They are not.

The difficult part is not proving that a signal contains information, but determining whether that information can survive the frictions required to monetize it: that is the challenge we take up in our second QuanTip piece.

We focus on fast-decaying alphas, meaning very short-horizon signals whose informational content is concentrated over seconds or minutes rather than days or weeks: such signals can look impressive before costs, which is precisely why they attract so much attention.

However, that first impression can be misleading. In many cases, the expected profit per trade is very close to trading costs.

In this piece, we study a mean-reversion signal that looks for short-term overreactions: after N consecutive 5-minute bars moving in the same direction, we take the opposite side on the next bar.

Read the Paper

For example, a gross-of-fee simulation of the N=1 specification yields striking results, with a CAGR of roughly 31.9% and a Sharpe ratio exceeding 2. However, once realistic transaction costs are added, the CAGR turns negative.

Does this mean the signal is useless? Not necessarily!

Our experience shows that a framework shift may be necessary: fast alphas often can’t stand alone, but they can improve the execution profile of a slower strategies that is already economically viable.

In fact, when this same signal is used to refine the entry and exit timing of a more robust trend strategy, we observe a meaningful net-of-fee improvement.

The base strategy already delivers a double-digit CAGR and a Sharpe ratio of approximately 0.86. With the fast-alpha overlay applied, the Sharpe rises to roughly 0.99 and net-of-fee CAGR improves by approx. 200 basis points.

Such evidence is not trivial!

It suggests that many short-horizon signals should not be judged only by their standalone, net-of-fee performance. A more appropriate question is how they behave when combined with other signals.

In many cases, their role is not to generate returns independently, but to support and refine the behavior of other strategies. Seen this way, their economic value lies not in isolation, but in how they interact with an existing signal structure.

If you are curious to dive deeper into the distinction between monetizable alpha and informational alpha, or explore the full methodology and details, we recommend reading the full piece.

Read the Paper

If you found this article useful, feel free to leave a comment. You can also reach out via direct message or email at info@concretumgroup.com for any questions.

We would expect them to achieve nearly identical results over time… right?

In our new work, we find that this seemingly innocent decision can have massive implications for realized wealth. This effect is commonly referred to as Rebalance Timing Luck (RTL), and a number of researchers (notably Hoffstein and Blitz) have over the years provided extensive evidence of the implications of this often overlooked aspect of portfolio management.

Over 33 years of history, we show that the single decision of changing the day of execution in a concentrated and monthly-rebalanced momentum portfolio can generate a difference in CAGR of nearly 350 basis points between the luckiest and unluckiest outcomes.

Read the Paper

For reference, an initial dollar invested in the least fortunate schedule grows to $99, whereas in the most fortunate, it grows to $272. The same strategy, the same signals, a difference of almost threefold in terminal wealth from a choice that most practitioners and researchers would describe as an operational footnote.

Moreover, this effect is not predictable. As prior literature has established, performance deviations induced by RTL are not expected to mean-revert.

They compound, and they persist.

A common approach to mitigate RTL is portfolio tranching: dividing the portfolio into N sub-portfolios, each rebalanced on a different staggered schedule, so that no single execution date dominates the outcome.

Our results confirm that tranching is effective, with RTL declining roughly in proportion to 1/N, while annual turnover remains broadly unchanged.

However, reducing RTL is not free.

We introduce a cost-aware framework, evaluated across different AUM levels and tranching schemes, applied consistently to the same concentrated momentum strategy. The results show that the optimal number of tranches depends critically on investor size.

In particular, smaller retail accounts and larger institutional portfolios sit on opposite ends of the tranching trade-off, and the optimal approach differs meaningfully between them…

If you want to find out where your portfolio sits on that trade-off, and how much terminal wealth may be at stake, we encourage you to read the full paper.

Read the Paper

If you found this article useful, feel free to leave a comment. For any questions, you can also reach out via direct message or email at info@concretumgroup.com

The hard part is designing a quantitative framework in which news relevance can be translated into actionable decisions: one that operates at scale and does not dissolve into discretion the moment the informational picture becomes complicated.

