Is Technical Analysis Bogus?

Pragmatic Observations on Randomness, Efficiency, and Predictability

Watch What You Trade

No matter what you trade, price is a crucial variable. Without it, you’re flying blind. Imagine being a realtor and ignoring the prices of houses in your neighborhood when determining your estimate. Imagine being a gas station that sets its price without considering the prices of competing stations. Ignore the price of the product you trade at your own risk. Drop the hubris and the desire to be right; you can benefit from observing and respecting the market.

Price Is Sentiment

Stock prices fluctuate based on investor sentiment. Technical analysis is the study of price action, and it offers pragmatic insights about investor preferences and fund flows. By looking at new highs (new lows), we can see which industries and themes investors are betting on (against). By looking at trends and prior volatility, we can observe momentum and mean reversion. Finally, certain price signals, like days with large price drops with record volume, can indicate a severe shift in sentiment that is worth respecting.

“If there’s a material change in price behavior… regardless of what the fundamentals are saying, it’s time to exit. The big players know/suspect something is wrong.” — Mark Minervini

Market sentiment is smart. Bond prices predicted all ten U.S. recession since 1955, with only one false signal. From 1802 to 2014, stock drops of 8% or more successfully predicted an imminent recession 65% of the time. Not surprisingly, many of the greatest traders of all time take price action seriously.

Stocks Aren’t Random

Critics claim that stock prices are random. This is false; stocks trade like items at an auction, and their prices fluctuate according to supply and demand for the stock. If stock prices are random, then investors’ buying and selling activity must be random too.

This is obviously untrue; millions of investors do not buy and sell securities randomly. They buy or sell based on their sentiment, which fluctuates in response to events as they unfold. Price drops indicate that investors’ sentiment about the company (and its stock) is souring. Price rises indicate that investors are feeling more optimistic. Positive headlines, like earnings surprises, are statistically likely to cause prices to rise more strongly than usual in the coming weeks, an effect known as “post-earnings announcement drift” (PEAD). Sentiment, and thus stock prices, is inherently trendy, though some periods are trendier than others.

Just because something resembles a random process does not mean that it actually is random or unpredictable. This flawed argument was popularized by Burton Malkiel in his poorly-reasoned book, “A Random Walk Down Wall Street.” He showed that stocks resemble randomly generated lines, and he concluded that they are random and not worth trying to forecast.

Even if stock prices were random, that would actually support the notion that prediction is possible. Random distributions have known characteristics (mean, standard deviation), tendencies, and probabilities. If a variable (ex: monthly returns) has a normal distribution, it has a 68% chance of falling within one standard deviation from its mean. If it is three standard deviations above its mean, there is a strong chance it will be closer to the mean (lower) in the future. Thus, we can create a probabilistic forecast that will be accurate on average.

Luckily, we can falsify the idea that stocks are unpredictable; statistical distributions confirm that price-based indicators can lead to a significantly different distribution of future returns compared to the population.

Some Indicators Have an Edge (Preview)

After hearing pundits mock the process of studying and predicting price action, I decided to try it myself. I used a statistical approach that incorporated reverse factor modeling and conditional distributions to find price-based metrics with predictive power. My goal was to find indicators associated with significantly elevated levels of future 20-day returns in the S&P 500 index.

I started with reverse factor modeling, comparing the distribution of a particular metric in the population overall to its distribution in the “best buy days” (days with high future 20-day returns). This technique is useful for discovering potentially valuable input variables. For example, my research found that “best buy days” appeared to have a noticeably different distribution of trailing volatility compared to the general population. This inspired me to consider trailing volatility as a key variable in my predictive indicator.

I also found that “best buy days” had a unique distribution of another metric compared to the population; 50Hdist, the % distance between the current stock price and its high over the past 50-days.

After trial and error, I eventually combined these variables and created a hybrid indicator with statistically different (and higher) future returns than a buy and hold strategy, with fairly consistent outperformance. The graph below shows the distribution of future returns for the hybrid indicator portfolio and compares it to the distribution for a buy and hold portfolio from 1928–2020.

This indicator improves average returns by 20% (on a relative basis) per year and is significant to a p-value below 0.01. These are strong results. The next step next step is to assess the persistence of outperformance, which requires testing its performance in each of the past nine decades and comparing it to a buy and hold strategy. It was remarkably consistent, winning in 8/9 test decades.

Another hybrid indicator I constructed, based on autocorrelation (momentum), had similarly strong results and performed even better in the 2010–2020 decade. I look forward to testing combinations of these indicators in the future. Clearly, both of them have an edge.

Developing these indicators took considerable research and effort and I consider them proprietary. However, I am passionate about education and helping aspiring traders acquire an edge. That’s why I share these indicators’ status daily, along with other interesting trading observations, for serious investors.

Interested? Check out Stock Savvy. It gives you the status of these quantitative indicators, and it highlights key trends and attractive trade ideas that I discover during my research process.

Prices are Informed

Market preferences are smarter than you think. Once the pandemic hit, industries that benefit in a pandemic environment have been the best performers, like online retail, leisure, packaged foods, medical/diagnostics, and defensive areas like precious metals and grocery stores. Industries most hurt by the pandemic, like travel and commercial real estate, were the worst performers.

Yet during this time, many billionaires, news outlets, and neighborhood pundits bashed the market for being “disconnected” from reality. Had they dug deeper and analyzed performance by industry, or by balance sheet health, they would have seen that the market was rewarding digital, defensive, cash-rich, and flexible companies. Stocks making new highs have been trending massively.

Folks complaining about disconnect betrayed their ignorance of which industries and themes were popular.

