Test a high-low forecast as one procedure, then cut the book

Fix the entry, the first touch, and the clock

The historical workflow treated each published high and low as one rule-based-entry. The book entered at the next session close after a forecast.

Exit followed a range-first-touch rule: the trade closed when price first reached either the stated high band or the stated low band. If neither edge was reached, a time-stop forced an exit on the 90th trading day.

What the mixed book showed

From October 2009 through July 2016 the collected book held nearly 800 recommendations. The realized average return was 0.78% per recommendation, and the average holding period was 21.4 days.

Excluding exchange-traded-fund trades raised the same-period average return from 0.78% to 1.78% and lengthened the average holding period to 26 days. Over matching holding periods the S&P 500 averaged a 1.18% total return.

Industry-group-slice and the large fund sleeve

Among industry groups with at least six trades, Independent Oil & Gas showed the highest listed return at 9.10% across 23 trades. Credit Services showed the lowest at -1.44% across 9 trades.

Exchange-traded funds supplied the largest trade count, 409, but posted a -0.15% return and 42.5% winning odds under the same first-touch rules.

Editorial: an industry-group-slice is one way to place the same rule-based-entry in a market-participant-analysis. The mixed book is not one participant.

Market-cap-bucket and winning-target cuts

Market-cap-bucket bands were defined as nano under $500 million, small from $500 million to $2 billion, mid from $2 billion to $10 billion, big from $10 billion to $200 billion, and mega over $200 billion.

Those slices produced average returns of 4.54% for nano-cap names, weaker small-cap results, then 1.67%, 1.97%, and 2.01% for mid, big, and mega names. The nano-cap sample was flagged as too small for statistical significance.

Forecasted winning-target buckets all showed positive average returns except the bucket with a predicted return above 20%. A winning-target is the upside percentage implied by the forecast high relative to the entry price.

Seasonality-analysis by calendar month

Average monthly results were strongest in June, at a 5.58% return and 60% winning odds. November, February, January, and July were also listed as relatively strong.

March and April were the weakest months. September, October, and December were also listed as months to avoid.

Seasonality-analysis groups the same rules by calendar month. Editorial: use that cut so a strong mixed print is not treated as a year-round property of the procedure.