Measure Actual Proc Rate From Combat Logs

You open a combat log after a boss fight and see 15 procs from a weapon enchant that claims a 20% chance on each swing. You swung 100 times. The tooltip says 20%, but you saw 15%. To measure actual proc rate from combat logs, divide the number of procs observed by the number of eligible trigger events. That ratio, p̂ = 15 / 100 = 0.15, is your observed rate. The question is how far 0.15 is from the tooltip's 0.20, and whether that difference means the enchant is broken or you just got unlucky. For any proc in Season of Discovery, that question is answered.

Take a Warrior using Victory Rush, the Rune that generates 20 Rage on a killing blow. The tooltip does not give a proc probability, Victory Rush works every time you land a killing blow, so this example uses a different proc. A Mage with Living Bomb Rune casts Fire Blast 200 times in a dungeon. The Rune causes the target to explode after 4 seconds, and the explosion is the proc. The combat log records 34 explosions. The eligible trigger events are the 200 Fire Blast casts that landed on a target that survived long enough for the explosion to occur. N_triggers = 200. N_procs = 34. p̂ = 34 / 200 = 0.17, or 17%.

Now compare that to the tooltip probability. Living Bomb's tooltip states a 20% chance per Fire Blast. Your observed 17% is 3 percentage points below the tooltip. That is not a huge gap, but you need to know whether 200 events is enough to trust that 17% is meaningfully different from 20%.

The calculation itself is simple division. p̂ = N_procs / N_triggers. For the Living Bomb example: p̂ = 34 / 200 = 0.17. The tooltip probability is 0.20. The difference is 0.03. The question is whether 200 events is enough to say this difference is real or just random variance.

You need a relative frequency calculator at this point. That calculator takes your observed ratio (0.17), the tooltip probability (0.20), and the number of trials (200), and returns a confidence interval or a p-value. For 200 trials, the standard error of the proportion is sqrt(0.20 * 0.80 / 200) ≈ 0.028. A 95% confidence interval around the tooltip probability runs from about 0.20 - 1.96 * 0.028 ≈ 0.145 to 0.20 + 1.96 * 0.028 ≈ 0.255. Your observed 0.17 falls inside that interval. That means the 17% you saw is not statistically different from the tooltip's 20% given only 200 swings. The proc rate is working as intended.

What Changes With More Data

If you had 1000 Fire Blast casts and saw 170 procs, the standard error drops to sqrt(0.20 * 0.80 / 1000) ≈ 0.013. The 95% interval shrinks to 0.174 to 0.226. A 17% rate on 1000 trials falls outside that interval. At that sample size, the observed rate of 17% is significantly lower than 20%. You would have evidence that the actual proc rate is lower than the tooltip claims.

The core reason you compare observed procs with a tooltip probability is to check whether a Rune, talent, or enchant is performing as advertised. Season of Discovery has no balance patches after Phase 8, so if a proc rate is wrong, it stays wrong. You cannot trust a forum post that says "Living Bomb procs 10% of the time" without checking the math yourself.

Three things can go wrong in the comparison. First, your combat log may miss trigger events. If your addon filters out certain swing types or the log truncates on a long fight, N_triggers is too low and p̂ is too high. Second, the proc may have hidden conditions the tooltip does not state. For example, a proc that only works on critical hits must have crits as the eligible trigger, not all hits. Third, the tooltip probability itself may be wrong. Blizzard has shipped incorrect tooltips before, and in maintenance mode they will not be fixed.

How to Identify Eligible Events Correctly

Open the combat log and count only the events that could produce the proc. For a weapon proc that says "on hit", misses do not count. For a Rune like Serpent Spread that activates on Multi-Shot, only Multi-Shot casts are eligible. For Void Plague, which replaces Shadow Word: Pain, the proc is the damage tick, and the eligible trigger is the initial cast. If you count the ticks as separate triggers, you inflate N_triggers and deflate p̂.

The most common failure when you measure actual proc rate from combat logs is missing trigger events. Combat logs in WoW Classic have a memory limit. On a long boss fight, the log may drop older events. If the log drops the first 50 swings of a fight but keeps the procs from the remaining 50, your N_triggers is 50 instead of 100. p̂ doubles from 0.15 to 0.30. You would conclude the proc rate is 30% when it is actually 15%.

To avoid this, export the full log file, not the in-game snippet. The file stores all events. If you cannot export, note the fight duration and check if the log file size seems small for that length. A 3-minute fight that produces a 200 KB log is probably truncated. A 3-minute fight that produces a 2 MB log is probably complete.

Another failure case: the proc has a hidden cooldown. Some Runes in Season of Discovery, like Divine Storm, have a 30-second cooldown. If you count every eligible swing as a trigger but the proc cannot fire during the cooldown, your N_triggers is too high and p̂ is too low. Check the Rune's cooldown in the tooltip or on Wowhead before counting triggers.

This method suits players who distrust third-party build sites and want to verify a Rune or enchant's performance using their own data. It suits theorycrafters comparing hybrid builds against pure specs who need to know whether a proc rate difference is real or noise. It suits returning Season of Discovery players who last played in Phase 3 or earlier and need to confirm that Runes added later are working as documented.

Skip this subject if you need current meta rankings or DPS sims. The comparison tells you whether the proc rate matches the tooltip, not whether the Rune is optimal for raiding. Skip it if you are playing TBC or Wrath Classic, where the talent trees and Rune systems are different. Skip it if you have fewer than 50 eligible events in your log, because the confidence interval will be too wide to draw any conclusion. A player with 30 swings and 5 procs cannot tell whether the true rate is 10% or 30%.

The single thing that most often goes wrong here is counting the wrong events as triggers. A player sees 15 procs, counts 100 swings, and gets 15%. But 20 of those swings were misses that could not proc. The real trigger count is 80, and p̂ is 18.75%, much closer to the tooltip 20%. Identify eligible events correctly before you run the calculation.