All investors had to do was watch prices and note which stocks were making new highs/lows, as well as stocks experiencing unusually high volume. But many investors focus on popular market indexes, like the S&P 500, which have become so skewed toward large companies that they are actively misleading. Cap-weighting has made them poor proxies for the market as a whole. Because of this, investors now must dig deeper and analyze internals, focusing on the industries, asset classes, themes, and companies that are performing the best and worst.

Prices are smart because investors are forward-looking, buying and selling based on their expectations for the future value of the company and its stock. Years ago, when Netflix IPO’d, Blockbuster stock immediately topped and began to break down violently; leadership constantly changes and investors often anticipate events and smell the roses before the headlines drop.

Market sentiment is so valuable that the Conference Board, a widely respected economic think tank, uses stock prices as a component in their index of leading economic indicators. Many of the greatest traders/investors of all time take price action seriously.

• Stanley Druckenmiller ran a hedge fund for 30 years, delivering a 30% average annual return without a single down year. He is adamant about the value of price action: “I never use valuation to time the market… The catalyst is liquidity, and hopefully my technical analysis will pick it up.”
• Mark Minervini delivered a 35,000% return within five years and won the 1997 U.S. Investing Championship. He has a deep respect for market sentiment: “Ultimately, opinions mean nothing compared with the wisdom and verdict of the market. Let the strength of the market, not your personal opinion, tell you where to put your money.”
• Scott Ramsay, a laid-back futures trader, achieved 11 years of positive returns averaging at 17.2%, and despite trading a leveraged instrument, achieved remarkably low volatility (12%): “I use the fundamentals to have a directional bias, and I use the technicals to confirm that bias.”

Even the greats struggle to avoid doubting the market. Druckenmiller admits that he incorrectly underestimated the trend and momentum in U.S. stocks once the pandemic hit.

A company’s stock price also reveals what investors in the market are willing to pay to own the company by buying all of its shares. This valuation is also known as the company’s “market capitalization,” which tells you how much the market thinks the company is worth. When the company’s stock price changes, its market cap changes too. This metric is surprisingly efficient based on available information. Aswath Damodaran, NYU Stern’s Chair in Finance Education, stresses the importance of humility and considering the consensus view of the market.

“What I have learnt in 35 years of investing is you can disagree with markets but you have to respect them… do not project your feelings onto the market.” — Aswath Damodaran

Technical Analysis Eliminates Feelings/Biases

One benefit of observing market prices is that it forces you to separate your own personal views/feelings and focus on what the market consensus is saying. For instance, I recently noticed increased market optimism toward Swedish stocks, with EWD breaking out into a new 52-week high. Even though I had macroeconomic and pandemic-related concerns, I ultimately decided it was a bullish price signal.

Damodaran cautions investors about ideological beliefs that lack evidence, so-called fairy tales, especially ones that claim the market is stupid. Hubris causes investors to underestimate the wisdom of the market. Humans tend to assume they are smarter than their peers yet are most likely to be average (by definition).

Love him or hate him, Jeff Bezos made a good point back in 2016 — reality is messy, so if you have rigid and constant beliefs, you are probably underestimating the complexity of the world. He has also expressed concerns about social media destroying nuance and spreading ideology. Rigid views based on hopes or dogma are an impediment to trading and forecasting; reality doesn’t care about your feelings. Predictions based on emotions and personal feelings are less likely to succeed compared to those built on probability, evidence, and logic.

Exploiting Patterns Requires a Statistical Approach

Critics are right to be skeptical of proposed methods to predict future returns. Without evidence, there’s no reason to expect success. We should be cynical toward forecasts of active money managers since the majority of them fail to outperform the overall stock market.

Backtests are misleading because early periods of outperformance can make results appear far better than they actually are due to compounding. A statistical approach is the best way to test whether an indicator has predictive power, and to assess the significance and persistence of its power.

For a signal to be valid, it must have a distribution of future returns that’s statistically different from the population. When converted into a buy/sell indicator, it should outperform in almost every single time period, ensuring that the predictive power isn’t entirely concentrated in one brief period.

Statistical methods allow investors to remove their assumptions and test hypotheses individually. For instance, many investors assume that price action is fundamentally different now than it was 50 years ago; we can test this hypothesis by simulating a strategy that did well back then and see how it does now. Surprisingly, some momentum-based indicators have impressive and consistent performance from 1928 to today.

Many strategies succeed longer than people expect. Investor behavior is less sophisticated and evolved than investors like to believe. Certain effects, like periods of suppressed volatility leading to periods of heightened volatility, persist over decades and likely stem from behavioral tendencies or hedging activities that are hard to measure.

3 Simple Practices Worth Following

Not everyone can take the time required for statistical forecasting. But several technical analysis practices take less than 15 minutes and still offer considerable insight into key investing trends and potential shifts.

1: Forecasting principles state that in the absence of information, we should use a naive forecast and assume that current behavior and trends will continue. This means we must watch monitor trends and watch for key shifts. We should be aware of which assets are making new highs/lows, which are breaking above/below their trend lines, and which have moved the most after the most recent peak or trough.

2: Investors can find interesting opportunities by looking for high-volume breakouts from flat, narrow ranges or periods of suppressed volatility. It’s especially noteworthy if momentum is strong and the breakout coincides with a new high or low. Those characteristics indicate that investors are drastically changing their outlook on the asset. It’s probably worth betting on that sentiment shift continuing, even when you don’t know the cause.

3: New highs are bullish, and new lows are bearish. Avoid betting against stocks that are hitting all-time highs with strong momentum (think mid-2020 TSLA); trends can last far longer than expected. U.S. stocks tend to rise over time. Countertrend trades should require at least some early signs of a sentiment shift, such as a high volume break below a moving average or the bottom of a long prior trading range, to compensate for their lower odds of success.

“Trades don’t have to start based on fundamentals. If you wait until you can find out the reason for the price move (technicals), it can be too late.” — Colm O’Shea