are the ones almost never taught.
Why this costs you money
Retail effort is allocated almost entirely to the 1 question that matters least.
Ask somebody what they are working on and the answer is what to buy. Ask them what percentage of their account is at risk right now, across every open position, and most cannot answer without a calculator and 10 minutes. Ask them what would make them reduce total exposure rather than exit a single position and most have no rule at all.
That allocation of attention produces a specific and repeated outcome. A portfolio built from good individual decisions fails as a group, because every position responds to the same variable and nobody measured that. Eight holdings, each entered with a sensible reason, each sized modestly, all of them sensitive to the same interest rate or the same sector or the same source of foreign flows. When the variable moves, they move together, and the person discovers their real position size for the first time on the day it hurts.
The second cost is the absence of an exit. A method for choosing what to own, without a matching method for deciding when the reason has stopped being true, produces portfolios that only grow. Losers are held because selling admits an error, winners are trimmed because a profit feels like it should be taken, and the portfolio slowly fills with the positions that did not work. Technical analysis is unusually good at supplying the missing half, because a price level is a fact and does not require you to have an opinion about the company.
How it works
Where technicals sit in a professional process
A professional investment process has 5 jobs. Technical analysis is used seriously in 4 of them and lightly in the fifth.
| Job | What decides it | Role of technicals |
|---|---|---|
| What to own | Research, valuation, business quality | Small. Mostly a ranking filter. |
| How much risk in total | Regime, volatility, drawdown state | Large |
| How much of each | Volatility and stop distance | Large |
| When to exit | A price level fixed before entry | Large |
| How to execute | Liquidity, spread, participation rate | Large |
Notice the pattern. The uses that survive are the ones where the input is observable and the output is a number. The use that does not survive is the one where the input is a shape and the output is an opinion.
Exposure: the regime question
The first portfolio-level use is deciding how much risk to carry at all.
The simplest version is a long-term trend filter. Hold the asset while its price is above a long moving average, and hold cash or short-term bonds while it is below. Meb Faber published a version of this in the Journal of Wealth Management in 2007, applying a 10-month moving average rule across 5 asset classes. The result was not a large improvement in return. It was a large reduction in the depth of the worst declines.
That distinction is the whole point and it is worth stating clearly. A trend filter is not a way to make more money. It is a way to be absent during the periods that end accounts. It pays for that by underperforming during rising markets, because it exits after a fall has already begun and re-enters after a recovery has already begun.
The cost is real and it is not small. A trend filter produces many false exits. In a market that falls 12% and recovers, you sell near the bottom and buy back higher, and you do it repeatedly. Anybody presenting a trend overlay without showing how often that happens is not describing the instrument.
Sizing: volatility as the input
The second use is deciding position size from measurable volatility rather than from confidence.
The standard technical input is Average True Range, which measures the typical size of a daily move including gaps. A position sized so that a move of 2 or 3 times the recent range equals a fixed percentage of the account will automatically be smaller in volatile instruments and larger in calm ones.
There is a broader version of this idea at portfolio level, called volatility targeting: scale total exposure so that the portfolio's expected volatility stays near a chosen number. Alan Moreira and Tyler Muir published evidence in the Journal of Finance in 2017 that scaling exposure inversely to recent volatility improved risk-adjusted returns across several asset classes. The mechanism is that volatility is persistent — a volatile month is usually followed by a volatile month — while returns are not, so you can forecast the risk without forecasting the return.
Exits: the part fundamental processes lack
The third use is the one where technicals contribute the most and receive the least credit.
A fundamental process produces a view about a business. Views about businesses are slow to change, and they should be, because a company does not become a different company in a week. That is a strength for entry and a weakness for exit, because it means nothing in the process ever says stop.
A price level does the job that the research process cannot. It is defined before entry, it is unambiguous, and it does not require anybody to admit the analysis was wrong. The position is closed because a condition set in advance was met. That is why many institutions with entirely fundamental selection processes still run technical exit rules over the top.
Execution: the least glamorous and most measurable use
The fourth use is getting in and out without paying more than necessary.