In this research project, in collaboration with Finreon AG (a leading quantitative investment boutique based in Zurich), we explore a specific application of large language models to momentum investing.

If a stock is already about to enter a momentum portfolio, can an LLM help us judge whether the recent flow of firm-specific news supports a continuation of that trend?

A canonical momentum signal tells us that a stock has been advancing, but it cannot tell us whether the news environment still supports that move. That gap is precisely where large language models may earn their place in a systematic process: not as oracles, but as interpreters of unstructured information that traditional price-based signals were never meant to read.

Read the Paper

We combine daily returns for S&P 500 historical constituents with high-frequency, firm-level news data, feeding prompt-engineered queries to GPT-4o mini that explicitly frame each request in the context of a momentum trading decision.

The model then outputs a score indicating whether recent news supports the continuation of a stock's existing relative strength: that score can condition both stock selection and portfolio weights. Notably, in our design the large language model does not replace the strategy, it attempts to augment it.

The empirical results offer meaningful support for this approach.

Over the full sample, the LLM-enhanced strategy increases annualized returns from 15% to 18%, improves the Sharpe ratio from 0.57 to 0.69, and reduces maximum drawdown from -33% to -31%, all net of conservative transaction cost assumptions of 2 basis points per trade.

The out-of-sample period begins in January 2024, which places it after the October 2023 pre-training cutoff of the model used. As a result, any improvements observed cannot plausibly be attributed to memorized historical information.

Importantly, these results should be interpreted with appropriate caution. The out-of-sample period covers only fifteen months, limiting statistical power, and the estimated annualized alpha of 3.26% is significant only at the 10% level. The analysis is further restricted to a single pre-trained model applied to large-cap U.S. equities, leaving open questions about generalizability to less liquid markets or other asset classes.

There is a second finding worth noting.

We find that the LLM augmented strategy performs best when it is relatively concentrated: at 25 holdings, the out-of-sample Sharpe ratio reaches approximately 1.3, declining as the portfolio broadens to 100 names. This suggests that the model’s informational edge is not uniformly distributed across the cross-section but is strongest where the news flow is clearest and most economically relevant.

If you are interested in the full methodology, prompt design, and portfolio construction details, and in understanding what this implies about the role of unstructured data in systematic investing, we invite you to read the paper.

Read the Paper

If you found this article useful, we’d be glad to hear your thoughts in the comments. For any questions, feel free to reach out via direct message or email at info@concretumgroup.com

Volatility traders, instead, ask a different question.

They do not care much whether the market rises or falls: they care whether it moves more or less than what options currently price in. That distinction may sound subtle, but it implies a fundamentally different source of return, one that, in principle, should behave independently of the equity market.

In this new paper, we analyze volatility trading with a particular focus on VIX ETNs, exchange-traded products that have grown increasingly popular in recent years, allowing traders to gain long and short exposure to the entire VIX futures complex.

Read the Paper

The framework we lay down progresses through four rule sets, each adding a layer of predictive content to the signal used.

A purely passive short-volatility allocation serves as our baseline, delivering single digit returns, high drawdowns, and a 72% correlation to the S&P 500. Its purpose is simply to capture the volatility risk premium in its most basic form.

We then progressively enhance the framework through three key improvements:

• Volatility risk premium filter: a signal based on the expected difference between implied and realized volatility

• VIX term structure signal: a more forward-looking indicator that captures information embedded in the futures curve

• Dynamic position sizing: a final refinement that adjusts exposure based on the current level of the VIX

Together, these enhancements transform the baseline strategy into a materially stronger framework: the Sharpe ratio increases from 0.48 to 1.00, CAGR rises from 6.2% to 16.3%, and correlation to the S&P 500 drops from 71.6% to just 14.8%, turning a fragile, equity-like exposure into a more robust and genuinely diversifying return stream.

Beyond the strategy itself, we examine how a volatility sleeve of this kind interacts with a passive exposure on S&P500 (SPY ETF), finding consistently attractive results across a range of allocation specifications.

That said, we would like to close with a note of genuine caution: the volatility risk premium has been persistent, but (like anything in markets) it is not guaranteed, and any type of strategy in the volatility space demands constant sizing discipline and an honest accounting of tail risk.

