Averages hide the thing you usually want to know. Two products can sell the same annual volume while one ships steadily and the other spikes once a quarter — and only the second one wrecks your inventory planning. Standard deviation is what separates them, and Deluge has everything needed to calculate it.

Before you start

• A form with twelve monthly numeric fields, plus Volumen, Promedio and Desviacion (decimal) to hold the results.
• Nothing else. This is pure Deluge with no external calls.

Deluge function

// ============================================
// SAMPLE STANDARD DEVIATION OVER MONTHLY VALUES
// ============================================

void Stats.calculateStdDev(int recordId)
{
    rec = DATA_ENTRY[ID == recordId];

    // --- 1. Gather the twelve months into one list ---
    months = List();
    months.add(ifnull(rec.Ene,0));
    months.add(ifnull(rec.Feb,0));
    months.add(ifnull(rec.Mar,0));
    months.add(ifnull(rec.Abr,0));
    months.add(ifnull(rec.May,0));
    months.add(ifnull(rec.Jun,0));
    months.add(ifnull(rec.Jul,0));
    months.add(ifnull(rec.Ago,0));
    months.add(ifnull(rec.Sept,0));
    months.add(ifnull(rec.Oct,0));
    months.add(ifnull(rec.Nov,0));
    months.add(ifnull(rec.Dic,0));

    // --- 2. Keep only months that actually traded ---
    values = List();
    total  = 0.0;
    for each m in months
    {
        if(m > 0)
        {
            values.add(m);
            total = total + m;
        }
    }

    // n is the COUNT of months with data. Not the sum of their values.
    n = values.size();

    if(n < 2)
    {
        info "Need at least two months of data for a sample standard deviation.";
        return;
    }

    // --- 3. Mean stays a decimal. Rounding here corrupts everything after. ---
    mean = total / n;

    // --- 4. Sum of squared deviations from the mean ---
    sumSquares = 0.0;
    for each v in values
    {
        diff = v - mean;
        sumSquares = sumSquares + (diff * diff);
    }

    // --- 5. Sample variance divides by n-1; population variance by n ---
    variance = sumSquares / (n - 1);
    stdDev   = variance.sqrt();

    rec.Volumen    = total;
    rec.Promedio   = mean.round(2);
    rec.Desviacion = stdDev.round(3);

    info "n=" + n + "  mean=" + mean.round(2) + "  sd=" + stdDev.round(3);
}

Notes

• Sample or population? Dividing by n - 1 gives the sample standard deviation, which is right when your twelve months are a sample of ongoing trading. Divide by n for the population figure, when those twelve months are the entire universe you care about. Using n - 1 is the safer default and is what spreadsheet STDEV does.
• Zero is not the same as missing. This treats a zero month as “no data” and excludes it, matching the original intent. If a genuine zero-sales month is meaningful in your data, drop the if(m > 0) filter and use all twelve.
• Why the original wrote (0.1 - 0.1). That odd-looking expression was a way to force a decimal accumulator, because a plain 0 is an integer in Deluge and would make every subsequent addition integer arithmetic. Writing 0.0 does the same job and reads better, but the trick is worth recognising in older scripts.
• Round at the end, never in the middle. Rounding the mean before squaring deviations is what broke the original. Carry full precision through the calculation and round only what you store.