A large order cannot be filled at once without moving the price. Volume profile, volume-weighted average price and intraday liquidity patterns exist to answer the question of when and how fast to trade. This is the one area of technical analysis where the results are measured precisely, every day, by every institution, against a benchmark that everybody agrees on.
For a retail reader the same principle applies at a smaller scale. Divide your intended position by the average daily volume. If it is above about 5%, your own order is part of the price you will get.
The one thing to understand about combining signals
Adding more technical signals usually does not add diversification. It adds correlation.
A trend filter on 10 assets is not 10 decisions. In a broad market decline every one of those filters turns negative within a few weeks of the others, because they are all reading the same market. A portfolio of trend signals is a single bet on trends persisting, and it will have long, deep periods of poor performance when they do not.
The honest way to state this: diversify across mechanisms, not across indicators. Two indicators reading the same price series are 1 input.
What it tells you, and what it does not
Portfolio-level technicals tell you your current total exposure and what will move it. That is a measurement, it is available today, and most people have never made it.
They tell you when a portfolio's positions have stopped being independent, which is visible when many positions cross their own trend filters within days of each other.
They do not tell you what to own. The evidence for chart-based security selection is weak, and this article does not improve it.
They do not protect against a gap. An exit rule fixed at a price executes at the next available price, and in a large overnight move that can be far away.
They do not remove the need for a decision about the total size of the account's risk. A trend overlay reduces the depth of declines and it does not make a badly sized portfolio safe.
And they carry a specific cost that is rarely disclosed. Every rule that reduces exposure also reduces participation. Over a long rising period, a portfolio with a trend overlay will usually end behind a portfolio without one. What it buys is a shallower worst case, and whether that is a good trade depends entirely on whether you would have stayed invested through the worst case without it.
The decision rule
Decide exposure first, size second, exit third, and selection last.
If the broad market is above its long-term trend filter and your portfolio's total open risk is within your limit, then new positions are permitted at full size.
If the broad market is below its long-term trend filter, take fewer positions and take them smaller. Relative strength in a falling market identifies what is falling least, which is information and is not a reason to buy.
Unless your positions all share a driver. In that case the number of positions is not the number of bets, and the limit must be applied to the driver, not to the position.
Every position gets an exit price written down before entry. A position without one is not a position, it is an opinion with money attached.
Try this now
Ten minutes, on your own holdings. It produces 2 numbers that most investors have never seen for their own account.
- Open your holdings list. Write down each holding and its current value. Include cash as a line.
- For each holding, open a daily chart and add a 200-day simple moving average. Write down whether the price is above or below it. This takes about 20 seconds per holding.
- Add up the value of everything above its 200-day average, and the value of everything below. Divide each by your total portfolio value including cash. Number 1 is the percentage of your money in holdings that are above their own long-term trend.
- Now open a chart of your broad index with the same 200-day average. Write down whether the index is above or below.
- Next, group your holdings by what would hurt them at once. Not by sector name — by driver. Use these groups: rate-sensitive, dependent on a single commodity, dependent on foreign revenue, dependent on domestic consumption, small and illiquid. A holding can be in 2 groups.
- Total the value in each group and divide by the portfolio value. Number 2 is the largest of those percentages.
What you should see. Number 1 is usually either very high or very low, and rarely in the middle. That is the finding. Your holdings are not independent. They are largely on the same side of their trends at the same time, because they are all reading the same market.
Compare number 1 with the index in step 4. If the index is below its own 200-day average and 80% of your money is in holdings above theirs, you are positioned against the market and should be able to say why in 1 sentence. If both are below and you are fully invested, you have made a decision by not making one.
Number 2 is the more uncomfortable one. For most portfolios the largest driver group covers 40% to 70% of the money. That number, not the count of holdings, is your real position size, and it is the number to compare against your risk limit.
Write both numbers down with today's date. Repeat the exercise in 3 months. The change tells you whether your portfolio is drifting toward concentration, which is what portfolios do when nobody measures them.