As a practical extension, we also include Python code showing how this strategy can be implemented and automated through Interactive Brokers, bridging the gap between research and execution.

For the full methodology, details, and Python implementation code, we encourage you to read the paper.

Read the Paper

If you found this article useful, feel free to leave a comment. You can also reach out via direct message or email at info@concretumgroup.com for any questions.

For us, that framing misses a more practical issue.

Once a market exists, becomes liquid (attracting both retail and institutional capital) we are interested in studying if its price behavior can be approached with the same disciplined tools that have worked elsewhere.

In this new paper, we take a classic long-only breakout methodology (well-known from decades of traditional trend-following) and apply it to both Bitcoin and a broader, survivorship-bias-free universe of cryptocurrencies.

Read the Paper

Specifically, our signal is built as an ensemble of Donchian Channels across nine lookback horizons, combined with a volatility-based position sizing approach targeting an annualized volatility of 25%.

Applied to Bitcoin from January 2015, the ensemble model delivers a net-of-fee CAGR of 30% and a Sharpe ratio of 1.56, while constraining maximum drawdown to 19%, compared with drawdowns exceeding 80% under passive exposure.

When the same framework is extended to a monthly rotation across the most liquid cryptos, a trend-managed portfolio targeting to hold the top 20 most liquid names produces a net-of-fee CAGR of 18% and a maximum drawdown of just 11%, delivering a statistically significant alpha of almost 11% versus Bitcoin.

We then examine how crypto trend-following relates to the broader managed futures category. The average rolling correlation between our diversified crypto program and the SG Trend Index over the past decade is approximately 7.4%, with extended periods where the two strategies are effectively uncorrelated or modestly negatively correlated.

That low structural overlap suggests that a crypto trend allocation may serve as a meaningful complement to traditional CTA programs, rather than merely replicating them in a more volatile wrapper.

All in all, crypto remains a young, constantly evolving asset class with volatility that has no real precedent in traditional markets… but that is precisely what makes our results interesting: classic trend-following frameworks, developed and refined over decades in traditional markets, appear to travel remarkably well into this environment, and they do so without meaningful modification.

If these results appear unsurprising because the strategy is implemented in a long-only form, the paper provides a broader perspective. We also examine a long–short implementation, where excess returns relative to a static Bitcoin exposure are even more pronounced.

For the full methodology, portfolio construction details, and implementation considerations (including on-chain and off-chain execution) we encourage reading the paper directly.

Read the Paper

If you found this article useful, we’d be glad to hear your thoughts in the comments. For any questions, feel free to reach out via direct message or email at info@concretumgroup.com

In 2005, Cole Wilcox and Eric Crittenden published a white paper tackling a simple but surprisingly underexplored question:

Does trend following work on individual stocks?

In this paper, written in collaboration with Cole Wilcox, we revisit and build upon their original research, drawing on a survivorship-bias-free dataset encompassing all liquid U.S. stocks (from 1950 through October 2024) and analyzing more than 66,000 simulated long-only trend trades.

Read the Paper

We find that fewer than 7% of trades essentially account for all cumulative profitability, while 56% of trades result in losses. The win rate across the full sample stands at 43.9%, with the average winning trade returning 1.90R versus an average loss of -0.70R, a payoff profile that exhibits the strong positive skew characteristic of trend-following strategies.

In other words, the outliers are not a random byproduct of the process. They are the process.

Average profitability per trade declined modestly after 2005, slipping from 0.39R to 0.31R. Yet cumulative annual returns improved in the post-publication period, driven by an expanded opportunity set. The edge, in other words, was not arbitraged away.

The second part of our work translates those signals and statistical properties into a tradable portfolio, backtested from 1991 to 2024.

The theoretical portfolio (excluding transaction costs and assuming fractional share allocation) achieves a CAGR of 15% against the market’s 11.2%, a Sharpe ratio of 0.85, and maximum drawdown of 31.8% compared to nearly 55% for standard buy-and-hold during the Great Financial Crisis.

Transaction costs, however, fundamentally alter the picture for smaller accounts. Without turnover mitigation policies, the $100,000 portfolio sees its CAGR collapse from 15% to 2.5%, with alpha turning negative at -4.6% per year as the result of high daily turnover driven by small rebalancing adjustments.

