MAR and MAD contain ISX60 drawdown, with no clear winner

The drawdown reduction against the ISX60 is the finding worth trading attention; MAR's advantage over MAD is much harder to defend. Zayerl and Ahmed report maximum drawdowns of 25.40% and 26.90% for their constructed portfolios, against 68.16% for the index. Their abstract instead leads with the MAR portfolio's "economic superiority" over MAD. The point estimates favour MAR, but the authors' own tests cannot distinguish the portfolios. They describe the advantage as economic and conclude that the metrics complement each other and should be used together. Even with that qualification, the abstract gives top billing to a superiority the tests cannot establish.

Risk measured two ways

The portfolios compete on different definitions of risk. Mean absolute deviation (MAD) takes the average absolute distance of returns from their mean. The MAR ratio divides compound annual growth by maximum drawdown, rewarding growth relative to the worst peak-to-trough loss along the realized path. Each measure supplies the objective for a sector-level portfolio. The study covers 39 companies from 2014 to 2025 and uses the ISX60 as its benchmark. Its stated sectors are Banking, Insurance, Industry, Hotels and Services.

Loss containment is the paper's case for this approach. The authors argue that Sharpe, Treynor and Jensen capture volatility or beta while missing the path wealth takes. They see that omission as a problem in a frontier market with skewed returns and extreme observations. MAD and MAR also dispense with a risk-free rate, a practical convenience in Iraq.

At 15.21%, Industry had the highest sector average return. The MAR portfolio returned an annual average of 11.18%, with CAGR of 9.59%, a MAR ratio of 0.377 and final wealth of 273.77. MAD's MAR ratio came to 0.270; the market's was -0.060. Maximum drawdown reached 25.40% for MAR and 26.90% for MAD, compared with 68.16% for the ISX60. The reported CAGR, drawdown and ratio for MAR agree with one another.

How much does 0.377 tell us against 0.270?

The authors acknowledge the statistical limitation. Their abstract says "no significant direct differences were found regarding returns," and reports the same finding for maximum drawdown. The conclusion supplies Wilcoxon p = 0.3652 for returns (H01) and exact permutation test p = 0.8750 for maximum drawdown (H02). A ranking analysis also failed to reject H03, although the authors identify economic differences in the rankings. Their response is "an economic preference for the MAR portfolio in terms of growth, final wealth, and the relationship between CAGR and MDD."

Look first at the drawdowns. The difference is 1.5 points, or 25.40% versus 26.90%. That 1.5-point gap alone cannot account for MAR ratios of 0.377 and 0.270. Growth therefore explains most of the difference, measured along a single eleven-year path. The p-value of 0.8750 equals exactly 7/8, a round fraction consistent with a test allowing few possible rearrangements. How many observations went into it is unstated. Growth, final wealth and CAGR over MDD, the three grounds for the economic preference, all come from the same realized path.

The comparison with the benchmark carries a different message. Both portfolios reject H04 against the market: MAD at p ≈ 0.0376 and MAR at p ≈ 0.0386. Those p-values are almost identical as well, consistent with the methods reaching much the same destination.

Eighty percent in two sectors

The MAR weights are reported as "40% to Industry, 40% to Agriculture, 10% to Hotels and Tourism, and 5% to each of Banks and Insurance." Yet Agriculture is absent from the five sectors named in the sample. Services appears in that list and receives no allocation. Readers are left unable to tell whether the sample description is wrong or a sector has been relabelled. Either explanation leaves 40% of MAR allocated to a sector outside the stated sample.

Another 40% goes to Industry, whose realized average return of 15.21% leads the sectors. Allocating 40% to the best-returning sector accounts for much of the portfolio's growth. A successful realized allocation, though, does little to establish that the drawdown objective earned the result. We found no statement that the authors estimated weights in an earlier window and tested them in a later one. If they fitted weights across the full 2014 to 2025 sample, MAR was selected on the path used to grade it, through realized CAGR divided by realized drawdown.

All reported weights are multiples of 5%, consistent with a coarse 5% grid that the paper leaves undescribed.

Can the published method be rebuilt?

The text available to us passes straight from a literature review of Sharpe, Treynor and Jensen to the conclusion. We could not find an optimization formulation, weight constraints or a rebalancing rule. Neither the return frequency nor the observation count for each test was apparent. The paper warns that closing-price data produces price returns excluding dividends, yet we could not determine which return measure it used. Its abstract says CAPM and the Treynor index were applied; we found results for neither. H01 through H04 arrive in the conclusion without formal definitions. We also found no transaction costs, an omission that matters for a thinly traded frontier market.

We could not run a test ourselves. We lack the ISX price and index history needed to rerun the study. A MAD-versus-MAR exercise on US stocks would test another market.

Zayerl and Ahmed offer practitioners a narrow result. Their portfolios kept maximum drawdown to 25.40% and 26.90%, against the ISX60's 68.16% during 2014 to 2025, apparently through an in-sample fit. That supports the authors' recommendation to report dispersion alongside drawdown. It leaves little basis for switching a mandate from MAD to MAR. The direct tests returned p = 0.3652 and p = 0.8750, and we could not find enough of the construction method to rebuild the portfolios. A walk-forward split in which MAR chose the Industry tilt before the fact would change our view.