Three real cases
1. Meb Faber, Journal of Wealth Management, 2007 — what a trend overlay actually delivers Faber tested a simple rule: hold an asset class when its price is above its 10-month simple moving average, and hold cash otherwise, applied to a set of 5 asset classes and rebalanced monthly. He tested it over a long historical period and published the rule before the following decade.
The result was a modest change in return and a large reduction in the depth of the worst drawdowns. That shape — similar return, much shallower declines — is the characteristic result of trend overlays and it has been reproduced in many later studies.
The case matters because of what happened next. The paper was published in 2007, and the rule's behaviour in the years afterwards is genuinely out of sample. It performed well in 2008 and poorly during several later years when markets rose with frequent shallow interruptions. Both facts belong in the same sentence, and only the first one appears in most presentations.
2. Managed futures in 2008 — a diversifier that worked when it was needed 2008 was among the worst years on record for global equities. Trend-following managed futures funds, as a group, had a strong positive year. Industry indices tracking this group, including those published by Societe Generale, recorded double-digit positive returns for 2008 while equity indices fell sharply.
The mechanism is straightforward. A trend-following programme holds positions in the direction of established moves across many markets. In 2008 the established moves were falling equities, falling commodities and rising bond prices, and the programmes were positioned that way for months.
This is the strongest documented case for technicals at portfolio level. It is also the case most often misused, because a single good year is not evidence of a persistent property. The relevant claim is not "trend following works". It is that trend following has a different return pattern from equities, which is what a diversifier is supposed to have.
3. Trend following in 2022, and in the decade before it — read both halves or neither In 2022 both equities and bonds fell substantially. Trend-following funds had a strong year, with industry trend indices recording large positive returns.
Now the other half, which is compulsory. Through much of the period from approximately 2011 to 2019, trend-following returns were poor, and many investors who had allocated after 2008 gave up before 2022 arrived.
The lesson is not about trend following. It is about what a diversifier costs. An approach that pays in the worst years must underperform in the ordinary ones, because if it paid in both, everybody would own it and it would stop paying in either. Anybody who shows you 2008 and 2022 without showing you 2011 to 2019 is showing you 2 numbers out of 15.
The question that resolves it
A novice asks: what should I buy next?
An expert asks: what is my total exposure right now, and what single event would move all of it at once?
The first question is asked continuously and answered easily. The second is asked rarely and takes 10 minutes to answer. Every serious portfolio failure in the case studies of this wiki — Barings, Long-Term Capital Management, Amaranth, Archegos — is a failure to answer the second question, in an institution where dozens of people were answering the first one all day.
What would make this wrong
The claim is that technicals are useful for exposure, sizing, exits and execution, and weak for selection.
The exposure claim would be wrong if trend filters stopped reducing drawdown depth. This is testable and it is being tested continuously in live products. The result is not guaranteed. Trend rules have long periods of poor behaviour, particularly in markets that fall and recover quickly, and there is a reasonable argument that faster policy responses in recent decades have made those V-shaped recoveries more common. If that continues, trend overlays will subtract value for long stretches, and the honest position would change.
The sizing claim rests on volatility being forecastable. It is, and that is one of the better-established results in finance. Volatility clusters. If it stopped clustering, volatility-based sizing would lose its justification.
The exit claim is the hardest to test and the easiest to defend logically. A process with no exit condition cannot terminate a losing position except by opinion. That is not a statistical claim, it is a structural one.
And the honest limit that cuts against this article. All of these portfolio-level techniques were developed and tested mostly on liquid futures and large-capitalisation equities, rebalanced mechanically, with institutional costs. A retail reader running them on a handful of positions with retail costs and retail taxes is not running the tested strategy. The direction of the evidence still applies. The magnitude does not.
In India
Four Indian features change how portfolio-level technicals are applied.
Exits are a weaker promise. Individual Indian shares have daily price bands. A share at its lower band may have no buyers, so a stop order does not execute because no trade occurs. That does not make exit rules useless. It makes position size the primary protection rather than the secondary one, because size is the only control that works when the exit does not.
