Choosing the right time horizon is of essential importance to find a viable edge for a trading strategy.
Many short-term traders tend to show an unhealthy obsession trying to optimize trade entries.
Long-term investors, on the other hand, often enter in an arbitrary fashion and could improve results by using broad entry criteria.
Exact entry points shouldn’t matter all that much as long as the general principle a strategy is based on is sound. If that is not the case and the strategy depends on an optimal entry, it is probably over-optimized and very unlikely to work in the future.
Let´s look at a generic trend strategy
which focuses on a medium- to long-term time frame.
The basic concepts of absolute momentum (trend) and relative momentum have been studied extensively by academics and used by traders successfully for many decades. Let´s accept for now, that these findings are evidence of a solid general principle at work in all asset classes which leads to an outperformance over time. The studies show that trends have a high probability to continue over time periods between 3 to 15 months.
Because the concept holds over such a wide time range and because there is such a large amount of random movement in the financial markets in general, it should make virtually no difference how you determine your entry points when following trends – actual differences in future outcomes will be due to unpredictable factors: pure luck.
When we look at backtests for different entry methodologies (eg. breakouts, moving average cross-overs etc.) using time periods between 3 and 12 months and at the actual long-term results of trend-following CTAs (for example here
) very similar risk-adjusted returns are exactly what we find. This parameter stability is an essential quality of a good trading system.
Looking more closely it is interesting to observe that the results (in terms of risk-adjusted returns) generally get better as we move from shorter parameter sets towards the longer term around 12 months after which returns slowly start to deteriorate.
This effect has gotten stronger in recent decades and, I think, it leads to an important principle.
The time horizon of an investing / trading strategy – the time frame it is based on – is an essential factor. In fact, even the unfortunate categorization between traders and investors is based primarily on strategy time frame. This distinction is counterproductive however, because it creates an artificial separation – often you either consider yourself a short-term trader or a long-term investor. From a portfolio perspective it is more productive to integrate good strategies from both the trading and the investing realm in the quest for superior risk-adjusted return, because correlations tend to be smaller.
In practice the time horizon has a strong, but gradual, influence: the shorter the time frame, the weaker the influence of the underlying principle of the strategy becomes. Exact timing matters more and random noise increasingly distorts the picture. To figure out, if there is sufficient strength remaining in the concept that gives your strategy an edge, is what really matters.
Very often descriptions of entry methods and whole trading systems come with the statement: “this works across all time-frames” – in my opinion that is usually false and I would need a lot of statistical evidence to accept such a statement.
The reason to bother with shorter time frames at all is the law of active management. It implies that, the more frequently you are able to bet on your edge, the higher your risk-adjusted return will be. Conversely the longer the time frame, the more reliable and strong the influence of the underlying principle and therefore your edge is.
We need to strike a balance.
Analyzing this thesis from the broad concept and then narrowing it down to the strategy level yields some interesting insights and tools for the development of trading strategies.
The very basic return drivers in the market are derived from the risk premia of different assets. If an investor could not expect to earn a premium for the risk he takes when investing his money over the long run, nobody would invest in the first place. The ultra-long uptrend of the stock market, the black line in the S&P 500 chart below, for example, is the equity risk premium that buy and hold investors harvest. It is aligned with economic growth – as long as the capitalist system doesn’t fall apart, it should continue to exist.
A majority of market participants exclusively rely on passively earning the long-term risk premia of different assets. A better benchmark for buy and hold investing than a market cap weighted US equity index (the S&P 500) would be an equal weighted portfolio of all global assets. This combination of all asset´s risk premia yields better risk-adjusted returns, specifically lower volatility, than stocks alone through the magic of diversification.
Zero Sum Game
The existence of this underlying principle has to be taken into consideration when developing a strategy as it is very powerful – profitable short equity strategies, for example, are very rare and difficult to build, because they fight against the strength of the equity risk premium.
But it also has another important implication: Risk premia returns are the total wealth created by investing in productive assets. Every outperforming strategy is bound to be difficult as it essentially attempts to create wealth out of thin air. Underperformers and outperformers must be equal (minus cost) in such a zero sum game.
Strength of the Equity Risk Premium
The table below illustrates how powerful risk premia are. Even though the equity risk premium´s reliability deteriorates with shorter time frames, it still gives a distinct edge even on a daily basis: the probability to have a positive return on any random day is 53,7%.
Other Forces in the Market
To make sustainable outperformance possible other recurring, predictable phenomena must be at play. If risk premia were a reliable constant, the blue S&P 500 price line in the chart above would hug the black line very closely, because investors would buy or sell as soon as prices deviated even a little bit from fair value causing them to mean-revert. These return drivers must work over a shorter time horizon causing time-varying risk premia (very different actual returns at different times) – an example would be changing levels of economic growth over time, but also human behavior causing fluctuations in valuation.
This is the realm of active investing strategies – the ones that work best are mainly based on the return drivers momentum (absolute and relative momentum strategies) and mean reversion (for example value or carry strategies)
. Other return factors, which have shown up in academic literature, might have some value, but are not as clear cut. Examples are illiquidity, size, low beta and quality. These concepts can be combined or used as standalone strategies, which have a high probability of long-term outperformance over buy and hold investments, if they are implemented with the necessary discipline.