To address this, we introduce a turnover control mechanism that skips sub-threshold rebalances, distributes large trades across multiple days, and filters adjustments whose expected commission exceeds a preset fraction of notional traded.

The impact is substantial!

Annual commission drag falls from over 10% to under 1%, CAGR recovers to 13%, and annualized alpha reaches a statistically significant 4.9%, with results remaining broadly stable across portfolio sizes from $100,000 to $100 million.

If you are interested in the full signal construction, statistical breakdown, and implementation details, we invite you to read the paper.

Read the Paper

If you found this article useful, we’d be glad to hear your thoughts in the comments. For any questions, feel free to reach out via direct message or email at info@concretumgroup.com

From coin clipping in the ancient world to the Ponzi schemes of the twentieth century, what holds these episodes together is not the complexity of the deception but its apparent plausibility: the ability to make an implausible claim look self-evidently reasonable.

We argue that the Martingale betting system, a method devised in 18th-century France for gambling, has quietly become one of the most effective tools for creating such illusions, and is increasingly finding its way into modern financial products designed to appear far more robust than they truly are.

Read the Paper

A Martingale system is typically designed to double the position size of a bet after every loss, on the premise that a single subsequent win will recover all prior losses.

Applied to a trading bot, this produces an equity curve that appears smooth and stable for long periods.

That apparent stability is precisely what attracts investors. What it hides is a structural exposure to rare but severe losses, meaning that investors are unknowingly taking on a significant risk of ruin.

In this paper, using QQQ as the trading instrument, we deploy three progressively refined strategies across 10,000 simulated accounts, each driven by purely random signals. We find that with randomly generated trading signals and a martingale structure, a strategy can still appear to deliver a fixed 20% annual return with nearly 80% probability.

The path looks reassuring, the results look systematic. The underlying process is neither.

Our work then extends into a fictional but precisely constructed scammer business plan. A fraudster opens ten accounts with $1,000 each, trades the full strategy, and presents the best-performing account as a certified track record (Account #8 in the table above). In our simulation, that account ends the year with 7% annualized volatility, a Sharpe Ratio of 3.0, and a maximum drawdown of just 2%, numbers that would attract serious interest from almost retail investors (and, at times, even institutional ones).

The fraudster then recruits 100 clients at $10,000 each, charging a 1% management fee and a 25% performance fee. Under a word-of-mouth growth model, the client base compounds from 100 to 4,912 over ten years. By the end of the decade, the bot provider has collected over $6.5 million in fees while aggregate net P&L across client accounts exceed a loss of $7.5 million.

Such numbers are striking not because they require elaborate manipulation… but because they do not. The entire architecture depends on nothing more than a random signal, a leverage rule, and the willingness of roughly 75–78% of clients in any given year to attribute their profits to genuine skill.

Those clients recommend the service to their peers, the losing minority exits quietly… and the cycle continues.

As trading bots proliferate in retail markets (a trend that accelerated visibly after COVID-19), what we want to outline is a general vulnerability: the capacity of a well-constructed track record to function as social proof even when it carries no information about future performance.

The complete methodology, simulation results, and a discussion of the regulatory challenges are presented in more detail in the paper.

Read the Paper

If you found this article useful, feel free to leave a comment. You can also reach out via direct message or email at info@concretumgroup.com for any questions.

Discretionary traders often rely on instinct and intuition, refined through years of exposure to market patterns; quants root decisions in measurement, rules, and statistical rigor.

The assumption has always been that these two worlds are irreconcilable.

But what if they don’t have to be?

We decided to put that assumption to the test alongside Marios Stamatoudis, a Greek trader whose verified 291% return earned him a top 5 finish in the 2023 US Investing Championship.

Read the Paper

We begin by studying 9,794 overnight gap events in U.S. equities from 2016 to 2023.

The initial results are unambiguous: trading all gaps on the long side consistently loses money.

More broadly, every rules-based variation we test fails to produce meaningful improvements. Systematically buying stocks after large gap-ups appears to be a structurally weak approach.