Lot sizes prevent precise sizing in derivatives. Futures and options trade in fixed lot sizes set by the exchange. The size your risk rule calls for often does not exist. If 1 lot risks more than your limit, the correct action is not to take the position, and that instruction has to be written down because in the moment it will not feel like an answer.
The index is concentrated, so index-based overlays are narrower than they look. A trend filter on the Nifty 50 is substantially a trend filter on a small number of very large companies, weighted heavily toward financial services. It is still useful. It is measuring a narrower thing than an S&P 500 filter measures.
Managing other people's money has explicit thresholds. Discretionary portfolio management for clients is regulated under the SEBI (Portfolio Managers) Regulations, 2020, with a minimum investment amount per client. Advising clients for a fee generally requires registration as an Investment Adviser. Alternative Investment Funds are regulated separately in 3 categories. These are not technical details for a retail reader. They decide whether a person offering to manage your money is permitted to do so.
Two more practical points. Short-term and long-term capital gains on listed equity are taxed at different rates, with the boundary at a holding period of more than 12 months. A rules-based overlay with frequent exits pays that difference repeatedly. And a share placed under the Additional Surveillance Measure or Graded Surveillance Measure framework can be moved to 100% margin or trade-for-trade settlement, which changes its liquidity for administrative reasons.
In the United States
The United States has the deepest set of instruments for applying these ideas and a different set of frictions.
Everything is purchasable at a fine grain. Fractional shares mean a sizing formula's answer can always be bought exactly. Sector, factor, bond, commodity and volatility exposures are all available as liquid, low-cost funds. A tactical allocation conclusion can be executed the same afternoon.
Halts are designed to reopen. Individual American shares are covered by volatility pause mechanisms that stop trading briefly and then resume with wider bands, rather than locking for the day. An exit is therefore more likely to execute, though not at a guaranteed price.
Leverage is standardised. Regulation T permits an initial margin level for share purchases, and brokers apply maintenance requirements above regulatory minimums. Portfolio margin is available to qualifying accounts. Standardised leverage makes portfolio risk easier to calculate and easier to take too much of.
Tax interacts with exit rules directly. The wash sale rule disallows a loss for tax purposes if a substantially identical security is repurchased within 30 days before or after the sale. A trend overlay that exits and re-enters within a month runs into this repeatedly, which is a reason many American implementations use funds rather than individual shares.
Managing money is regulated by size and by state. An adviser generally registers with the Securities and Exchange Commission above an assets under management threshold and with a state below it. The fiduciary standard under the Investment Advisers Act of 1940 applies to registered advisers.
Where they differ, and what that tells you
Three differences, and each changes the method rather than describing it.
The order of protections is reversed. In the United States, an exit rule is the primary protection and position size is the backup, because halts reopen and fractional shares let you size exactly. In India, position size is the primary protection and the exit rule is the backup, because a share at its band cannot be sold at any price. That is not a difference in emphasis. It changes which control you must get right first, and Indian readers who learn the American ordering will protect themselves in the wrong order.
A sector view is an instrument in one market and a company decision in the other. An American tactical allocation ends in a fund purchase and takes no company risk. An Indian one usually ends with choosing companies inside a sector, because sector funds are limited and often thin. The Indian version therefore carries a risk the analysis never examined, and needs an extra check that the American version does not.
Rebalancing costs different amounts. American exchange traded funds and fractional shares make frequent, small adjustments cheap. Indian adjustments pay securities transaction tax on the sell side, face a wider short-term capital gains rate, and often face wider spreads outside the largest companies. That means the optimal Indian rebalancing frequency is lower than the optimal American one, and copying an American rebalancing schedule transfers a cost that was never in the original calculation.
Carry this
- Technicals do 4 jobs well at portfolio level: exposure, sizing, exits and execution. Selection is not one of them.
- A trend overlay does not raise returns. It reduces the depth of the worst declines, and it pays for that with false exits.
- Two indicators reading the same price series are 1 input, not 2. Diversify across mechanisms.
- The number of positions is not the number of bets. Group by driver and total the largest group.
- Every position gets an exit price before entry. In India, size matters more than the exit, because a share at its band cannot be sold.