A long-term investor could improve his results by systematically integrating these findings in his approach to create better than random entries. He could, for example, increase asset exposure in an uptrend and when valuation measures, e.g. the CAPE ratio for equities, indicate cheap asset prices and vice versa.
Evidence of Short-term Phenomena
For shorter-term strategies the big question we must ask is: how well do these principles hold up at lower time frames, where we might bet more frequently to increase our return? At which point does random noise take over and are there other reliable concepts at those shorter time periods?
Many studies agree on the medium- to long-term time line:
- mean reversion: 1 month
- momentum: 3-15 months
- mean reversion: 1 to 5 years
Below one month reliable evidence for these phenomena becomes scarce.
Market inefficiencies caused by participants who lose willingly (e.g. hedgers) or are forced to behave as they do (e.g. institutions or central banks), are likely to be very persistent and play a role on a weekly to monthly time-frame as well as over longer horizons.
There exist alternative risk premia based on carry
(the expected return on an asset assuming that market conditions, including its price, stay the same) that have shown strong persistence. For example, the volatility risk premium
is based on persistently high demand for insurance and hedges in equities as well as real assets.
These carry premia take time to play out – I found a period around one month to offer a very good reward / risk ratio.
Lack of Evidence for Intraday Strategies
In the realm of day-trading strategies the influence of basic risk premia ceases to matter, as we found empirically by looking at the equity risk premium (table above). Other predictable and persistent patterns must be found, otherwise results will be random as they have no edge. High frequency trading is an example where the market structure at the lowest possible time frame is used to generate superior returns, but even here fierce competition has driven those returns down and costs up.
Unfortunately the majority of studies, I found, that look at traditional technical short-term trading in equities, currencies and futures markets, conclude that there are no superior returns to be had – not even before trading costs. Trading lore of outsized gains may have been true a long time ago, but it looks like these have long since been arbitraged away.
While there are plenty of descriptions of technical patterns, fibonacci numbers, wave formations and the like, I could find few reliable, positive strategy performance statistics from the intraday space in my research – if you know differently, please send them to me and I´ll add the data to an article update.
Most traditional setups transfer known phenomena, as described above, into shorter time frames without considering that increasing noise overshadows the valid information that may be contained in market prices. The concept of support and resistance, which lies at the core of most trading methodologies, targets breakout momentum and mean reversion.
Can this simple transfer work? With the equity risk premium we have seen empirical evidence that it deteriorates into pure randomness – a valid signal is reduced to random noise below a daily period. Increasingly specific, optimized strategy rules become necessary, which runs afoul of the idea that parameter stability is the premise for a durable strategy.
Still the myth of a quick road to riches via day-trading is persistent and stories abound – probably because it is such a desirable dream and a lot of money can be made on commissions and the sale of day-trading methods. What general information points to day-trading profitability either being a myth or an attainable goal?
- Studies point to no more than 1% to 5% of day-traders being able to consistently return outsized profits after cost. Even these outliers could be simply due to luck as the percentage is just so low: take the analogy of the seemingly skillful coin tosser – given a large enough population, just by coincidence someone must be in the top 1% of the heads-only throwers for many consecutive random throws. Even if the return persistence is due to extraordinary skill, the low percentage of winning traders still means that it is very unlikely we will ever reach that level.
- A direct reason such a large number of day-traders lose money could simply be, that most don´t have any edge – they are noise traders. I would go so far as to conclude that a strategy that relies on very exact strategy rules, e.g. entry points, to work is over-optimized to randomness and trades primarily on noise rather than signal.
- The human mind is made to find patterns in randomness – we are pattern recognition machines even if there are no reliable patterns there. What gives us a survival advantage in the real world cannot simply be transferred to the financial markets.
- Machines are better than human beings at finding small, statistically significant patterns in an ever changing environment. Algorithms are a major competitor in the trading space, which is likely to lead to unstable patterns in the future, if the edge a day-trader uses is based on human behavior.
- There are limits to possible returns. While one-offs or even several heads in a row are certainly possible, sustainable compounding at 50%, 100% or more annually would lead to unimaginable wealth within a few years. There is no evidence such wealth creators exist – the visible super-rich investors posted returns much closer to normal market returns over long periods of time (20% to 30% per year). If actual day-trading returns are closer to longer term strategies´ results, isn´t it more reliable and easy, with less mistake prone screen-time, to concentrate on those time frames?
- Most day-trading success stories use momentum setups – trading breakouts under high volume. The best statistical evidence I have found so far comes from very specific mean-reversion strategies, though.
Not many positive points here to endorse a day-trading career, unfortunately. Please do send me the evidence you have seen, backtested, traded etc..
All successful investing and trading is based on an underlying principle that provides an edge that makes positive returns possible. The question we must ask of any strategy we trade is: is our trading based on a persistently strong market force and how much is left of it at the time-frame we target – is it still a reliable concept?
If there are different phenomena at shorter time-frames, we need to figure out what is happening and why – are they dependable or are we seeing patterns where there are none?
For me the easiest conclusion from the lack of evidence so far is to stay away from the very short-term space – there is plenty to be gained within the longer time horizons.