And yet, within this unpromising universe, a small tail of outlier trades delivers exceptional returns.

To test whether a skilled practitioner (like Marios) could identify those setups in advance, we designed a custom software environment that removes all sources of bias. Each chart was fully anonymized: no dates, no tickers, no price levels, no volume, no sector information, and no news context.

The video below explains how the software works in more detail.

Marios was shown only two years of prior price action and asked to approve or reject each setup based solely on that information. No context. No clues.

In total, only 18% of eligible gaps were approved. Yet this single act of filtration materially altered the outcome. Restricting the strategy to the approved subset significantly improved performance.

Applying the strategy exclusively to the approved subset improved profitability in a material way, and when Marios was allowed to also manage entries and exits bar by bar within the same bias-free environment, the results became even more striking: a hypothetical $100,000 portfolio grew to over $4,000,000 over eight years, with a CAGR of 59.1% and a Sharpe ratio of 1.70.

If you are interested in the full design of the experiment and in understanding how discretionary judgment can complement systematic frameworks, and what this implies about the limits of purely systematic trading, we invite you to read the original paper.

Read the Paper

If you found this article useful, we’d be glad to hear your thoughts in the comments. For any questions, feel free to reach out via direct message or email at info@concretumgroup.com

There is no shortage of research documenting its effectiveness across asset classes.

Yet one question remains surprisingly underexplored:

Can we exploit trends across U.S. industries in the same way?

In this paper, written in collaboration with Gary Antonacci, a pioneer in momentum investing, we approach this question using a classic channel breakout system, closely aligned with how trend-following is implemented in practice by systematic managers.

The work was recognized with the 2025 Dow Award for Best Paper of the Year by the CMT Association.

Using 48 U.S. industry portfolios from 1926 through 2024, we examine if a long-only industry trend-following program can remain both profitable and resilient over almost a full century.

The appeal of industries, as opposed to single stocks, stems form the fact that they sit at an interesting intermediate level: broad enough to capture general relative strength dynamics across economic cycles, yet specific enough to express idiosyncratic exposure without simply loading up on generic market beta.

Read the Paper

We find that our Industry Timing portfolio compounds at 18.2% per year with an annualized volatility of 12.6%, with a resulting Sharpe ratio of 1.39. Over the same period, the U.S. equity market has returned 9.7% per year with volatility of 17.1% and a Sharpe ratio of 0.63.

The gap is already significant on its own, but what makes the result more interesting is their risk profile: the active strategy’s maximum drawdown sits at roughly 33%, compared to around 84% for passive market exposure

That is why simulations spanning such a long sample are so interesting.

It is not just that returns remain strong: it is that our trend framework continues to function as the underlying economic conditions change almost beyond recognition. A result like that does not guarantee the edge will be permanent, but it does make it hard to dismiss trend trading (across industries) as a historical accident or a postwar anomaly.

For us, the deeper implication is that industry-level trend trading may deserve more attention than it typically receives.

Stock selection and broad asset allocation dominate most of the literature and most practical discussion. Industries, by contrast, may offer a level of aggregation that captures economic leadership more cleanly, particularly for investors who want something more targeted than an index but still less idiosyncratic than pure stock picking.

That leaves an interesting question open: if momentum appears this durable at the industry level, perhaps the real debate is not whether the effect exists, but whether investors have been looking for it at the wrong level of granularity.

If you are keen to explore the full details and methodology, including how closely such portfolios can be replicated via modern ETFs, we suggest reading the whole paper here:

Read the Paper

If you found this article useful, feel free to leave a comment. You can also reach out via direct message or email at info@concretumgroup.com for any questions.

That is only part of the picture.

Our experience as practitioners tells us that markets can exhibit much shorter bursts of directional movement, often within the trading day. What makes this particularly interesting, from a research standpoint, is determining whether these moves are simply noise or something genuinely exploitable.

In this paper, we focus on SPY, one of the most widely traded and closely observed ETFs in the world, and investigate whether abnormal intraday demand–supply imbalances can be captured using a systematic trend-following framework.

