Eighty percent in three days: reading the concentration number

Every audit prints one line that readers quote more often than any other. It reads like 80% of profit made in 3 days, and it looks like a verdict on how bursty a strategy is. Across the 26 default audits in the catalog the count runs from 3 days to 452 days, and the same low number turns out to have two opposite meanings. One of them is the finding people expect. The other one is a warning about the size of the profit, not about its timing. Telling them apart takes one extra number, and it is printed on the same page.

How the number is built

The rule sits in our detector and has no free parameters beyond the 80% mark. Every closed trade is booked on the calendar day of its exit, the day totals are sorted best first, and the engine counts how many of those days it takes before the running sum reaches 80% of the run's total net result. The share that follows is that count divided by the number of days on which the EA closed at least one trade. The lamp thresholds live in the same function, caution below 8% of active days and red below 3%.

The decisive detail is the denominator of that 80%. It is the net result after every losing day, not the sum of the winning days. A run that finishes thin has a small target to reach, so a handful of good days clears it. A run that finishes fat has a large target, and the count grows accordingly. The metric therefore compares the best days against the total, and the total is free to be almost nothing.

What that does to a thin result

Lizard closed trades on 4,242 days between 2003.05.05 and 2026.08.16 and finished at 183.99 USD net. Its three best days are 2025-12-18 with 75.57 USD, 2025-11-28 with 61.94 and 2026-05-12 with 58.04. Those three sum to 195.55 USD, which is more than the entire net of the run. The audit prints 80% of profit made in 3 days. That is 0.1% of 4242 trading days. and lights the lamp red. For scale, the worst single day of the same run cost 56.58 USD.

Every part of that is correct, and it is still not a statement about bursts. It is a statement about twenty three years of trading that netted less than three good days. Note also where those three days fall. All of them sit inside the last nine months of a run that started in 2003.

And what it does to a real profit

Gold Atlas needs 54 of 4,636 active days, 1.2% of them, to cover 80% of a net of 8,239.71 USD, and its best day alone made 276.36 USD. Prop Firm Gold EA needs 67 of 5,533 days, also 1.2%, for a net of 6,769.79 USD. Both sit far below the red threshold, both made money, and here the number means exactly what the reader assumes. The result really does live in a few dozen sessions out of two decades.

So Lizard and Gold Atlas carry the same lamp colour for reasons that have nothing in common. That is the whole point of this article. Read the count next to the net.

The shortcut fails in the other direction too

A thin net does not produce a low count on its own. Waka Waka closed trades on 170 days, needed 95 of them for its 80%, and finished at 166.44 USD. That is a share of 55.9%, the highest of the 23 default audits that carry the figure, on a net smaller than Lizard's. Its best day made 3.69 USD. What drives the count down is a best day that is large against the total, and this run has a best day worth about two percent of its result. Anyone inferring the count from the size of the profit alone gets this case backwards.

Where the number is missing on purpose

Three of the 26 default audits carry no concentration figure at all. Logan finished at -24,334.65 USD, Scalping Robot Pro at -53,234.68 USD and Smart Gold Hunter at -1,418.55 USD. When a run ends below its deposit there is no profit to concentrate, so the field is empty rather than zero. Those three are exactly the three default audits with a negative net result, which is the cleanest way to see what the metric depends on.

The upper end of the scale

At the other extreme the result is spread out. CryonX needs 452 of 3,191 days (14.2%), Quantum Athena X needs 254 of 902 (28.2%) and Quantum Queen X needs 258 of 820 (31.5%). A spread result is a different property from a good one, and the audits keep those questions apart. Quantum Queen X carries an ok concentration lamp and red lamps for data quality and structure in the same document.

Run these checks yourself

• Put the count next to the net result. A count of 3 on a net of 183.99 USD and a count of 54 on a net of 8,239.71 USD produce the same lamp and are not the same finding.
• Compare shares, not raw counts. Lizard's 3 days sit in 4,242 active days, Waka Waka's 95 sit in 170. The raw counts are not on the same scale and the shares are.
• Ask what being absent costs. A count of N out of M means the result lives in a window you have to be present for. Our own flag text puts the second consequence next to it, that consistency rules at prop firms punish exactly this profile.

The concentration figure for every audited EA sits on its audit page with the deal level evidence behind it. Your own tester report? The browser check is free.