Read the Paper

Our signal detection begins by defining what we call the noise area: a time-varying equilibrium zone built from SPY’s average absolute move from the open at each intraday timestamp over the previous 14 sessions.

These historical average moves define an upper and lower boundary around the day’s opening level, so that the threshold required to signal imbalance is typically narrower early in the session and wider later in the day.

In other words, these bands are designed to separate ordinary market oscillation from abnormal buying or selling pressure. As long as price stays within that zone, we assume the market is in equilibrium and the strategy stays flat. Only when price breaks above or below those boundaries do we treat the move as evidence of a genuine intraday trend worth acting on.

What has emerged clearly through our research is that capturing these intraday moves effectively is not about applying a single idea, but about building a framework across coherent and incremental layers.

A basic version of the strategy, with a relatively simple stop policy, produces an annualized return of 6.2% and a Sharpe ratio of 0.61. Once the exit logic is tightened, moving to a current-band plus VWAP stop, annualized return rises to 9.7% and Sharpe roughly doubles to 1.24.

The final version, which adds dynamic position sizing on top of the refined trailing-stop framework, reaches a total net return of 1,985% from 2007 to early 2024, with an annualized return of 19.6% and a Sharpe ratio of 1.33.

Beyond the core results, the paper explores several additional dimensions, including conditional profitability and day-of-the-week effects, and concludes with a detailed FAQ section aimed at practitioners seeking to further refine and enhance the strategy.

For everyone interested in the full details, the methodology, and in discovering more about intraday momentum effects, we suggest reading the paper:

Read the Paper

If you found this article useful, we’d be glad to hear your thoughts in the comments. For any questions, feel free to reach out via direct message or email at info@concretumgroup.com

Despite this, they have received relatively limited attention in academic research, which is precisely what motivated us to pursue this study.

By focusing on a modified version of the classic 5 minute opening range breakout, we analyze a dataset containing more than 7,000 U.S. stocks traded between 2016 and 2023.

Rather than treating every stock at the open as equally relevant, we ask a more targeted question:

Does the true potential of an ORB framework emerge only when applied to stocks exhibiting unusually high trading activity in the first minutes of the session?

Read the Paper

We refer to these as “Stocks in Play”, and the empirical results point to a clear distinction.

When the 5-minute ORB signal is applied broadly across the market, the result is somewhat disappointing: about 3.2% annual return, roughly 6.6% annualized volatility, and a Sharpe ratio of 0.48, which lags a passive S&P 500 benchmark.

Once the universe is restricted to Stocks in Play, the risk-return profile of the portfolio changes materially.

The enhanced strategy produces more than 1,600% in total net return from 2016 to 2023, with a Sharpe ratio of 2.81 and an annualized alpha of 36%: over the same period, the S&P 500 gained roughly 198%.

Notably, the improvement we observe does not come from making the breakout signal more complex: it comes from analyzing supply and demand imbalances in a more accurate way.

Our results suggest that not every opening move deserves attention: only Stocks in Play appear to reflect a genuine repricing phenomenon that persists after the first few minutes of the session.

We believe these findings raise a more fundamental question for both practitioners and researchers. When evaluating a day trading strategy on equities, should the focus be on entry and exit rules, or on the more critical question of where to deploy risk in the first place?

As always, for the full details and methodology, you can read our paper here

Read the Paper

If you found this article useful, feel free to leave a comment. You can also reach out via direct message or email at info@concretumgroup.com

In this paper, we set out to take this famous indicator out of folklore and place it under more rigorous evaluation. It is unquestionable that many traders watch VWAP: the real challenge is understanding whether it can serve as a core signal within a directional intraday trading framework.

The rules we define are deliberately simple: using one-minute intraday data, our strategy takes a long position when price closes above VWAP, a short position when it closes below, and ensures end-of-day liquidation to avoid any overnight exposure.

This simplicity is central to our approach: if the results are strong, they are less likely to be the product of overfitting.

Read the Paper

When applied to QQQ, the results are substantial. A $25,000 starting account grows to $192,656 net of commissions, a total return of 671%, with a maximum drawdown of 9.4% and a Sharpe ratio of 2.1. Over the same period, passive QQQ returns 126%, with a drawdown of approximately 37% and a Sharpe ratio of 0.7.

Extending the strategy to TQQQ (the triple-leveraged replica of QQQ) produces even more striking results: the same starting capital reaches $2,085,417, representing an 8,242% total return and an average annual return of 116%.

The next question we address is whether VWAP is merely a proxy for other trend-type indicators and not a signal with independent predictive merit.

Running the same entry and exit logic on SMA9, SMA20, SMA100, and SMA200, we find that VWAP outperforms all of them.

One possible explanation is structural: because VWAP resets daily and incorporates volume directly, it may respond more effectively to intraday supply–demand imbalances than time-weighted averages.

Notably, these positive results are achieved with a hit ratio of only 17%, meaning the system depends heavily on payoff asymmetry rather than frequent accuracy.

Importantly, transaction cost assumptions are critical, and the relatively high turnover of the strategy can materially impact its standalone profitability, particularly as account size increases and market impact becomes more relevant.

For this reason, we believe the most appropriate interpretation of these results is not that VWAP should be viewed as a standalone trading system, but rather as a source of fast alpha: a short-horizon signal that can be used to refine execution within broader frameworks.

In particular, VWAP may serve as a confirmation tool or as a dynamic trailing stop in strategies where turnover is more controlled. We have explored this approach in our paper “Beat the Market: An Effective Intraday Momentum Strategy for the S&P 500 ETF (SPY)”, where VWAP-based exit logic proved to be an effective component of the overall framework.

Seen in this light, VWAP is not merely a reference line, but a practical building block that can enhance the implementation of more robust trading systems.

For full details, benchmark comparisons, and methodology, we invite you to consult the paper.

Read the Paper

If you found this article useful, feel free to leave a comment. For any questions feel free to reach out via direct message or email at info@concretumgroup.com

Studies from Brazil and Taiwan have shown that the overwhelming majority of retail traders lose money over time, and critics are not wrong to point out the real frictions involved: commissions, taxes, leverage risk, and the psychological cost of being wrong frequently.

Against that backdrop, the question in our title is not a provocation: it is a genuine empirical problem.

We investigate the profitability of a modified 5-minute Opening Range Breakout strategy applied to QQQ, the most liquid instrument replicating the Nasdaq-100, over January 2016 to February 2023.

Read the Paper

The point of our study is not to construct the best possible system: it is to test whether a transparent, rules-based intraday trading process can beat a passive benchmark on a risk-adjusted basis.

Starting with an initial capital of $25,000, the ORB portfolio on QQQ reaches a total return of 675% by February 2023, against 169% for passive QQQ. Annualized alpha is 33% net of commissions, with a p-value of 0.25%. Across 1,795 trades, we record a split of 51% long and 49% short, with a win rate of roughly 24%.

Notably, our ORB framework is profitable not because it is accurate, but because our profit-and-loss distribution exhibits a strong positive skew: this shows how a low win-rate compensated by an asymmetric payoff can still yield attractive results.

There is, however, a structural constraint we address directly.

U.S. brokers cap intraday leverage at 4x net liquidation value, resulting in roughly 60% of our trades being sized approximately 40% below their theoretically optimal exposure.

Given that the unconstrained version of the same ORB strategy would have returned 1,630% over the same period, we also extend the study to TQQQ, the 3x leveraged Nasdaq-100 ETF. Because TQQQ amplifies the daily move of its underlying, a $25,000 account can achieve an exposure profile that would otherwise require a much larger one.

Overall, we recognize that day trading faces significant operational challenges. In practice, factors such as execution, slippage, discipline, and leverage constraints can easily turn a seemingly profitable strategy into an unprofitable one.

At the same time, the evidence suggests that simple, well-structured systems can exhibit a meaningful and statistically significant edge, even with little to no exposure to market beta.

The framework presented in this paper should therefore be viewed as a baseline. It is not intended as a final, fully optimized solution, but rather as a starting point from which more robust and sophisticated trading strategies can be developed.

As always, for further details, results, and methodology, we invite you to consult the full paper.

Read the Paper

If you found this article useful, feel free to leave a comment. You can also reach out via direct message or email at info@concretumgroup.com for any questions.